# FIFTY SHADES OF GREY: EXPLORING THE DARK SIDES OF LEADERSHIP AND FOLLOWERSHIP

EDITED BY : Susanne Braun, Ronit Kark and Barbara Wisse PUBLISHED IN : Frontiers in Psychology

### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-695-6 DOI 10.3389/978-2-88945-695-6

### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

Frontiers in Psychology 1 January 2019 | Fifty Shades of Grey

# FIFTY SHADES OF GREY: EXPLORING THE DARK SIDES OF LEADERSHIP AND FOLLOWERSHIP

Topic Editors:

Susanne Braun, Durham University, Durham, United Kingdom Ronit Kark, Bar Ilan University, Israel Barbara Wisse, University of Groningen, Netherlands and Durham University, United Kingdom

Cover image: Kerdkanno/Shutterstock.com

The scientific field of leadership and followership is fast evolving and has seen several interesting developments over recent years. The early heroic views of leadership are slowly turning into more nuanced perspectives, including the understanding that leadership and followership are mutually dependent on each other. Likewise, there is a growing awareness that the focus on the positive side of leadership and followership can be fruitfully complemented by a focus on the darker sides of these constructs.

According to the latest research plenty of "grey areas" exist, where further insights into leadership and followership are needed. We seek to emphasize the different shades of dark leadership by taking leaders, followers, and their interaction in specific contexts into account.

Accordingly, many of the findings presented in this Research Topic align with a deviation away from the idea of the omnipotent leader. Not only leaders' dark traits such as narcissism and psychopathy, but also followers' Machiavellianism emerged as hindering factors for positive organizational functioning. Other results presented in this Research Topic will be fruitful to explain what drives leaders towards

dark-side behaviors, the consequences of dark-side leader behaviors (e.g., different types of destructive leadership), and how followers respond to them (e.g., follower attributions of perceived abusive supervision). Contributions to this Research Topic are also pushing the boundaries of current theorizing, shedding further light on the "shades of grey," when it comes to the possibly unintended negative consequences of leadership and followership.

In sum, the dark sides of leadership and followership are a natural part of an organizational reality that many employees face day in and day out. The aim of this Research Topic is to encourage an integrative view of leadership and followership and their dark sides, for a better understanding of complex organizational systems and implications for better practice.

Citation: Braun, S., Kark, R., Wisse, B., eds. (2019). Fifty Shades of Grey: Exploring the Dark Sides of Leadership and Followership. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-695-6

# Table of Contents

*05 Editorial: Fifty Shades of Grey: Exploring the Dark Sides of Leadership and Followership*

Susanne Braun, Ronit Kark and Barbara Wisse

### SECTION 1

### DARK-SIDE LEADER TRAITS AND BEHAVIORS


Ellen A. Schmid, Armin Pircher Verdorfer and Claudia V. Peus


# SECTION 2

### THE INTERPLAY BETWEEN LEADER AND FOLLOWER TRAITS


Dick P. H. Barelds, Barbara Wisse, Stacey Sanders and L. Maxim Laurijssen

*103 Angels and Demons: The Effect of Ethical Leadership on Machiavellian Employees' Work Behaviors*

Frank D. Belschak, Deanne N. Den Hartog and Annebel H. B. De Hoogh

# SECTION 3

### NOVEL EXTENSIONS OF THEORY AND INTEGRATIVE FRAMEWORKS

*115 'Killing Me Softly With His/Her Song': How Leaders Dismantle Followers' Sense of Work Meaningfulness*

Petra Kipfelsberger and Ronit Kark

*129 I am as Incompetent as the Prototypical Group Member: An Investigation of Naturally Occurring Golem Effects in Work Groups* Alex Leung and Thomas Sy

# Editorial: Fifty Shades of Grey: Exploring the Dark Sides of Leadership and Followership

### Susanne Braun<sup>1</sup> \*, Ronit Kark <sup>2</sup> and Barbara Wisse1,3

*<sup>1</sup> Durham University Business School, Durham University, Durham, United Kingdom, <sup>2</sup> Department of Psychology, Bar Ilan University, Ramat Gan, Israel, <sup>3</sup> Department of Psychology, University of Groningen, Groningen, Netherlands*

Keywords: abusive supervision, dark side, destructive leadership, followership, Machiavellianism, narcissism, psychopathy

**Editorial on the Research Topic**

### **Fifty Shades of Grey: Exploring the Dark Sides of Leadership and Followership**

"I could stand in the middle of Fifth Avenue and shoot somebody and I wouldn't lose any voters. Okay. It's like incredible!" Donald J. Trump, President of the United States, in January 2016 at a campaign rally in Iowa.

In light of corporate and political turmoil and subsequent questions raised about leaders' dark sides, this Research Topic is particularly timely. We set out to contribute to theoretical, empirical and methodological advancements, focusing on dark sides of personality, processes, and perceptions, and how they relate to leader-follower relationships. Studies of the dark side of leadership follow a long-standing tradition (Conger, 1990), and initially focused mainly on negative leader traits such as narcissism (Braun, 2017) and leader behaviors such as abusive supervision (Hogan and Kaiser, 2005; Tepper, 2007; Schyns and Schilling, 2013; Tepper et al., 2017). The particular potential for toxicity to unfold at the intersections of leadership and followership has been noted (Padilla et al., 2007), yet research into this domain remains largely underdeveloped. While followership theories receive increasing attention (Uhl-Bien et al., 2014), the potential dark sides of followership or followers' impact on dark-side leaders remain unclear. Deviating from the unidimensional view that leaders are omnipotent and to be blamed for negative outcomes, we seek to place emphasis on the different "shades" of dark leadership by focusing on how dark leadership can be explained by taking leaders, followers, and their interaction in specific contexts into account.

In line with the purpose to explore the intersections between dark-side leadership and followership, we saw three main themes emerging from the articles published in this Research Topic. The first theme revolves around leader traits and behaviors. It focuses on questions such as what makes a "dark-side" leader and what "dark-side" leaders do. The second theme accounts for the interaction between leaders' and followers' characteristics, and zooms in on the extent to which this interaction may affect the negative impact of "dark-side" leadership or followership. Finally, the articles also reflect novel ideas, extensions and integration of current theories at the interface between leadership and followership.

# LEADER TRAITS AND BEHAVIORS

The conceptual paper by de Vries reviewed personality traits and their links with dark leadership styles. The Three Nightmare Traits (TNT), leaders' dishonesty, disagreeableness, and carelessness, were found to be aligned with low honesty-humility, agreeableness, and conscientiousness. Using a Situation-Trait-Outcome-Activation (STOA) model the author argued that specific situations should attract TNT leaders, activate their dark-side traits, and result in (mainly but not exclusively) negative outcomes in relation to the recognition, perception, and attribution of leadership.

### Edited and reviewed by:

*Ilias Kapoutsis, Athens University of Economics and Business, Greece*

\*Correspondence:

*Susanne Braun susanne.braun@durham.ac.uk*

### Specialty section:

*This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology*

Received: *08 August 2018* Accepted: *13 September 2018* Published: *05 October 2018*

### Citation:

*Braun S, Kark R and Wisse B (2018) Editorial: Fifty Shades of Grey: Exploring the Dark Sides of Leadership and Followership. Front. Psychol. 9:1877. doi: 10.3389/fpsyg.2018.01877*

In addition, three of the published articles gave primary attention to the question of what dark-side leaders do and how they affect followers at work and in terms of their personal lives. This widens our scope of leadership behaviors that are perceived as negative, and allows us to explore more discrete types of negative behaviors and their outcomes.

Three different types of destructive leadership and their effects on follower outcomes were assessed in an experiment and a field study by Schmid et al. Differentiating between distinct types of negative leadership their research focuses on followerdirected (abusive supervision), organization-directed, and selfinterested (exploitative) destructive behaviors. All three forms of dark-side leader behaviors predicted followers' negative affect. However, abusive supervision elicited the highest levels of fear. In relation to turnover intentions, exploitative leadership and abusive supervision affected calculative and immediate turnover intentions similarly.

Nauman et al. extended the research to explore how dark-side leadership affects the private sphere of life of the employees. They assessed despotic leadership (i.e., tendencies toward authoritarian and dominant behavior in pursuit of selfinterest, self-aggrandizement, and exploitation of others) and its negative effects, which the authors hypothesized would transcend from the workplace to subordinates' personal lives (increased emotional exhaustion and work-family conflict, and decreased life satisfaction). The results confirmed their hypotheses. They show that negative forms of leadership can also affect our personal lives, homes and families and opens up a new field of research at the work-life interface. The work also connects with our second theme, the interplay between traits of leaders and followers. In this study, followers' anxiety increased the negative impact of despotic leadership.

Schyns et al. extended the perspective from dark-side leader behaviors to follower perceptions and attributions of these behaviors. Comparing different levels of abusive behavior (constructive leadership, laissez-faire leadership, mild to strong abuse), they analyzed follower perceptions of abusive supervision and follower attributions as moderators. The three-study series employed manipulations of leaders' abusive behaviors and established attributions of the leaders' intentionality in the behavior and the level of his/her control as moderators. Relationships between abusive supervision perceptions and outcome variables (loyalty, turnover, and voice) were largely buffered by the attribution of leader intentionality. In Study 3, a survey of abusive supervision perceptions, however, control attributions strengthened the relationships with loyalty and voice.

# THE INTERPLAY BETWEEN TRAITS OF LEADER AND FOLLOWER

Three articles in this Research Topic provided a largely new angle. They considered the relevance of follower traits when confronted with dark-side leadership, but also followers' own dark-side traits.

Looking at leader narcissism, Nevicka et al. analyzed the interface between self-absorbed, entitled narcissistic leaders and insecure follower, who make "easy targets" for narcissists. The authors conducted two field studies. Followers with low self-esteem and low core self-evaluations perceived narcissistic leaders as more abusive than those with high self-esteem or high core self-evaluations. Abusive supervision perceptions in turn related to lower follower performance and higher experiences of burnout, pointing to risks of leader narcissism for vulnerable followers.

Barelds et al. also studied followers' self-esteem, but in terms of how it affected the relationship between leaders' psychopathy and their self-serving behaviors. The authors first conducted an experimental study, in which they manipulated follower self-esteem, measured leader psychopathy, and assessed their combined effects on leader self-serving behavior using an ultimatum game. They also conducted a multi-source field study using questionnaires to assess leader psychopathy, follower selfesteem, and perceived leader self-serving behavior. Across both studies they found that leader psychopathy was positively related to leader self-serving behaviors, but only when their followers had low rather than high self-esteem. Again, these findings show that that the degree to which dark-side traits of leaders are reflected in their behavior depends on the characteristics of their followers. Follower characteristics can mitigate the negative impact of darkside leadership.

However, not only leaders' dark-side traits pose risks to organizations; followers' dark-side traits may do the same. Belschak et al. studied ethical leadership as a potential remedy for negative behaviors of Machiavellian followers. Followers with high Machiavellianism are goal-driven to the extent that they use all possible means to achieve desired ends. Machiavellianism predicted reduced helping behavior and increased knowledge hiding and emotional manipulation, but only when ethical leadership was low. That is, ethical leadership served as a buffer of the negative outcomes of dark-side followership.

# NOVEL EXTENSION OF THE THEORY AND INTEGRATION

Two articles challenged current theoretical thinking at the interface of leadership and followership. One article focused on the conditions under which leaders' positive efforts can in fact backfire, and the other one addressed the relevance of negative followership theories at the group level.

Kipfelsberger and Kark developed a theoretical model to explain the conditions under which leaders' meaning making efforts, despite their good intentions, can "kill" followers' experiences of meaningfulness at work. The authors applied a wide angle taking into account leaders' characteristics, followers' characteristics and the context. They argued that leaders harm followers' work meaningfulness when followers' experiences of coherence, purpose or significance of work are diminished. The six conditions that can affect the reduction of followers' sense of meaningfulness included in the model capture leaders' personality traits, leaders' behaviors, the relationship between leader and follower, followers' attributions, followers' characteristics, and job design. The negative consequences of diminished meaningfulness comprise cynicism, disengagement, and decreased well-being.

Leung and Sy extended the established construct of implicit followership theories to the group level showing that Golem effects can occur as a consequence of negative beliefs held within teams. Golem effects capture a special case of self-fulfilling prophecies, the idea that negative performance expectations result in low performance. The authors studied naturally occurring Golem effects in the form of negative implicit followership theories, specifically incompetency schemas that are shared within groups. Results confirmed showed groups who shared negative group-level Implicit Followership Theories (GIFTs) affected follower performance negatively through decreased self-efficacy and effort.

### CONCLUSION

We see the extension and integration of leadership and followership theories in the dark-side realm as one of the major contributions of this Research Topic. The work presented places particular emphasis on the role that followers can play in

# REFERENCES


dark-side leadership, whether through their own traits, implicit theories or attributions. We also see the importance of the context as one major aspect for further investigations. Future research should add to the understanding of how leaders, followers, their relationships and the context interact within the dynamic of dark sides in organizations. Moreover, future research can look into how negative leadership affects different life spheres of the followers, as well as of the leaders themselves, We see particular strengths of the empirical papers presented here in their methodological rigor, including experimental as well as survey data, gathered from multiple sources and in multiplestudy series. Better understanding the dark sides of leadership and followership is, so we believe, timely. Future research may decipher more unique and discrete types of dark leadership and followership, focus on toxic relationships and their consequences, and find ways to reduce the harmful effects. In other words, there can be at least "50 shades of gray" in dark-side leadership.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Braun, Kark and Wisse. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Three Nightmare Traits in Leaders

### Reinout E. de Vries 1,2 \*

<sup>1</sup> Department of Experimental and Applied Psychology, Vrije Universiteit Amsterdam, Amsterdam, Netherlands, <sup>2</sup> Department of Educational Science, University of Twente, Enschede, Netherlands

This review offers an integration of dark leadership styles with dark personality traits. The core of dark leadership consists of Three Nightmare Traits (TNT)—leader dishonesty, leader disagreeableness, and leader carelessness—that are conceptualized as contextualized personality traits aligned with respectively (low) honesty-humility, (low) agreeableness, and (low) conscientiousness. It is argued that the TNT, when combined with high extraversion and low emotionality, can have serious ("explosive") negative consequences for employees and their organizations. A Situation-Trait-Outcome Activation (STOA) model is presented in which a description is offered of situations that are attractive to TNT leaders (situation activation), situations that activate TNT traits (trait activation), and the kinds of outcomes that may result from TNT behaviors (outcome activation). Subsequently, the TNT and STOA models are combined to offer a description of the organizational actions that may strengthen or weaken the TNT during six career stages: attraction, selection, socialization, production, promotion, and attrition. Except for mainly negative consequences of the TNT, possible positive consequences of TNT leadership are also explored, and an outline of a research program is offered that may provide answers to the most pressing questions in dark leadership research.

### Edited by:

Barbara Wisse, University of Groningen, Netherlands

### Reviewed by:

Shane Connelly, University of Oklahoma, United States Chiara Ghislieri, Università degli Studi di Torino, Italy

> \*Correspondence: Reinout E. de Vries re.de.vries@vu.nl

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 13 October 2017 Accepted: 14 May 2018 Published: 04 June 2018

### Citation:

de Vries RE (2018) Three Nightmare Traits in Leaders. Front. Psychol. 9:871. doi: 10.3389/fpsyg.2018.00871 Keywords: HEXACO, leadership, personality, contextualized personality, dark triad, STOA, attraction-selectionattrition, career stages

# INTRODUCTION

Interest in leadership traits and the relations between leader personality and leadership styles has waxed and waned over the decades, following the rise and fall in popularity of situational (nurture) and behavioral genetic (nature) explanations of personality and leadership (Judge et al., 2009). Although most researchers nowadays adopt an integrated ("nature in nurture") stance (e.g., Plomin et al., 2016), models that integrate personality traits, leadership styles, and situations that account for—or can counter—the activation of personality traits and leadership styles, are still rare. This is especially true when considering the dark side of personality and leadership. Although—especially in the wake of several high-profile corporate scandals (e.g., Enron, WorldCom, Volkswagen)—a bourgeoning field of research on dark personality traits (Hogan and Hogan, 1997; Paulhus and Williams, 2002; Chabrol et al., 2009; Buckels et al., 2014) and dark leadership styles (Tepper, 2000; Reed, 2004; Rosenthal and Pittinsky, 2006; Hauge et al., 2007; De Hoogh and Den Hartog, 2008; Schmid Mast et al., 2009; Ghorbani et al., 2010; Schyns and Schilling, 2013; Boddy, 2017; Schmid et al., in press) has emerged, these two fields of research remain by-and-large separate.

As its main contribution the following review offers a theoretical, empirical, and practical integration of personality and (dark) leadership research (1) by proposing that they can—and should—be integrated by conceptualizing leadership styles as contextualized personality, (2) by introducing the so-called "Three Nightmare Traits" (TNT; de Vries, 2016)—i.e.,

**8**

dishonesty (low honesty-humility), disagreeableness (low agreeableness), and carelessness (low conscientiousness)—as an overarching conceptualization of dark side personality and leadership, (3) by using the Situation-Trait-Outcome Activation (STOA) model (de Vries et al., 2016b) as a framework to explain the effects of TNT leaders on, in, and through situations, and (4) by providing recommendations for organizations how to deal with TNT leaders in different career stages using an extended Attraction-Selection-Attrition (ASA; Schneider, 1987) model.

Although most of this review will focus on the TNTs among leaders (hereafter referred to as "TNT leadership"), one of the core assumptions of this review is that leadership styles can be interpreted as contextualized personality traits. That is why, before focusing on the TNT leadership, the following section offers a more general explanation of why there is reason to assume that all leadership styles—not only those that are related to the TNT—can be considered contextualized personality traits.

# LEADERSHIP STYLES AS CONTEXTUALIZED PERSONALITY

In the following, I will specifically focus on leadership styles. Among leadership scholars, leadership styles—or behavioral tendencies—probably constitute the most common research area. Still, it can be considered a subset of a broader leadership domain, which encompasses, among others, leader knowledge, skills, and abilities [e.g., (emotional) intelligence, leader experience, and leader expertise; (Podsakoff et al., 1983; Cavazotte et al., 2012)], motivation to lead (Chan and Drasgow, 2001), leadership roles (Denison et al., 1995), and leader-subordinate relational quality (Dulebohn et al., 2017) research. Leadership style, as used here, refers to the way a "leader" (i.e., somebody who has gained position power through a process of legitimation) tends to act toward people he or she directs or supervises. Popular leadership styles in the literature include—for example autocratic and democratic leadership, directive and participative leadership, task- and relation-oriented leadership, charismatic leadership, and transformational and transactional leadership (Bass and Bass, 2009), but next to these mostly "bright" leadership styles, dark leadership styles have received an increasing amount of attention in the last two decades (Schyns and Schilling, 2013).

Contextualization occurs when a relevant context (or frameof-reference) is added to a (generic or non-contextualized) personality questionnaire (Schmit et al., 1995; Bing et al., 2004). Contextualization can be accomplished by completely rewriting personality items or by using a contextual "tag" to reflect a certain context (e.g., work, home, school, sports, etc. . . ). In the case of leadership, a leadership-contextualized personality questionnaire can be constructed by rewriting personality items to reflect behaviors expressed by somebody in a hierarchical position or to add a tag such as "as a leader" to items. For instance, when contextualizing using a tag, a generic HEXACO (reversed) Agreeableness item "People sometimes tell me that I am too critical of others" (Ashton and Lee, 2009) would become "As a leader, people sometimes tell me that I am too critical of others."

Contextualized versions of personality scales have been found to be strongly (generally ≥ 0.65) related to their respective generic versions (Bing et al., 2004; Holtrop et al., 2014a,b; Robie et al., 2017) and they generally offer better validities than generic personality scales (Bing et al., 2004; Lievens et al., 2008; Holtrop et al., 2014a; Robie et al., 2017), mainly because contextualized scales reduce within-person inconsistencies in item responding (Lievens et al., 2008). Consequently, leadershipcontextualized personality questionnaires are likely to offer better validities in the prediction of leader-relevant outcomes than generic personality questionnaires.

In the following, I will offer five arguments why leadership styles can be considered contextualized personality traits. (1) The content domain of leadership styles can be considered a subset of personality traits. Whereas personality provides a parsimonious description of all possible human behaviors that are psychologically meaningful in all possible situations, in line with common definitions of leadership, leadership models restrict themselves to behaviors in a subset of situations, i.e., those that are relevant to the goal-directed (hierarchical) influence of one individual vis-à-vis a group of other individuals. (2) In so far leadership items refer to behavioral tendencies (or: leadership styles) instead of attributions made by subordinates, they are formulated equivalent to personality items. Terms that have been used to describe prototypical leadership, such as determined, decisive, organized, responsible, honest, and fair (Lord and Maher, 1993) are the very same terms that have been used in lexical personality studies (Goldberg, 1990; Ashton et al., 2004). Items in leadership questionnaires that describe actual behaviors (e.g., "criticizes poor work;" Fleishman, 1953; de Vries et al., 2002) instead of subordinates' leadership attributions or evaluations, are highly similar to items in personality questionnaires that describe behaviors (e.g., "criticizes others' shortcomings;" Goldberg et al., 2006; see also the HEXACO Agreeableness item above). (3) Empirical evidence shows that leadership styles—like personality traits—are stable across time (Harris and Fleishman, 1955; Dvir and Shamir, 2003; Nielsen et al., 2008; Tafvelin et al., 2011) 1 . (4) Leadership styles show similar levels of heritability and genetic correlations show "that there is a strong common source [italics added] of genetic variation underlying leadership and personality" (Johnson et al., 2004, p. 31). And, last but not least, (5) there are strong relations between personality traits and leadership styles (de Vries, 2012).

Although the first four arguments are theoretically and empirically straightforward, this may not be the case for the last argument. In fact, one of the consistent findings in most studies has been the relatively weak observed relations between personality traits and leadership styles (Judge and Bono, 2000; Bono and Judge, 2004; Lim and Ployhart, 2004; De Hoogh et al., 2005; DeRue et al., 2011), which has led Bono and Judge (2004) to hypothesize that "leadership behaviors are more malleable, more

<sup>1</sup> In contrast to research on the stability of personality, studies that have investigated the stability of leadership styles are relatively few. Apart from Harris and Fleishman (1955), the other three referenced articles included two waves of leadership measurement in the context of a longitudinal design, but these articles only used single subordinate reports of leadership.

transient, and less trait-like than one might otherwise believe" (p. 906). However, as I've argued elsewhere (de Vries, 2012), the main reason for these relatively weak relations is the fact that all studies included in Bono and Judge's (2004) meta-analysis used leaders' self-ratings of personality and subordinate-ratings of leadership, which introduces an important cross-source upper limit restriction, i.e., that the maximum possible correlation between two different variables obtained from two different sources is equal to the minimum cross-source correlation of one of these two variables.

The upper limit of cross-source correlations of the same variable (i.e., self-other agreement) in work settings is generally low; not surpassing the r = 0.25 level for leadership (Warr and Bourne, 1999; Ostroff et al., 2004) and r = 0.30 for personality (de Vries et al., 2008; Connelly and Ones, 2010) 2 . The fact that none of the meta-analytic zero-order correlations in Bono and Judge's (2004) cross-source meta-analysis surpassed r = 0.17 (between extraversion and charismatic leadership), is thus understandable when taking the cross-source upper limit into account. When correcting for low cross-source correlations, de Vries (2012) obtained strong—and consistent—estimates of the relations between personality and leadership styles. That is, charismatic, supportive, and ethical leadership were strongly related to respectively extraversion (β = 0.76), agreeableness (β = 0.74), and honesty-humility (β = 0.50), with only taskoriented leadership having a somewhat weaker relation with conscientiousness (β = 0.33).

These corrected relations offer strong support for a contextualized interpretation of leadership style scales. According to de Vries (2012), charismatic leadership can be considered a contextualized version of extraversion because of the social self-esteem, social boldness, energy, and enthusiasm typical for both extraversion and charismatic leadership; ethical leadership can be considered a contextualized version of honestyhumility because both involve behaviors expressive of sincerity, fairness, and greed avoidance; supportive leadership can be considered a contextualized version of agreeableness (but also some extraversion), because both involve gentleness, patience, flexibility, and tolerance when dealing with subordinates' problems; and finally, task-oriented leadership can be (partly) considered a contextualized version of conscientiousness, because both have to do with order, discipline, and perfectionism when carrying out tasks. Consequently, these four leadership styles—when operationalized as behavioral tendencies—seem to overlap to a large extent with personality traits commonly found in personality models and they may be, accordingly, regarded as contextualized versions of these four traits.

In the following, I will argue that the "negative" pole of three of these four traits are associated with what I will call the "Three Nightmare Traits" (TNT). That is, especially leaders who are characterized by low honesty-humility (henceforth called "leader dishonesty"), low agreeableness ("leader disagreeableness"), and low conscientiousness ("leader

carelessness") may have important negative effects on their subordinates, their organization, and in some cases even society at large.

# THREE NIGHTMARE TRAITS (TNT) IN LEADERSHIP

To explore the TNTs, it is necessary to first introduce the HEXACO personality model, from which these three traits are derived. The HEXACO model—here applied to leadership—has its basis in lexical personality research. The main assumption of lexical personality research is that anything that can be said about personality is codified in language, and that sufficiently large dictionaries contain a great number of words that may be used to describe somebody's personality (Galton, 1884; Goldberg, 1981). Factor analyses on self- and/or other ratings using these words (most often adjectives) have been applied to uncover the main dimensions of personality. In first instance, lexical personality research (Goldberg, 1990) yielded five main dimensions of personality that are commonly known as the "Big Five." However, follow-up studies (Ashton et al., 2004; Saucier, 2009; De Raad et al., 2014) have shown that a six-dimensional structure more optimally captures the largest possible cross-culturally replicable personality space in lexical datasets. The dimensions that span this six-factor personality space are commonly known by the HEXACO acronym, i.e., Honesty-humility, Emotionality, eXtraversion, Agreeableness, Conscientiousness, and Openness to experience. Of these six personality dimensions, honestyhumility is least well-captured by the Big Five model, but some of the content associated with emotionality and agreeableness is rearranged in the HEXACO model. The most prominent feature of this rearrangement is that content associated with anger is associated with low HEXACO agreeableness instead of low Big Five emotional stability and content associated with sentimentality is associated with high HEXACO emotionality instead of high Big Five agreeableness (see Ashton et al., 2014, for more details). In this paper, note that when I refer to (leader) dishonesty, disagreeableness, and carelessness, I'm referring to the opposite poles of three of the six HEXACO factors, i.e., low honesty-humility, low agreeableness, and low conscientiousness<sup>3</sup> .

Leader dishonesty, the first of the TNT as applied to leadership, is straightforwardly defined as the opposite pole of HEXACO honesty-humility, i.e., the tendency of somebody (in a leadership position) to be insincere, unfair, greedy, and

<sup>2</sup> Self-other agreement on personality is much higher among close friends (r = 0.47 in Connelly and Ones, 2010, and r = 0.59 in de Vries et al., 2008), family members (0.49 and 0.62), and partners (0.58 and 0.69).

<sup>3</sup> In contrast to most leadership constructs, personality constructs are conceptualized using items that cover both poles of the constructs, suggesting that low levels of a trait (e.g., introversion) are the opposite of high levels of that same trait (e.g., extraversion). Operationalizations of dimensional constructs that include items from both poles have the advantage that they reduce response biases (Ashton et al., 2017). In this manuscript, traits are conceptualized as density distributions, i.e., the tendency to act in one way or another in terms of the likelihood/frequency of trait-related behaviors. For example, a dishonest leader may very well often act in an honest way. However, as I will explain in section The STOA Model of TNT Leadership and **Table 1**, such a leader will be more likely to act in a dishonest way than an honest leader if the trait gets activated in a situation that allows for deception.

immodest. Leader dishonesty may be especially problematic for organizations because it may induce, encourage, and/or exacerbate an unethical organizational culture with low trust, low satisfaction, and high turnover. Furthermore, when unchecked it may be associated with serious economic, organizational, and legal costs for an organization. In the popular press, much attention has been devoted to the serious negative effects of dishonest leader behaviors in cases such as the Enron, WorldCom, Volkswagen, and Bernie Madoff scandals, in which CEOs and/or CFOs acted fraudulent and/or condoned fraudulent behaviors. Although there is not much leadership research using HEXACO constructs, HEXACO personality research and leadership research using concepts related to low honesty-humility seem to support the deleterious consequences of leader dishonesty. In personality research, low honestyhumility has been found to be associated with higher levels of counterproductive work behaviors (Zettler and Hilbig, 2010; Wiltshire et al., 2014), workplace delinquency (Lee et al., 2005; de Vries and Van Gelder, 2015), and unethical business decisions (Ashton and Lee, 2008; de Vries et al., 2017). Unethical leadership, which is—when taking into account the self-other agreement problem (see above)—strongly negatively related to honesty-humility (de Vries, 2012), has been found to be related to a more unethical climate or culture (Demirtas and Akdogan, 2015; Eisenbeiss et al., 2015), higher levels of organizational units' deviance/unethical behaviors (Mayer et al., 2009, 2012), lower levels of Organizational Citizenship Behaviors (OCBs; Mayer et al., 2009), lower top team effectiveness (De Hoogh and Den Hartog, 2008), lower levels of trust in the supervisor (Chughtai et al., 2015), lower job satisfaction (Kim and Brymer, 2011; Palanski et al., 2014), lower affective commitment and effort (Brown et al., 2005), and higher intentions to quit (Palanski et al., 2014; Demirtas and Akdogan, 2015). The consequences of leader dishonesty thus seem to be vast, ranging from negative consequences for individual employees and teams to negative consequences for the entire organization's performance (Eisenbeiss et al., 2015).

Leader disagreeableness, the second TNT applied to leadership, is defined as the opposite of HEXACO agreeableness, i.e., the tendency of somebody (in a leadership position) to be unforgiving, overly critical, inflexible, and impatient. Leader disagreeableness may be problematic for organizations because it may induce a culture of fear and retaliation, which may, in turn, lead to high levels of job dissatisfaction, turnover, and costs associated with conflict management and conflict-related lawsuits. It is important to clarify that disagreeableness in the HEXACO model is more closely associated with reactive aggression (vs. reactive cooperation) than with instrumental or proactive aggression (vs. active cooperation). The former is somewhat more closely associated with HEXACO agreeableness, whereas the latter is somewhat more closely associated with honesty-humility (Book et al., 2012; Hilbig et al., 2013; Thielmann et al., 2014; Zhao and Smillie, 2015). Honestyhumility has been found to be more strongly related to premeditated rather than immediate revenge reactions, whereas agreeableness has been found to be fairly equally related to premeditated and immediate revenge reactions following transgressions (Lee and Ashton, 2012). Although it is difficult to extrapolate from Big Five agreeableness because it does not make a clear distinction between active and reactive forms of aggression, thus rendering it unclear whether the following applies to HEXACO agreeableness, teams with lower levels of agreeableness do seem to suffer from lower performance, lower levels of cohesion, more conflicts, and lower levels of workload sharing (Barrick et al., 1998; Peeters et al., 2006; Bell, 2007). In teams, persons with the lowest level of agreeableness seem to have the most negative impact; that is, the least agreeable person in a team has been found to have a greater negative effect on team outcomes than the average agreeableness of a team (Bell, 2007).

In leadership research, HEXACO agreeableness (and not HEXACO honesty-humility) was found to be by far the strongest predictor of leader supportiveness, a measure of the extent to which a leader is considerate toward his/her subordinates, willing to share power, and is non-despotic (de Vries, 2012), and thus leader disagreeableness seems to be associated with low leader support and high leader despotism. Apart from despotic leadership (De Hoogh and Den Hartog, 2008), several other leader constructs exist to measure concepts akin to leader disagreeableness, such as abusive (Tepper, 2000), autocratic/authoritarian (Lewin et al., 1939), destructive (Einarsen et al., 2007), and tyrannical (Hauge et al., 2007) leadership. Despotic leadership (De Hoogh and Den Hartog, 2008) has been found to be negatively related to job performance, OCB, and employee creativity (Naseer et al., 2016). Abusive supervision (Tepper, 2000; Tepper et al., 2017), which has been found to be most strongly related to Big Five agreeableness (Tepper et al., 2001), has been found to be related to higher levels of supervisor-directed, organizational, and interpersonal deviance (Mitchell and Ambrose, 2007; Tepper et al., 2008, 2009), lower levels of perceived interactional or procedural justice and lower levels of employees' OCB (Zellars et al., 2002; Aryee et al., 2007), lower job satisfaction (Tepper, 2000; Tepper et al., 2009), and higher psychological distress and emotional exhaustion (Tepper, 2000; Wu and Hu, 2009). In line with findings on abusive leadership, destructive and tyrannical leadership styles have also been found to be consistently related to negative follower and organizational outcomes (Schyns and Schilling, 2013). A related construct, but with a somewhat different focus, is the construct of autocratic (or: authoritarian) leadership (Lewin et al., 1939). Autocratic leadership, which is defined by unilateral leader decision making and intolerance of disagreement, has been found to result in lower levels of satisfaction (Gastil, 1994), higher levels of cynicism (Jiang et al., 2017), and higher levels of role conflict and role overload (Zhang and Xie, 2017). Probably mostly the intolerance of disagreement inherent in autocratic leadership is associated with higher levels of abusive supervision, making autocratic (i.e., authoritarian) leadership positively related to abusive supervision (r = 0.37; Mackey et al., 2017). Boys in Lewin et al. (1939) camp did not perform worse under an autocratic supervisor but reacted more dependent on him and exhibited higher levels of aggression and frustration once the autocratic leader became unavailable (White and Lippitt, 1960).

Note that most of the "abusive" constructs do not separate dishonesty from disagreeableness, and thus most—if not all are probably related to both leader dishonesty and leader disagreeableness. For instance, abusive leadership was found to be almost equally negatively related to HEXACO honestyhumility and agreeableness (Breevaart and de Vries, 2017), and thus it may be unclear, when investigating its effects, which consequences are due to leader dishonesty and which are due to leader disagreeableness.

Leader carelessness, the third of the TNT traits applied to leadership, is defined as the opposite of HEXACO conscientiousness, i.e., the tendency of somebody (in a leadership position) to be sloppy, lazy, negligent, and impulsive. Leader carelessness may be problematic for organizations, because it may be associated with an accident-prone culture, in which rules and regulations are disregarded and in which industry standards, necessary for optimal performance, are violated. More generally, it may lead to a culture in which low, instead of high, performance is the norm. When related to leadership, conscientiousness as a personality variable has been found to be most closely associated with task-oriented or structuring leadership (de Vries, 2012; Babiak et al., 2017), although relations with ethical leadership and leader consideration have also been noted (DeRue et al., 2011; Babalola et al., in press). One of the most notable characteristics of "careless" people with low levels of conscientiousness is their enhanced level of procrastination, i.e., their tendency to delay tasks that need to be done. In a meta-analysis by Steel (2007), procrastination was very strongly negatively related (r = −0.62) to conscientiousness. Another characteristic of carelessness is low levels of self-control. Of all personality traits, conscientiousness has been found to be by far the strongest correlate of self-control (e.g., r's > 0.50; de Vries and Van Gelder, 2013). A third characteristic of careless people is that they are more likely to make errors and to be involved in accidents because they are less motivated to follow safety regulations (Wallace and Vodanovich, 2003; Clarke and Robertson, 2005; Christian et al., 2009). Consequently, careless leaders are more likely to put things off until tomorrow which should be done today, they are more likely to lack a sense of urgency and discipline, they are more likely to make errors or let errors go unnoticed, and they are more likely to seek out pleasurable activities instead. Such a profile of low self-control, high procrastination, and high error proneness is probably best reflected in laissez-faire leadership. Meta-analyses seem to confirm a negative relation between conscientiousness and laisser-faire leadership (Bono and Judge, 2004; DeRue et al., 2011). In turn, task-oriented leadership and laissez-faire leadership have been found to be important predictors of outcome variables. That is, low task-oriented leadership has been associated with low levels of leader effectiveness (but not lower levels of job and leader satisfaction) and high levels of laissez-faire leadership has been associated with both low levels of leader effectiveness and low levels of job and leader satisfaction (DeRue et al., 2011).

One might question whether passive leadership such as laissez-faire leadership and lack of task-oriented leadership constitute such a liability to the organization to call leader carelessness a "nightmare trait." As Einarsen et al. (2007) argue, the answer should be an unequivocal "yes," because passive leadership not only constitutes shirking functional responsibilities, which can thus be considered stealing company time, but because it may also result in highly negative consequences for organizations when crucial errors are made or when important safety regulations are violated. Given the fact that passive leadership (cf. leader carelessness) has been strongly negatively associated with positive organizational outcomes (DeRue et al., 2011), it may be appropriate to label it following Einarsen et al. (2007)—as a destructive leadership style. Although too high levels of conscientiousness may be (but only slightly) "too much of a good thing" (Le et al., 2011), and too high levels of leader perfectionism may result in negative consequences associated with micromanagement, too high levels of leader carelessness seem to result in much worse outcomes in terms of decreased individual, team, and organizational effectiveness.

# COMBINING THE TNT WITH EXTRAVERSION, EMOTIONALITY, AND OPENNESS TO EXPERIENCE

The three remaining HEXACO dimensions, extraversion, emotionality, and openness to experience, do not seem to be associated to the same degree with negative leadership outcomes as the TNTs (but see Judge et al., 2009 for possible negative leadership outcomes associated with either low or high extraversion, emotional stability, and openness to experience). However, in some instances, combinations of the three remaining traits with the TNT may be associated with even worse outcomes. The most important of the remaining traits is extraversion. Extraversion is one of the most robust correlates of leader emergence, transformational/charismatic leadership, and leadership effectiveness (Judge et al., 2002; Bono and Judge, 2004; de Vries, 2012). However, in combination with leader dishonesty, leader disagreeableness, and leader carelessness, an extravert leader may turn out to be even more destructive, showing characteristics of what has been called a personalized (i.e., self-aggrandizing, non-egalitarian, and exploitative) charismatic leader (McClelland, 1975; House and Howell, 1992), who misuses his/her charisma and dominance to obtain personal goals at the expense of others. Interestingly, House and Howell (1992) described in detail the pattern of personalized charisma using narcissism, Machiavellianism, and authoritarianism traits that are associated with leader dishonesty and leader disagreeableness. Together with psychopathy, narcissism and Machiavellianism form the so-called dark triad, which are associated with grandiosity, entitlement, and feelings of superiority (narcissism), manipulativeness and deception (Machiavellianism), and antisocial tendencies, glibness, lack of empathy, and irresponsibility (psychopathy). Recently, a fourth trait, sadism, has been added to the dark triad to form the dark tetrad (Chabrol et al., 2009; Buckels et al., 2014), the core of which is formed by the enjoyment of physical and/or emotional pain in innocent others through aggressive and/or cruel acts.

Although the dark triad (and tetrad) have been found to be related to especially low agreeableness in the Five-Factor Model (FFM; O'Boyle et al., 2015), the most important correlate of the dark triad (and tetrad) is HEXACO honesty-humility. Through the inclusion of honesty-humility, the HEXACO model has been able to outperform the Big Five model (or: FFM) in the explanation of not only the dark triad (Lee and Ashton, 2005, 2014), but also the dark tetrad (Book et al., 2016). Although the common core of the dark triad/tetrad traits, which are generally strongly related to each other, is formed by honesty-humility, each of the dark traits have some residual relations with other HEXACO traits. That is, besides honesty-humility, narcissism has also been found to be positively related to extraversion, Machiavellianism negatively to agreeableness, and psychopathy negatively to emotionality and to conscientiousness (Lee et al., 2013). Sadism has been found to be most closely related to low honesty-humility and low emotionality, but also (but less strongly) low agreeableness and low conscientiousness (Book et al., 2016).

The core of these dark triad/tetrad traits thus seems to be formed especially by low honesty-humility (i.e., dishonesty), but also somewhat low agreeableness (i.e., disagreeableness) and low conscientiousness (i.e., carelessness). A profile that combines high levels of extraversion with leader dishonesty is indicative of leader narcissism whereas a profile that combines low levels of emotionality with the TNT (i.e., leader dishonesty, leader disagreeableness, and leader carelessness) is indicative of psychopathic leadership. Consequently, the most "dangerous" leaders seem to be those leaders who combine the TNT traits with high extraversion and low emotionality, resulting in a narcissistic-psychopathic leadership profile.

It is somewhat less clear what the results may be of a leader profile, which combines the TNT with low or high openness to experience. Openness to experience, like extraversion, has been found to be positively related to leader emergence, leader charisma, and leader effectiveness (Judge et al., 2002; Bono and Judge, 2004), and thus it may be true that, just like extraversion, high openness to experience strengthens the negative effects of the TNT on individual, team, and organizational outcomes. On the other hand, high openness to experience, when expressed through new ideas and methods, may also distract or even compensate for some of the negative effects associated with the TNT.

Although there is, at present, not much evidence on profiles that combine the TNT with the other three personality traits, some studies suggest that outcomes may be worst when combining low honesty-humility with extraversion. For instance, Gylfason et al. (2016) found that respondents high on extraversion and low on honesty-humility were most likely to send deceiving messages in a "cheap talk" game. Similarly, in two of the three samples investigated, Oh et al. (2011) found that extraversion and honesty-humility interacted in the prediction of workplace deviance, such that the highest level of workplace deviance was observed for those high on extraversion and low on honesty-humility. Furthermore, narcissistic leadership, a leadership style which combines high extraversion with low honesty-humility, has been found to be associated with problematic organizational and/or societal outcomes, such as higher levels of tax evasion (Olsen and Stekelberg, 2016), higher numbers of lawsuits (O'Reilly et al., 2018), higher levels of actual fraud (Rijsenbilt and Commandeur, 2013), and more volatile and extreme (both negative and positive) return on assets (Chatterjee and Hambrick, 2007).

With respect to psychopathic leaders (i.e., those leaders who combine the TNT with low emotionality), Babiak et al. (2010) found—based on observational ratings—that 5.9% of their sample consisting of managers and executive had a "potential psychopathy" score. Although, based on 360◦ ratings, these managers were perceived to have good communication skills and innovative ideas (indicative of respectively high extraversion and high openness to experience), psychopathy scores correlated negatively (r = −0.41) with supervisory performance ratings. That is, although people with psychopathic profiles were able to successfully climb the corporate ladder, probably due to their high extraversion and high openness to experience, they were found to have a negative impact on the team and the organization when considering their performance evaluations.

# THE STOA MODEL OF TNT LEADERSHIP

Whether and how people emerge as leaders, act as leaders, and are effective as leaders, can only be ascertained by taking situational contexts into account. People act on, in, and through situations, and thus any model that describes leadership needs to also describe how the personality of leaders "unfolds," i.e., what situations (potential) leaders seek out, in what way they behave in these situations, and what the effects are of their behaviors. The STOA model posits three activation mechanisms that describe the way personality unfolds: (1) a situation activation mechanism, (2) a trait activation mechanism, and (3) an outcome activation mechanism (de Vries et al., 2016b). First of all, based on their personality, people perceive, select, manipulate, and/or evoke situations to "fit" their personality (Buss, 1987, 2009). To become a leader, persons have to first of all select situations that afford them to become a leader. People who avoid social settings, because they feel less comfortable in groups or because they are less interested in social situation, are unlikely to become leaders in the first place. People low in extraversion and high in emotionality/anxiety are not only less interested in social situations (Holtrop et al., 2015), with extreme levels of these traits they may also be more likely to actively avoid such situations because of social phobia (Kotov et al., 2010). Highly extraverted people, in contrast, seek out social situations, not only because such situations are rewarding or because they like social occasions, but especially because they seek social attention (Ashton et al., 2002). Thus, by virtue of their personality, extraverted people are more likely to seek out situations in which they can fulfill a leadership role.

Social situations, in turn, afford the expression of leadershiprelated traits. Trait activation, the second of the proposed mechanisms, is predicated on trait activation theory (TAT; Tett and Burnett, 2003), which maintains that traits only get activated when situations allow these traits to be expressed. Social situations may activate several traits, but for leadership, especially three personality dimensions seem to be most relevant: extraversion, conscientiousness, and openness to experience. These are the three personality dimensions that have been found to be most strongly and positively related to leader emergence (Judge et al., 2002; Ilies et al., 2004; Reichard et al., 2011). These are also the three personality dimensions that have been found to be most strongly related to proactive personality (a.k.a. proactivity, see de Vries et al., 2016c), which includes taking charge, networking, voice behaviors, and career initiative; behaviors that can only be expressed in social situations and that are viewed as indicating leader potential (Fuller and Marler, 2009). With respect to extraversion, especially social boldness may play a role. People who are socially bold are more likely to take charge in groups. With respect to conscientiousness, especially diligence and organization may play a role. People who are diligent and organized, work hard and plan carefully in order to have a better chance to reach their goals; traits that also seem to help groups to become successful (Peeters et al., 2006; Bell, 2007). With respect to openness to experience, especially creativity, and innovativeness may play a role, behaviors that may help groups distinguish themselves through new and original solutions. People who have such a profile of high extraversion, high conscientiousness, and high openness to experience are likely to be viewed as an important asset to a group (i.e., obtain "idiosyncrasy credits;" Hollander, 1992), and are consequently more likely to emerge as a leader.

Two other traits that have been proposed to be relevant to leader emergence and that may be activated in social situations are narcissism and self-monitoring. As noted above, narcissism has been found to be related to both (low) honesty-humility and (high) extraversion (Lee et al., 2013). Several studies have argued that narcissism is related to leader emergence (Paunonen et al., 2006; Nevicka et al., 2011), even when correcting for Big Five extraversion (Brunell et al., 2008), suggesting that low honestyhumility (especially low modesty) may play a role. Similar to narcissism, self-monitoring has also been found to be related to (low) honesty-humility and (high) extraversion (Ogunfowora et al., 2013), and also similar to narcissism, self-monitoring has been found to be positively related to leader emergence (Ellis, 1988; Zaccaro et al., 1991). Furthermore, Foti and Hauenstein (2007) found that a pattern that combined high levels of (social) dominance (which has been conceptualized as a facet of extraversion; e.g., Lee and Ashton, 2004), intelligence, selfefficacy, and self-monitoring had the strongest correlation with peer and superior ratings of leadership impressions. However, a recent meta-analysis on the relation between narcissism and leader emergence found that, when correcting for extraversion, the positive relation between narcissism and leader emergence turned to near zero (Grijalva et al., 2015). Because selfmonitoring relates to extraversion as well, it looks as though variance associated with extraversion is the only real and substantial correlate of leader emergence in these two traits.

Outcome activation, the third of the proposed mechanisms, pertains to the effects that activated traits have. Three kinds of effects may be distinguished: (1) recognition, (2) perception, and (3) attribution. In the first place, one of the main outcomes of socially bold, disciplined/organized, and/or creative behaviors is that group members take notice. That is, people only get "recognized" as a potential leader if they show prototypical leader behaviors. Second, the more a person acts socially bold, disciplined/organized, and/or creative, the higher the chance that group members act upon that person's suggestions, which strengthen leadership perceptions. And third, if—by following the suggestions of somebody who shows prototypical leader behaviors—a group becomes successful, the results are likely to be attributed to the person who has shown leaderlike behaviors, resulting in even stronger leadership perceptions (cf. the Romance of Leadership theory, Meindl, 1995; see also de Vries, 2000). In general, holding everything else constant, socially bold, disciplined, and creative behaviors (i.e., proactivity) are more likely to result in positive outcomes for a group than behaviors that are their opposites (i.e., socially phobic, unorganized, and uncreative). That is, proactive personality has been shown to be one of the most important predictors of job performance and business success (Rauch and Frese, 2007; Fuller and Marler, 2009; Thomas et al., 2010).

Apart from conscientiousness, the main drivers of leader emergence thus appear to be traits that are not aligned with the TNT. However, apart maybe from carelessness (i.e., low conscientiousness) which may be associated with higher number of mistakes Wallace and Vodanovich, 2003; Clarke and Robertson, 2005; Christian et al., 2009, there does not seem to be anything in the two remaining TNT traits, i.e., dishonesty and disagreeableness, that prevents people who exhibit these traits to rise through the ranks and to obtain a leadership position. Elsewhere (de Vries, 2016; de Vries et al., 2016b), it has been argued that some situations are sought out by people who are characterized by higher levels of dishonesty, disagreeableness, and carelessness because these types of situations allow people to more readily express these traits, free from constraints. That is, people high on dishonesty are more likely to seek out situations that allow for exploitation (Sherman et al., 2015), because in such situations they can more readily express dishonest behaviors (Hilbig and Zettler, 2009; Hilbig et al., 2012) and because in such situations, they are more likely to obtain "sex, power, and money" (Lee et al., 2013). People high on disagreeableness are more likely to pay attention to negative events (Bresin and Robinson, 2015) and seek out situations that allow for (interpersonal) obstruction (Rauthmann, 2012; de Vries et al., 2016b), and are consequently more likely to have relationship conflicts (Bono et al., 2002). Disagreeableness may result in positive outcomes for a person if s/he has enough power and status to get more easily what s/he wants using disagreeable behaviors (Sell et al., 2009). Last of all, people high on carelessness are more likely to seek out situations in which they can shirk duties and avoid planning and goal-setting, because especially in situations in which they have to set goals and perform (e.g., in most school and organizational settings), carelessness (i.e., low conscientiousness) is associated with lower performance (Barrick and Mount, 1991; Poropat, 2009).

In fact, some studies suggest that norm violating behaviors (i.e., dishonest, disagreeable, and/or careless behaviors) may be perceived as leaderlike, because they suggest to others that the norm violator has the power to act free from social constraints (Van Kleef et al., 2011). Power derived from a leadership position, in turn, may free people to "do as they please" (Galinsky et al., 2008), resulting in a greater likelihood to express norm violating behaviors. That is, individuals high on the TNT are more likely to seek out situations in which they can freely express counternormative traits (situation activation). Combined with high levels of extraversion (social boldness) and openness to experience (creativity), the expression of TNT behaviors may be perceived as more leaderlike, which may make it more likely for them to emerge as a leader. In turn, when they have power, TNT leaders may feel less constrained, resulting in more frequent and open expression of the TNT (trait activation), which may result in positive outcomes for the self in terms of "sex, power, and money" (outcome activation; Lee et al., 2013), especially when there are no countervailing powers (i.e., checks and balances). Evidence of the STOA mechanisms is, for instance, found in a study by Wisse and Sleebos (2016), who observed a positive relation between supervisor-rated Machiavellianism and his/her perceived position power (indicative of both situation activation and outcome activation) and who found that Machiavellianism interacted with perceived position power in the prediction of subordinate-rated abusive supervision. That is, abusive supervision of Machiavellian leaders was higher when the supervisor had more position power. Together with the finding that Machiavellianism is positively related to career success in terms of a (higher) leadership position (Spurk et al., 2016), the results seem to suggest that norm violation may indeed be beneficial for perpetrators.

# NIGHTMARE CAREERS

What should organizations do when faced with a TNT leader? And are there ways to prevent TNT leaders to rise through the ranks? In the following, I'll use an extended version of the ASA model of Schneider (1987), including six (instead of Schneider's three) career phases, i.e., attraction, selection, socialization, production, promotion, and attrition, to describe possible actions organizations can take to prevent TNT applicants for leadership positions to become—in the end—TNT CEOs. Following de Vries (2016), attraction is the phase in which recruitment efforts take place, selection the phase in which a candidate is chosen from the available applicant pool, socialization the phase in which a new leader formally and informally gets to know his/her team and organization, production the phase in which a leader performs in his/her job, promotion the phase in which a leader qualifies for an even higher-level position, and attrition the phase in which a contract is (voluntarily or involuntarily) terminated. In **Table 1**, an overview is offered of the TNTs, in what situations these traits are activated, what possible negative outcomes are associated with the TNTs, and what organizations can do to prevent situation, trait, and outcome activation of these traits among leaders<sup>4</sup> .

### Attraction

To attract employees for leadership positions, firms are likely to use a great number of recruitment channels to find motivated candidates (Russo et al., 2000). From the perspective of the recipients of the recruitment messages, these messages may either generate interest in the organization or not. In terms of the STOA model, situation activation is the main mechanism in the attraction phase. Prospective employees are mainly attracted to organizations based on the perception of the nature of work and the organizational culture (Boswell et al., 2003; Chapman et al., 2005). Whereas vocational interests are the most important determinant of vocational (job) choice (Tracey and Hopkins, 2001; Volodina and Nagy, 2016), which plays a role in the earlier phase of a career, personality may play an important role in determining organizational culture preference in later career stages. Only few studies have been conducted on the relations between personality and organizational culture preference, and none have been conducted using the HEXACO model, but findings do suggest that personality plays an important role in line with the TNT described above. That is, of all relations explored between self- and peer-reported personality and selfreported organizational culture preference, Judge and Cable (1997) found agreeableness to be the most important negative predictor of an aggressive organizational culture preference, suggesting that people with a high level of TNT disagreeableness are more likely to apply for an organization which is more likely to condone aggression. The second most important relation was between conscientiousness and preference for an outcomeoriented culture, suggesting that careless people are more likely to apply for an organization that is less outcome-oriented. In a sample of students, attractiveness of a sales job with "out of town travel" was highest among students with low conscientiousness and low agreeableness (Stevens and Macintosh, 2003), suggesting that careless and disagreeable people are more likely to apply for organizations that offer these types of "away-from-work" fringe benefits. With respect to dishonesty, low scorers on honestyhumility are motivated by wealth, privilege, and status (Lee and Ashton, 2004), so it may seem logical to assume that organizations that "flaunt" these kinds of characteristics, are more likely to be attractive to dishonest people. Empirical evidence suggests that this is indeed the case; i.e., people low on honestyhumility are more likely to be attracted to power and money than people high on honesty-humility (Lee et al., 2013). Furthermore, Ogunfowora (2014) found that people low on honesty-humility, but not people high on honesty-humility, were more likely to be attracted to an organization with a CEO who was morally questionable.

### Selection

From an organizational perspective, trait activation is the most important mechanism in the personnel selection phase. In this phase, organizations provide candidates with situations (e.g., questions in interviews and selection assessments) that activate traits and skills that are deemed relevant by the organization. With respect to the TNT, there is convincing evidence that especially carelessness (i.e., low conscientiousness), but also dishonesty (low honesty-humility) are associated with higher

<sup>4</sup>This Table is an—for TNT leadership—adapted version of Table 1 in de Vries (2016).

TABLE 1 | Implications of the TNT for attraction, selection, socialization, production, promotion, and attrition in organizations.


TABLE 1 | Continued


/ : Actions of the organization that may strengthen/weaken nightmare traits.

counterproductive behaviors and lower job performance (e.g., Barrick and Mount, 1991; Ones et al., 1993, 2007; Ones and Viswesvaran, 1998; Schmidt and Hunter, 1998; Sutin et al., 2009; Fine et al., 2010; Brown et al., 2011). Although there is no evidence for the negative effects of disagreeableness from personnel selection studies, team studies seem to suggest that one disagreeable team member can have a strong negative effect on team cohesion (Barrick et al., 1998; Peeters et al., 2006; Bell, 2007; O'Neill and Kline, 2008), which may be exacerbated when the team leader is disagreeable. Apart from the TNT, extraversion has been found to be related to career success and leadership position, in part because of its relation with perceptions of charisma (Vergauwe et al., 2017). As argued above, although organizations might like to select on extraversion to recruit potential leaders, especially the combination of extraversion with the TNT may have negative consequences for an organization. Both narcissism (indicative of leader dishonesty and extraversion) and Machiavellianism have been found to be associated with positive career outcomes for the employee him-/herself, such as higher salary (narcissism) and higher leadership position (Machiavellianism) (Spurk et al., 2016), but mostly negative outcomes for the organization (Spain et al., 2014). Consequently, doing a thorough background check and making sure that the selection procedure allows the measurable expression of the TNT through (reliable and valid) structured interviews, questionnaires, or assessment tools, seems to be important to select non-TNT leadership candidates and, consequently, to prevent potential toxic organizational consequences.

### Socialization

In the socialization (or: onboarding) phase, new employees (including those who applied for a leadership position) get to evaluate the actual level of trait activation and outcome activation that the job and the organization offer. This phase is important for the establishment of a psychological contract (Kotter, 1973), an informal set of reciprocal expectations between an employee and his/her organization. These expectations cover the kind of behaviors that are allowed and/or expected at work and the kind of outcomes expected of an employee. Based on these informally and/or formally communicated expectations, new employees/leaders learn whether the organization affords or constrains TNT-based behaviors and what outcomes result from such expressions of the TNT. An example of an onboarding activity is ethics training. Although the effect of limited ethics training has been found to be transient (Richards, 1999), more exhaustive and in-depth ethics training has been found to have a longer lasting effect on ethical decision-making (Mumford et al., 2008) and to have a positive effect on the perceived ethical culture of an organization (Valentine and Fleischman, 2004). Although it is unlikely that ethics training changes a person's personality, it does make an employee aware of the norms and values of an organization, which may limit the expression of nightmare traits (i.e., prevents trait activation) and which may limit expectations that positive outcomes may result from the expression of nightmare traits (i.e., prevents outcome activation). Because the socialization phase for leadership positions is often short and new leaders are often expected to make changes to their team and/or organization, a potential danger is that ethics training or attention to ethical dilemmas have limited effect and that the first thing TNT leaders do is to try to make their mark by changing the culture of the organization to fit their personality.

# Production

All three STOA mechanisms play a role in this phase. That is, a new leader is likely to try to seek out certain organizational situations and/or to change them to fit his/her personality (situation activation), these situations are likely to activate (combinations of) his/her traits (trait activation), which may result in positive and/or negative outcomes for him/her and/or for the organization (outcome activation). For organizations, job crafting (Wrzesniewski and Dutton, 2001) may take on a negative meaning when employees and leaders with a TNT profile use job crafting to adapt their job and the organization to their personality. That is, during this phase, a TNT leader is likely to want to find a personal niche in the organization or to change his/her job and organization for egocentric reasons (a) in order to enrich him-/herself (dishonesty), (b) in order to have no restrictions when dealing with people who oppose him/her (disagreeableness), and (c) in order to be unhampered by rules, regulations, plans, and goals (carelessness). As an example, narcissistic (i.e., high extraversion and low honesty-humility) CEOs have been found to be able to increase the earning gap

between them and the other top managers in their team (O'Reilly et al., 2014).

What can an organization do to prevent TNT leaders from inflicting harm on the organization? First and foremost, surveillance and an ethical culture have been found to be negatively related to delinquent work behaviors (de Vries and Van Gelder, 2015). Top managers' ethical leadership was found to have a "trickle-down" effect through supervisory ethical leadership on employees' OCBs, organizational commitment, and reduced deviance two hierarchical levels down (Mayer et al., 2009; Ruiz et al., 2011). The reverse is also true. Abusive management was found to have a trickle-down effect on employees two hierarchical levels down, such that work group interpersonal deviance was higher in employees when management used an abusive leadership style (Mawritz et al., 2012). Second, activation of the TNT is more likely when TNT behaviors are rewarded. Compared to people high on honesty-humility, people low on honesty-humility were more likely to cheat or to contribute less to a public good when there was no chance of being caught (Hilbig and Zettler, 2015) and when punishment for uncooperative behaviors was unlikely (Hilbig et al., 2012). Additionally, people high on honestyhumility were more likely to be cooperative than people low on honesty-humility when others were cooperative as well (Zettler et al., 2013). Thus, when higher management sets an ethical example, supports virtuous behaviors, and makes sure negative consequences result from counterproductive (TNT) behaviors, it is less likely that TNT—and more likely that virtuous—behaviors are activated.

# Promotion

Promotion is an important outcome for those with a TNT profile, because higher positions are more likely to be accompanied with a higher income and more status, power, and autonomy (outcome activation), which are associated with fewer constraints on trait expression (Galinsky et al., 2008). Especially those low on honesty-humility are more likely to use impression management techniques (e.g., ingratiating superiors) in politicized organizations, which may ultimately help them to advance (Wiltshire et al., 2014). The higher the position, the more harm a TNT leader can do to the organization, and thus the more important it is to have adequate promotion selection mechanisms in place.

For promotion the same applies as for selection, but generally more information about the person from within the organization is available during a promotion trajectory, and thus in theory it should be easier for an organization to determine whether the TNT are present or not. However, during this phase, the organization can mistakenly interpret TNT behaviors in terms of leadership attributes, i.e., leader dishonesty in terms of "cunningness," leader disagreeableness in terms of "toughness," and leader carelessness in terms of "willingness to delegate." Furthermore, the organization can mistakenly only rely on supervisory instead of 360◦ reports. Whereas TNT leaders are less likely to let their supervisors become aware of dishonest, disagreeable, and possibly even careless behaviors, subordinates are more likely to be confronted with such behaviors. Ambition, which is related to career success and a higher income (Ashby and Schoon, 2010; Judge and Kammeyer-Mueller, 2012), is often regarded as a positive attribute, but it may also be indicative of greed, a facet of low honesty-humility (Lerman, 2002; Lee and Ashton, 2004). Arguably, organizations should select on humility instead. Leader humility has been found to improve interpersonal team processes, which, in turn, has been found to result in greater team performance (Owens and Hekman, 2016).

# Attrition

Another possible outcome of the TNT is a person's voluntary or involuntary attrition (outcome activation). Meta-analyses and longitudinal studies have shown that job performance is negatively related to turnover (McEvoy and Cascio, 1987; Williams and Livingstone, 1994; Griffeth et al., 2000; Zimmerman and Darnold, 2009), and this relation seems to be true even when turnover is involuntary (Shaw et al., 1998). Consequently, organizations seem to rely to some extent on job performance indicators to discharge dysfunctional personnel. However, as noted above, some TNT employees, such as psychopaths, seem to be found relatively frequently in the boardroom (Babiak et al., 2010), suggesting that not all organization are able to adequately deal with low performing managers. Research suggests that organizations that have a highly develop HR system with high selection rates (Shaw et al., 1998) and performance-contingent rewards (Williams and Livingstone, 1994; Griffeth et al., 2000) have a stronger relation between job performance and turnover, and thus more extensive HR systems may be associated with a reduced chance for TNT employees to turn into TNT boardroom members.

# HOW BAD IS TNT LEADERSHIP?

Are TNT leaders uniformly bad? And how bad are they? In the following section, I'll discuss (a) possible situations in which nightmare traits may have positive consequences and (b) whether "bad is stronger than good" when talking about leader nightmare traits.

According to some authors, Dark Triad traits (psychopathy, narcissism, and Machiavellianism) in leaders may be beneficial in some contexts (Judge et al., 2009; Spain et al., 2016). According to Judge et al. (2009), the strategic and flexible use of people, resources, and influence tactics by Machiavellian leaders may be associated with positive outcomes for themselves and for their followers. For instance, Machiavellianism among US presidents has been found to be positively related to rated performance (Deluga, 2001) and to the number of legislative achievements (Simonton, 1986). When operating in a corrupt environment, it may be impossible to rise through the ranks and be effective as a leader without being tainted by corruption. For instance, in a case study of political leadership in Lebanon, Neal and Tansey (2010) showed that Rafik Hariri could only rebuild Beirut with "effective corrupt leadership." In some instances, narcissism has also been equated with greatness. When a workrelated area is important for their self-esteem, narcissists may be especially strongly motivated to do their best (Harms et al., 2011). Furthermore, narcissists are more likely to favor attentiongrabbing, big, and bold actions; actions that may result in large gains or large losses (Chatterjee and Hambrick, 2007). And finally, managers and executives with psychopathic profiles, although less positively rated on performance and management style, were found to be rated more positively than those lower on psychopathy on communication skills and strategic thinking (Babiak et al., 2010).

The findings on the Dark Triad, however, need to be treated with caution, because some of these findings may be indicative of the effects of other trait dimensions than the TNT. That is, some of the positive effects noted above may be due to the positive effects of extraversion or cognitive abilities instead. It may be true that only highly extravert and intelligent TNT leaders are able to make it to the top, being able to adequately "neutralize" the accusations and conflicts that they encounter on the way up. As noted above, the effects of narcissism on leader emergence disappeared once the effects of extraversion were controlled (Grijalva et al., 2015). Similarly, potential positive effects of narcissism on leader effectiveness may disappear when controlled for extraversion. Note that earlier, I argued that extraversion may aggravate the relations between the TNT and outcomes. Some of these negative (fraudulent, self-enhancing, chaotic) effects may be especially apparent when the environment is conducive of such leadership (Padilla et al., 2007) but not when sufficient checks and balances are in place to control for the toxic effects of TNT leadership. When sufficient checks and balances are in place, extraversion may account for most if not all of the leadership effects, which may thus turn out to be positive (Judge et al., 2002) rather than negative.

Because the Dark Triad are most strongly related to (low) honesty-humility, these findings may indicate that in some circumstances leader dishonesty may have positive consequences, although it is questionable whether the results are as positive for the team, organization, or society as they are for the leader him-/herself. With respect to leader disagreeableness, it may be an effective conflict strategy for a powerful leader (Sell et al., 2009), although it is questionable whether the shortterm gains associated with leader disagreeableness are not offset by long-term losses, associated with higher levels of task and relationship conflicts (Bono et al., 2002; De Dreu and Weingart, 2003; De Wit et al., 2012). With respect to leader carelessness, one might argue that some leaders might get their work done by delegating responsibilities, especially in "mature" teams (Hersey and Blanchard, 1996). But even such delegation would entail an active instead of a careless or laissez-faire response of the leader, the latter which is generally found to be generally ineffective (Einarsen et al., 2007). Thus, although in some specific contexts (e.g., in corrupt environments, when resolving a conflict in a powerful position, and/or when dealing with a "mature" team), leader dishonesty, disagreeableness, and carelessness may have less negative or even somewhat positive consequences, overall the effects of TNT leadership seem to be mostly negative.

Is "bad stronger than good" when applied to leadership? Baumeister et al. (2001) have argued that bad events have a stronger effect than good events and that this holds across a broad range of psychological phenomena. It is welldocumented that ethical, transformational, supportive, and instrumental leadership are positively related to individual and organizational outcomes such as subordinate satisfaction and team or organizational effectiveness (e.g., Judge and Piccolo, 2004; Judge et al., 2004; Burke et al., 2006; Dumdum et al., 2013). But what does this entail for the nightmare traits? Some scholars have compared the effects of constructive leadership styles (e.g., individualized consideration) with destructive leadership styles (e.g., abusive supervision) but did not find support for the "bad leadership is stronger than good leadership" notion (Schyns and Schilling, 2013; Brandebo et al., 2016). However, to conclude, based on these studies, that bad is not stronger than good, may be premature. For a proper investigation of this notion, it is not adequate to compare the effect sizes of constructive and destructive operationalizations of leadership. Instead, one should compare the effects of both constructive and destructive operationalizations of leadership at the negative pole of outcomes with those at the positive pole of outcomes. For instance, one should investigate whether those who have a leader low on constructive leadership (or: high on destructive leadership) suffer more from the negative consequences (when compared to a neutral position) than those who have a leader high on constructive leadership (or: low on destructive leadership) gain from the positive consequences (when compared to a neutral position). That is, good and bad leadership should be treated as a bipolar continuum, in which gains from the "positive" pole are compared to losses from the "negative" pole to find out whether bad is stronger than good.

# CONCLUSIONS AND DISCUSSION

Surprisingly enough, given the similar background, items, and genetic origin of leadership styles and personality traits and given the fact that leadership behaviors are a subset of behaviors referred to in personality models, only relatively few scholars (Kirkpatrick and Locke, 1991; Hogan and Kaiser, 2005; de Vries, 2008, 2012; Judge et al., 2009; Antonakis et al., 2012; Zaccaro, 2012) have called for a closer integration of leadership and personality research. Even though personality perspectives on leadership have been around for some time (e.g., Stogdill, 1948), a unifying perspective is still lacking. Especially when considering the overwhelming number of (dark) leadership styles that have been proposed, an integration of these two perspectives is more than ever needed. In this article, I suggest that an integration of the dark side of leadership with personality can be achieved by considering three so-called nightmare traits, leader dishonesty, leader disagreeableness, and leader carelessness. First of all, I have argued that commonly used leadership styles can be considered contextualized personality traits. Operationalizations of (dark) leadership styles are highly similar to operationalizations of personality, albeit in a contextualized format. Second, I have shown that low levels of three HEXACO traits, honesty-humility, agreeableness, and conscientiousness, underlie the main negative effects of the destructive leadership styles proposed in the

literature (e.g., abusive, despotic, authoritarian, laissez-faire, etc. leadership). Third, I have argued that these TNTs, when combined with high extraversion and low emotionality, may have even greater destructive effects (cf. the effects of psychopathicnarcissistic leadership). Fourth, I have introduced the STOA model to account for the process by which the nightmare leadership traits manifest themselves. Fifth and subsequently, I have used the STOA model to delineate the actual effects of TNT leadership in organizations and how to react to them throughout six career phases, i.e., attraction, selection, socialization, production, promotion, and attrition. And finally, I have discussed potential positive effects of the TNT and whether bad leadership is stronger than good leadership.

Although great strides have been made in our understanding of personality and (nightmare) leadership, there are still several research gaps to be filled. First of all, research is warranted which integrates leadership styles—or leadership-contextualized personality—with non-style leadership research, such as research on leader (emotional) intelligence (Cavazotte et al., 2012), leader expertise (Podsakoff et al., 1983), and motivation to lead (Chan and Drasgow, 2001). Whereas cognitive ability has been found to be by-and-large unrelated to personality (Joseph and Newman, 2010), intelligence has been found to be related to general perceptions of leadership (Lord et al., 1986) and to perceptions of transformational leadership (Cavazotte et al., 2012), although ability-based emotional intelligence has not been found to be related to transformational leadership when ratings were derived from different sources (Harms and Cred, 2010). Furthermore, personality—especially extraversion and agreeableness—has been found to be related to the motivation to lead (Chan and Drasgow, 2001). A further integration of leadership-contextualized personality (or leadership styles), competence, motivation, and affect perspectives on leadership is warranted to explain specific leader behaviors and outcomes. Such an integration necessitates largescale multi-time, multi-methods, multi-raters generalizability studies (Shavelson et al., 1989) to disentangle different sources of variance and to estimate the strength of the relations between leaders' contextualized personality/style, competence, motivation, affect, specific behaviors, and outcomes.

Second, a great number of leadership scales, and especially those that pertain to "dark styles" are problematic because they are highly (negatively) evaluative and pertain to low-base behaviors (e.g., Breevaart and de Vries, 2017). It is known, among others based on studies on low base-rate personality disorders, that answers to items on evaluative scales are more biased than answers to more neutrally formulated items (de Vries et al., 2016a; Ashton et al., 2017). Thus, when creating a contextualized leadership version of the main (HEXACO) personality dimensions, each dimension should preferably be represented by a matched number of positive and negative formulated items, reducing response biases typically observed in answers to leadership questionnaires.

Third, when such a contextualized leadership questionnaire is created, it will be better feasible to disentangle the relative effects of leader dishonesty, leader disagreeableness, and leader carelessness on leader effectiveness and subordinate outcomes. Self-other agreement tends to be higher on personality traits than on leadership styles (de Vries, 2012) and so a first question would be whether this is also true for contextualized leadership scales. Additionally, affect and liking has been found to be strongly related to leader ratings (Brown and Keeping, 2005), so a second question would be whether target variance is increased and relationship variance is decreased in contextualized leadership scales when compared to commonly used leadership instruments (Livi et al., 2008; de Vries, 2010). Furthermore, when using different sources, the next main question would be whether contextualized—and more neutrally formulated—leadership scales are better able to predict important outcomes than existing instruments.

Fourth, with respect to the TNT and the three non-TNT dimensions, an important question would be whether TNT and non-TNT scales interact in the explanation of leadership outcomes. By combining the TNT, non-TNT, and Dark Triad/Tetrad in one analysis, it is also possible to determine whether the effects of the Dark Triad/Tetrad variables are just due to the TNT or to a combination of TNT with non-TNT variables. If the latter is the case, a follow-up question is whether profiles that combine the TNT with high levels of extraversion and low levels of emotionality are more likely to result in worse outcomes for organizations than profiles that combine the TNT with low levels of extraversion and high levels of emotionality. Such an analysis may be problematic, because it would also need to resolve whether checks and balances interact with the outcomes of such profiles. The expectation would be that especially in contexts in which there are insufficient checks and balances, TNT leadership, combined with high extraversion and low emotionality, is especially explosive. Furthermore, investigations of the effects of such profiles over time (i.e., when do the effects of the TNT unfold, and are narcissistic leaders well-liked at first only because of their higher levels of extraversion?) and the differential effects of the TNT on subordinates, colleagues, and supervisors, would greatly help delineating the circumstances in which TNT leadership has the strongest impact.

Fifth, such research would be greatly helped if we could find out what organizations in which industries are more likely to be attractive to TNT applicants to leadership positions. In line with the STOA model, I have argued that organizations that offer greater opportunities for quick advancement, freewheeling, and quick monetary gains, which are slack on goal-setting and planning, which have a lower levels of surveillance, and which see harsh treatment as a sign of leadership, are more likely to be attractive to TNT leaders because such organizations fully allow them to freely express their traits and to gain desirable outcomes from these traits. The HR department in organizations might benefit from a full analysis of each of their career stages in order to find out whether they attract, select, socialize, promote, or (fail to) attrite TNT leaders.

Sixth and finally, more research needs to be carried out to distinguish circumstances in which TNT leadership may play a positive role and whether "bad" leadership is really worse than "good" leadership. As argued above, the latter should be investigated using another design than a design in which the effect sizes of destructive leadership styles are compared to the effect sizes of constructive leadership styles (Schyns and Schilling, 2013; Brandebo et al., 2016). Note, however, that this might be hard to ascertain, because one would have to carefully delineate what a "neutral" leadership effect is and what the objective costs and benefits of destructive and constructive leadership styles are.

There are certain aspects in our current time that seem highly beneficial for TNT leaders in organizations, i.e., in a global world, it is easier to select niches that allow some people to exploit a great number of other people; organizations can grow tremendously practically overnight, and because of the fast pace of change, it is practically impossible to control our most important resource, the people who work in our organizations and the leaders who influence them. Awareness of the leadership

### REFERENCES


traits that make organizations a nightmare to work in, may constitute the first step in preventing an important reason for stress and burnout among employees (Schyns and Schilling, 2013). Distinguishing the three most important traits that seem to underlie the dark side of leadership—leader dishonesty, leader disagreeableness, and leader carelessness—, and getting a grip on the steps that organizations can take to deal with these traits, may go a long way in helping create a more optimal work environment.

# AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and approved it for publication.


multiple group situations. J. Appl. Psychol. 76, 308–315. doi: 10.1037/0021-9010. 76.2.308


**Conflict of Interest Statement:** A Dutch version of the HEXACO-PI-R has been released for commercial purposes. A percentage of the profit from sales is used by the University to support the research of the author.

Copyright © 2018 de Vries. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Different Shades—Different Effects? Consequences of Different Types of Destructive Leadership

Ellen A. Schmid\*, Armin Pircher Verdorfer and Claudia V. Peus

TUM School of Management, Technische Universität München, Munich, Germany

Destructive leadership comes in many shapes and forms. From reviewing the literature, we conclude that three major forms of destructive leader behaviors are described: (1) follower-directed destructive behaviors, i.e., genuine abusive forms of destructive leadership, (2) organization-directed behaviors, i.e., behaviors such as stealing from the organization or embezzlement, and (3) self-interested destructive leader behavior, i.e., leader who exploit others to reach their goals. One can easily imagine that these three types of leader behavior have very different effects on followers. Unfortunately, so far, there is no empirical evidence to support this, since comparative research in the field of destructive leadership is scarce. With this paper, we aim to address this gap: In two studies, an experimental and a field study, we examine the differential impact of these three different destructive leader behaviors on two important outcomes: first, their impact on different emotional reactions of followers, the most proximal outcome to a social interaction. Second, we examine a key outcome in leadership research: followers' turnover intention. The results suggest that different types of destructive leader behavior do impact followers differently. Whereas all three behaviors had a positive relationship with negative affect, follower-directed destructive behaviors had the strongest relation out of the three. As expected, all three types of destructive behavior relate to turnover intention, yet, the results of our study suggest that different types of destructive leader behavior relate to different urgencies of turnover intention. We conclude that a tailored approach to destructive leadership, whether in research or practice, seems necessary, as diverse types of destructive leader behaviors affect employees differentially.

Keywords: destructive leadership, differential effects, dark side of leadership, exploitative leadership, turnover intention

# INTRODUCTION

The media frequently reports stories about so-called "bad bosses." On a closer look, these destructive leader behaviors come in many forms. Recently, Jeff Bezos, CEO of Amazon, was announced as winner of the world's "worst boss" award at the 3rd International Trade Union Confederation World Congress in Berlin, because Amazon is said to exploit its workers. Microsoft was also in the press, when a senior manager was arrested on federal charges for stealing more than 9 million USD from the company to pay for a lavish lifestyle. Steve Jobs of Apple, on the other hand, was known for an aggressive leadership style, shouting at and humiliating others (Isaacson, 2011).

Edited by: Ronit Kark, Bar-Ilan University, Israel

### Reviewed by:

Laura Venz, Universität Mannheim, Germany Cristina García-Ael, Universidad Nacional de Educación a Distancia (UNED), Spain

> \*Correspondence: Ellen A. Schmid ellen.schmid@tum.de

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 31 October 2017 Accepted: 05 July 2018 Published: 25 July 2018

### Citation:

Schmid EA, Pircher Verdorfer A and Peus CV (2018) Different Shades—Different Effects? Consequences of Different Types of Destructive Leadership. Front. Psychol. 9:1289. doi: 10.3389/fpsyg.2018.01289

It is intuitively compelling that an abusive leader, who shouts, has a different effect on a follower than a leader who exploits followers, or a leader who violates organizational rules. Unfortunately, we do not have empirical evidence to know if this is simply a lay assumption or if followers do have different reactions to different types of destructive leader behaviors. One reason for this is that comparative research in the field of destructive leadership is scarce. Rather, empirical work in the field is characterized by isolated investigations of separate destructive leadership constructs, resulting in a body of evidence that seems somewhat scattered and disconnected. This is unfortunate for both theory and practice. From a theoretical perspective, we still know too little about the unique and relative contributions of different destructive leader behaviors regarding negative follower outcomes. As a consequence, practitioners have little guidance when it comes to distinguishing, detecting, and managing different forms of destructive leadership in organizational contexts.

This is further aggravated since a broad body of research evidence suggests that negative information has a stronger influence on us and that we perceive and process negative events in a more nuanced way than positive ones (Baumeister et al., 2001; Unkelbach et al., 2008). This "bad is stronger than good" phenomenon has important implications for the domain of leadership. Not only are destructive leader behaviors likely to have a far stronger impact on followers than constructive behaviors, but the adverse impact of such destructive behaviors is likely to outweigh the benefits gained from positive relationships (e.g., with coworkers or customers). Negative interactions with a leader are likely perceived as more nuanced and more dissimilar from each other than in the case of positive information about the leader (Unkelbach et al., 2008). In our view, this makes understanding the differential effects of different destructive leader behaviors even more urgent. Thus, our main purpose in this article is to investigate whether and to what degree different types of destructive leadership may affect followers in a distinct way. In our theoretical model, we draw on the work of Einarsen et al. (2007) and Schyns and Schilling (2013). We argue that the target of the leader behavior and the level of hostility are key factors in understanding the potentially unique effects of different types of destructive leader behavior on followers. Specifically, we focus in our study on three constructs of destructive leadership: (1) abusive supervision (Tepper, 2000), as a behavior high on hostility focusing on the follower; (2) exploitative leadership (Schmid et al., 2017), as a behavior low on hostility focusing on the follower; and (3) organization-directed destructive leadership (Thoroughgood et al., 2012), as a behavior low on hostility focusing on the organization.

In order to answer the question of how far these different destructive leader behaviors elicit different reactions in followers, we draw on emotions as the first reaction to an interaction with a leader (Dasborough, 2006). Furthermore, we investigate the intention to leave, one of the most well-researched outcomes in destructive leadership research (Schyns and Schilling, 2013) and highly relevant to organizations. We thus deem these outcomes as most suited to understanding different follower reactions to destructive leadership.

# THEORETICAL BACKGROUND

Leadership is one of the most important relationships in the workplace and the way leaders give direction, assign tasks, and handle conflict has a strong influence on followers (Yukl, 2012). With this, it becomes particularly important to consider what social research refers to as "negativity bias." In a seminal article famously titled "Bad is Stronger than Good," Baumeister et al. (2001) cite extensive evidence showing that bad events and interactions "have more impact than good ones, and bad information is processed more thoroughly than good" (p. 323). To account for this phenomenon, Baumeister et al. (2001) draw on evolutionary selection: in order to survive threats, it was important for organisms to recognize and remember negative information more strongly than positive. As a consequence, negative information has greater emotional and motivational significance. This has important implications for the study of destructive leadership, since destructive leaders should therefore have a strong influence on followers' emotional state and their motivation to act.

Related to this, more recent research indicates that there is a significant difference between how we generally process positive versus negative information. Unkelbach et al. (2008) describe this in the density hypothesis. They argue that information is generally perceived as more similar to other positive information compared to negative information's similarity to other negative information (i.e., negative information is perceived as more dissimilar to other negative information). Thus, while destructive leadership generally impacts followers more strongly, followers may also be very sensitive to the unique features of different destructive leader behaviors.

Against this backdrop, a great deal of attention has been given to the nature and processes of destructive leadership over the last 15 years (for a review, see Schyns and Schilling, 2013). Different definitions and constructs of destructive leadership exist, all describing different behaviors. The most widely researched construct is abusive supervision (Schyns and Schilling, 2013). This refers to repeated hostile and aggressive yet nonphysical behaviors toward followers (Tepper, 2000). One of the most recent constructs describes a more prevalent form: exploitative leadership (Schmid et al., 2017) refers to genuinely self-interested leader behaviors, such as using followers for personal gain and taking credit for followers' work. Other researchers have pointed to destructive leader behaviors such as accepting bribes, stealing, or making personal use of company property (Einarsen et al., 2007; Thoroughgood et al., 2012).

In short, the literature on destructive leadership describes a multitude of different constructs (for a review, see Schyns and Schilling, 2013). At the same time, efforts have been made to integrate and organize these different approaches (Einarsen et al., 2007; Thoroughgood et al., 2012; Krasikova et al., 2013; Schyns and Schilling, 2013). In the present work, we follow the seminal taxonomy provided by Einarsen et al. (2007), who describe destructive leadership behavior along two dimensions: destructive leader behaviors targeting the followers versus destructive behaviors that target the organization. This distinction is well established and commonly used when it comes to organizing empirical evidence on destructive leadership (Aasland et al., 2010; Thoroughgood et al., 2012; Schyns and Schilling, 2013). In addition, we follow the work of Schyns and Schilling (2013), who concluded that the core of destructive leadership lies in the hostile or hindering nature of the leader's behavior. They defined destructive leadership as "a process in which over a longer period of time the activities, experiences and/or relationships of an individual or the members of a group are repeatedly influenced by their supervisor in a way that is perceived as hostile and/or obstructive" (2013, p. 141).

Taken together, these two aspects (i.e., the target of behavior and the level of hostility) offer a useful basis for differentiating constructs. Cross-tabulation of the two dimensions results in four theoretical destructive leadership behavior categories, as shown in **Figure 1.** The underlying rationale for these categories is presented below.

# Follower-Directed Behaviors High in Hostility

Constructs describing follower-directed destructive leader behaviors usually stem from the bullying literature (Tepper, 2000) and refer to genuinely abusive forms of leadership, high in hostility. The most widely researched construct appears to be abusive supervision (Schyns and Schilling, 2013). Abusive supervision refers to "subordinates' perceptions of the extent to which their supervisors engage in the sustained display of hostile verbal and nonverbal behaviors, excluding physical contact" (Tepper, 2000, p. 178). Other variants of this notion are, for instance, petty tyranny (Ashforth, 1994), social undermining (Duffy et al., 2002), strategic bullying (Ferris et al., 2007), or despotic leadership (De Hoogh and Den Hartog, 2008). While none of these constructs conceptualizes follower-directed destructive leadership in exactly the same way, they all have in common that they describe leaders who behave in a hostile and aggressive (yet nonphysical) manner toward followers.

This includes repeatedly intimidating and belittling followers. However, the most established assessment of these constructs is abusive supervision (Tepper, 2000).

# Follower-Directed Behaviors Low in Hostility

Recently, Schmid et al. (2017) have introduced the concept of exploitative leadership to describe a prevalent leadership behavior that targets the followers but is not inherently hostile or aggressive. Exploitative leadership describes behaviors "with the primary intention to further the leader's self-interest by exploiting others, reflected in five dimensions: genuine egoistic behaviors, taking credit, exerting pressure, undermining development, and manipulating" (Schmid et al., 2017, p. 26). Self-interested behaviors, such as taking credit for followers' work or undermining the development of followers to benefit the leader, are low in regard to hostility. Schmid et al. (2017) posited that exploitative leadership may even be overtly friendly toward followers. Certainly, we can imagine situations where the self-interested behaviors of a leader may even benefit the organization. If a leader's goals and the organization's goals align, the leader may push followers to achieve higher targets. This may be done in a seemingly friendly way, and not by being directly abusive.

# Organization-Directed Behaviors Low in Hostility

Thoroughgood et al. (2012) described organization-directed destructive leadership around behaviors that violate the established rules and social norms of conduct in an organization. There is a broad variety of behaviors that fall under this category—for instance, theft (e.g., stealing small materials such as pens, but also money or time), talking negatively about the organization, using company properties for personal gain, as well as fraud or corruption, and even substance abuse at work (Thoroughgood et al., 2012). While these behaviors certainly vary in terms of their seriousness and harmfulness for the organization, they are not high on hostility as such.

# Organization-Directed Behaviors High in Hostility

Behaviors that fall under the category of organization-directed leadership characterized by high levels of hostility have not been explicitly described in the destructive leadership literature. However, from a theoretical viewpoint and borrowing from research in the field of workplace deviance (Martinko et al., 2002), such behaviors refer to acts of genuine aggressiveness toward the organization. Examples would be sabotage, equipment destruction, or vandalism (e.g., spreading computer viruses). We assume that this type of destructive leadership represents a low base rate phenomenon. While this is in part true for all forms of destructive leadership, such explicitly hostile behaviors against the organization are likely to be performed particularly covertly and thus remain unseen by others. As such, they are less likely to elicit effects on followers. Thus, in the current study, we focus on those behaviors that are more prevalent and feasible to assess and that are established constructs in the destructive leadership literature.

In conclusion, we propose that two important differentiating factors of destructive leadership are: (1) the level of hostility and (2) the target of the behavior. Based on this, in the next section we develop different hypotheses for three recurring destructive leadership behaviors: abusive supervision, exploitative leadership, and organization-directed destructive leadership.

# DIFFERENT EFFECTS OF DIFFERENT DESTRUCTIVE LEADERSHIP BEHAVIORS

In this part of our article, we delineate the proposed different effects of different destructive leadership behaviors on relevant follower outcomes.

When assuming that the target of the leader's behaviors and the level of hostility are the differentiating factors between different types of destructive behavior, these two factors would naturally impact how an employee reacts. As mentioned before, negative information, such as destructive behavior of a leader, has higher emotional and motivational significance than positive information (Baumeister et al., 2001). Thus, we first assume that followers' emotions, as the most proximal reaction (Sy et al., 2005; Bono and Ilies, 2006; Bono et al., 2007) when confronted with destructive leadership, are likely to differ as a function of different destructive leadership behaviors.

Secondly, we follow the argument by Baumeister et al. (2001) that negative information has a strong motivational significance, in that it triggers an action (e.g., avoiding a negative stimulus). Thus, when relating this to destructive leadership, different levels of hostility are likely to have a different impact on the motivation to leave a leader. We thus propose to focus on emotions and the intention to leave the leader (i.e., turnover intention) in analyzing the different effects of abusive supervision, exploitative leadership, and organization-directed destructive leadership.

A very proximal effect a leader's behaviors have is on their followers' emotions (Sy et al., 2005; Bono and Ilies, 2006; see, for example, Bono et al., 2007). As such, all experiences of destructive leadership are likely paralleled by negative emotions. However, the extent of the negative affect is thought to vary, depending on the level of hostility and if the follower is targeted directly. Several scholars (e.g., Schaubhut et al., 2004; Tepper, 2007; Thau and Mitchell, 2010) have argued that destructive leadership is destructive since it is a threat to the self-worth of the followers. Abusive supervision is described as rather high on hostility. By targeting the follower—for instance, by ridiculing followers in front of others or even telling them they are incompetent abusive supervisors would very directly harm the self-worth of followers (Burton and Hoobler, 2006). Accordingly, hostile and aggressive behaviors, such as described in abusive supervision, have been consistently related to negative affect in empirical studies (Aquino et al., 1999; Tepper, 2007). In line with this, we posit that abusive supervision has a strong impact on employees' negative affect.

Exploitative leaders, on the other hand, will take credit for work or manipulate followers to further their own self-interest. Such behaviors, while still targeting the follower directly, are lower on hostility and should thus have a less detrimental effect on followers' self-worth. While being exploited would certainly relate to negative affect, we posit that it does so less strongly than abusive supervision. On the other hand, leaders that show antiorganizational behaviors (Thoroughgood et al., 2012) will show negative behaviors that are not a direct attack on the followers' self-worth. Stealing from the organization, or talking negatively about it, primarily targets the organization, and is rather distal from the follower. We posit that this should have the least strong effect on followers' negative affect.

Thus, we specify the following predictions:

Hypothesis 1: All three destructive leader behaviors (i.e., abusive supervision, exploitative leadership, and organizationdirected destructive leadership behaviors) will have a positive relationship with follower negative affect.

Hypothesis 1a: Abusive supervision will have a stronger positive relationship with followers' negative affect in comparison to exploitive leadership and organization-directed destructive leadership behaviors.

Hypothesis 1b: Exploitative leadership will have a stronger positive relationship with followers' negative affect in comparison to organization-directed destructive leadership behaviors.

Tepper et al. (2009) argued that when followers are confronted with self-worth threatening interactions, they feel a need to empower themselves. A very strong way to empower themselves is turnover, since a follower who intends to leave the job is less dependent on their supervisor (Tepper et al., 2009). We expect exploitative leadership, just like abusive supervision and organization-directed destructive leadership, to relate to general turnover intentions, as previous research has shown (Tepper, 2000; Schyns and Schilling, 2013; Schmid et al., 2017). We therefore predict that all three leadership styles will cause followers to reconsider their employment options. However, the degree of self-worth threat is assumed to vary depending on the level of hostility and how directly a follower is targeted by the behavior. We thus expect that the urgency of the turnover intentions will vary. Since abusive supervision represents a more direct attack on the follower with high levels of hostility, this should relate to followers considering immediate turnover (i.e., leaving the situation immediately). We argue that exploitative leadership, as a less hostile behavior, poses less of a self-worth threat to followers, resulting in a less immediate need to leave the situation. Therefore, followers under exploitative leadership will take a rather more calculative approach and consider staying until, for example, the next career level is reached. Since organization-directed destructive leadership behaviors are more distal and do not target the follower directly, the effect is more difficult to predict. It may be that a leader harming the organization confronts followers with behaviors that run against their feeling of what is right and wrong. On the other hand, the anti-organizational behavior of the leader may be too distant; as Thoroughgood et al. (2012,

p. 18) put it "...such behaviors might not increase turnover intentions as quickly as overtly abusive acts." Therefore, we only specify hypotheses for abusive supervision and exploitative leadership.

Hypothesis 2: All three destructive leadership behaviors (i.e., abusive supervision, exploitative leadership, organization directed destructive leadership) will have a positive relationship with general turnover intentions.

Hypothesis 2a: Exploitative leadership will have a stronger positive relationship with followers' calculative turnover intentions in comparison to abusive supervision and organization-directed destructive leadership behaviors.

Hypothesis 2b: Abusive supervision will have a stronger positive relationship with followers' immediate turnover intentions in comparison to exploitative leadership and organization-directed destructive leadership behaviors.

Our research model is shown in **Figure 2.**

# METHODS

To test the hypotheses under investigation, we conducted two studies with different designs. In Study 1, we used a working sample and adopted a scenario-based approach to manipulate destructive leadership (i.e., abusive supervision, exploitative leadership, organization-directed destructive leadership). Then, respondents were randomly assigned to one of three conditions and provided self-reports on affective reactions and turnover intentions. Study 2 was a field study in which employees from various occupations and organizations rated their immediate supervisor in terms of destructive leadership (i.e., abusive supervision, exploitative leadership, organization-directed destructive leadership). In line with Study 1, self-reports of affective reactions and turnover intentions were collected.

We certify that the research presented in this manuscript has been conducted within the ethical standards of the DGP (German Psychological Society) regarding research with human participants and scientific integrity. We adhere to the ethical standards of the DGP, since in Germany there is no legal regulation for approval of research through a research ethics committee for the social sciences, but ethics questions are addressed within a framework by professional associations.

### Study 1

### Sample and Procedures

Building on prior research on leadership that has successfully used the vignette method (e.g., De Cremer, 2006; Van Dierendonck et al., 2014), we created three hypothetical scenarios for abusive supervision, exploitative leadership,

and organization-directed destructive leadership behaviors by covering the core elements of each construct (see Appendix).

Participants for this study were recruited via an open online survey conducted within the network of three Master's students. On the first page of the online survey, participants were informed that participation was voluntary and by continuing to the second page, they consented to participating in the study. A prerequisite for participating in the survey was that participants were employed full time. In total, 297 participants took part in the online survey and were randomly assigned to one of the three experimental groups (92 in the exploitative leadership, 113 in the abusive supervision, and 92 in the organization-directed destructive leadership condition). In total, 136 respondents were female, the mean age was 25.64 (SD = 7.04), and the majority of the participants (95.6 percent) worked in the for-profit sector.

### Measures

### **Manipulation check**

After presenting respondents with the scenarios, they were asked to rate them in terms of abusive supervision, exploitative leadership, and organization-directed destructive leader behaviors to test whether the manipulation of the independent variable was successful. Exploitative leadership was assessed by six items taken from the exploitative leadership scale (α = 0.87) introduced by Schmid et al. (2017). These six items covered the five dimensions of exploitative leadership (i.e., egoism, taking credit, exerting pressure, undermining development, manipulating). An example item was "This leader prioritizes their own goals over the goals and needs of followers." Abusive supervision was measured by six items taken from the abusive supervision scale by Tepper (2000; α = 0.89). An example item was "This leader puts me down in front of others." Organizationdirected destructive leadership behaviors were captured with seven items from the anti-organizational leader behavior subscale developed by Thoroughgood et al. (2012; α = 0.92); a sample item was "This leader violates company policy/rules." All leadership items were rated on a five-point scale (ranging from 1= strongly disagree to 5 = strongly agree).

### **Emotional reactions**

Emotional reactions Emotional reactions were measured by using the German version (Krohne et al., 1996) of the 20-item Positive and Negative Affect Schedule (PANAS; Watson et al., 1988). The PANAS contains two ten-item sub-scales to measure both negative and positive affect. In the current study, both subscales showed sufficient reliability (α = 0.75 for both sub-scales). Respondents were instructed to indicate the extent to which they felt this way (e.g., active, interested, or excited for positive affect versus distressed, upset, or guilty for negative affect) toward the leader described in the scenario. Responses were given on a fivepoint scale (ranging from 1= very slightly or not at all to 5 = extremely).

### **Turnover intentions**

We assessed three indicators related to turnover intention. Firstly, we adapted two items from Kirchmeyer and Bullin (1997) to assess general turnover intentions ("I would start looking for a new job") as well as immediate turnover intentions ("I would hand in my notice immediately"). Moreover, we developed an item to measure calculative turnover intentions ("I would wait for the next career step is reached before leaving"). Responses were anchored on a five-point continuum (ranging from 1= strongly disagree to 5 = strongly agree).

### Results

### **Manipulation check**

The manipulation check was tested by a one-way analysis of variance (ANOVA), including post-hoc comparisons using the Tukey HSD test. The results revealed a significant effect of leadership style manipulation on the perception of exploitative leadership [F(2, 235) = 7.41, p < 0.001]. Post-hoc comparisons indicated that the exploitative leadership manipulation was indeed perceived as being more exploitative (M = 4.32; SD = 0.76) compared to abusive supervision (M = 3.94; SD = 0.80) and organization-directed destructive leadership behaviors (M = 3.87; SD = 0.75). Similarly, there was a significant effect of leadership manipulation on the perception of abusive supervision [F(2, 237) = 109.33, p < 0.001]. Post-hoc analysis indicated that the abusive supervision vignette was indeed perceived as being more abusive (M= 4.44; SD = 0.64) than the exploitative leadership condition (M = 3.02; SD = 0.83) and the organization-directed destructive leadership condition (M = 3.00; SD = 0.75). Finally, we found a significant effect of leadership manipulation on the perception of organizationdirected destructive leadership behaviors [F (2, 238) = 109.66, p < 0.001]. Post-hoc analysis showed that the organizationdirected destructive leadership behaviors condition was indeed perceived as being more organization-directed destructive (M = 4.10; SD = 0.78) than the abusive supervision condition (M = 2.37; SD = 0.80) and the exploitative leadership condition (M = 2.40; SD = 0.84). Taken together, this pattern shows that the leadership manipulations were successful.

### **Hypothesis tests concerning followers' emotional reactions**

Next, we tested our hypotheses regarding the proposed different effects of the three destructive leader behaviors. The first set of hypotheses refers to affective reactions. Although the focus of our analysis was the effects of destructive leader behavior on negative affect, we deemed it useful to account, too, for the effect on positive affect. The mean scores pertaining to the three conditions are shown in **Table 1**.

A multivariate analysis of variance (MANOVA) with destructive leadership as independent variable and negative and positive affect as dependent variables showed a significant multivariate effect [F(4, 504) = 7.20, p < 0.001; Wilk's 3 =.89, η 2 = 0.05]. Yet, univariate testing found the effect to be significant

TABLE 1 | Mean scores of emotional reactions (Study1).


only for negative affect [F(2, 253) = 13.28, p < 0.001], and no significant effect was found for positive affect [F(2, 253) = 2.92, p = 0.06]. Post-hoc comparisons using the Tukey HSD test indicated that negative affect was significantly higher in the abusive supervision condition (M = 3.60, SD = 0.66), as compared to the exploitative leadership condition (M = 3.11, SD = 0.54) and the organization-directed destructive leadership condition (M = 3.34, SD = 0.62). No difference between the exploitative leadership and the organization-directed destructive leadership conditions was revealed. Thus, hypothesis 1 was supported.

Next, we adopted an explorative perspective and examined whether the different types of destructive leadership under investigation would be related to specific facets of negative affect. Specifically, building on the work of Mehrabian (1997), Janke and Glöckner-Rist (2014) found evidence that the negative affect items of the PANAS reflect two sub-dimensions, upset and afraid. The upset dimension contains the upset, hostile, and irritable items, whereas the afraid dimension includes the guilty, ashamed, afraid, nervous, jittery, distressed, and scared items. Using the Pleasure-Arousal-Dominance (PAD) emotion model (Mehrabian, 1996) as a framework, Mehrabian (1997) found that the upset dimension is characterized by high levels of displeasure (i.e., genuine negative emotional state) and, though less heavily, by arousal (i.e., mental and/or physical activity level). In contrast, the afraid dimension relates less strongly to displeasure, more to arousal, and also more to submissiveness (i.e., lack of control over others or situations).

The mean scores for the two sub-dimensions that we obtained in the current study are shown in **Table 2**. Again, a MANOVA with destructive leadership as the independent variable and the upset and afraid dimensions as dependent variables revealed a significant multivariate effect [F(4, 504) = 15.08, p < 0.001; Wilk's 3 = 0.80, η <sup>2</sup> = 0.11].

Separate one-way ANOVAs for each dimension showed the following pattern. For the upset dimension, we found a significant effect of the leadership manipulation [F(2, 252) = 10.62, p < 0.001]. Post-hoc analyses revealed no significant difference between the exploitative leadership condition (M = 4.16, SD = 0.79) and the abusive supervision condition (M = 4. 25, SD = 0.72). Yet, both conditions were significantly different from the organization-directed destructive leadership condition (M = 3.73, SD = 0.81). Next, also for the afraid dimension, we found a significant main effect [F(2, 253) = 17.20, p < 0.001]. Respondents scored similarly high in the abusive supervision (M = 3. 32, SD = 0.85) and the organization-directed destructive leadership conditions (M = 3.18, SD = 0.69), which were both



significantly different from the exploitative leadership condition (M = 2. 67, SD = 0.63).

### **Hypothesis tests concerning followers' turnover intentions**

The MANOVA we conducted showed a statistically significant difference in turnover intentions based on the leadership manipulation [F(6, 482) = 2.69, p < 0.05, Wilk's 3 = 0.93, η <sup>2</sup> = 0.03]. Separate ANOVAs showed the following pattern. For general turnover, we found a significant effect of the leadership manipulation [F(2, 243) = 3.70, p < 0.05]. Post-hoc analysis using the Tukey HSD procedure revealed a significant difference between the abusive supervision (M = 4.52, SD = 0.68) and the organization-directed destructive leadership conditions (M = 4.24, SD = 0.84). For the other combinations, no significant differences were revealed. Overall, general turnover intentions were substantially high in all three conditions, thus confirming hypothesis 2. Next, for calculative turnover intentions, we found no significant effect of the leadership manipulation [F(2, 243) = 1.24, p = 0.32]. Therefore, hypothesis 2a was not confirmed. Finally, immediate turnover significantly differed between the conditions [F(2, 243) = 3.58, p < 0.05]. Post-hoc analyses showed that immediate turnover was lower in the exploitative leadership (M = 2.01, SD = 0.90) than in the abusive supervision (M = 2.39, SD = 0.92) condition, thus confirming hypothesis 2b. The organization-directed destructive leadership condition (M = 2.30, SD = 0.96) did not significantly differ from the other two groups.

### Brief Discussion

This study revealed a series of distinct effects, in particular for exploitative leadership and abusive supervision. As predicted, abusive supervision emerges as the strongest precursor to overall negative affect.

Both abusive supervision and exploitative leadership are associated with stronger feelings of displeasure (i.e., upset) compared to organization-directed destructive behaviors. Yet, with regard to the afraid dimension of the PANAS, an interesting difference was revealed, with lower scores for exploitative leadership relative to the abusive supervision condition. This suggests that abusive supervision is more strongly related to anxiety among followers, reflected in increased arousal and feelings of submissiveness (Mehrabian, 1997). With regard to turnover, all three forms of negative leadership were related to high general turnover intention. While the level of calculative turnover intention was inconspicuous among the three conditions, abusive supervision tends to relate to higher immediate turnover reactions.

Overall, the results were only partly as expected. This may be because of the hypothetical nature of the scenarios. Therefore, in Study 2 we designed a field study to test the same hypotheses.

### Study 2

### Sample and Procedures

We gathered valid responses from 167 employees from various organizations in Germany who rated their immediate leaders in terms of destructive leadership and provided self-reports on emotional reactions and turnover intentions. Respondents were contacted via snowball sampling, starting with the authors' professional network. The majority of the participants (72 percent) worked in the for-profit sector (28 percent worked in non-profit organizations or in the public sector). The mean age was 36.22 years (SD = 12.13) and 63.30 percent of the respondents were male. On average, the respondents had been working for their current supervisor for 4.89 years (SD = 5.48) and organizational tenure was 7.98 years on average (SD = 8.69). In terms of education, 67 percent of the respondents held a university degree.

### Measures

### **Destructive leadership measures**

Exploitative leadership was assessed with the full 15-item exploitative leadership scale developed by Schmid et al. (2017). Abusive supervision was measured according to the full 15 item abusive supervision scale by Tepper (2000). Organizationdirected destructive leader behaviors were captured with the measure developed by Thoroughgood et al. (2012). Sample items can be seen in the description of measures in Study 1. Respondents rated the frequency of destructive leader behaviors on a five-point scale ranging from 1 ("not at all") to 5 ("frequently if not always").

### **Outcome measures**

For the outcomes (i.e., emotions and turnover), we used the same items with the same response format as in Study 1.

### Results

### **Validity of measures**

**Table 3** reports the descriptive statistics and correlations among the study variables.

Prior to testing the hypotheses under investigation, we examined whether the measures we used represented valid tools to assess our target constructs. To this end, we used confirmatory factor analysis (CFA) in AMOS and tested the factorial integrity of our measures. In a first step, we conducted CFA on the item level for each measure separately (i.e., exploitative leadership, abusive supervision, organization-directed destructive leadership) and examined the factor loadings and item reliabilities. While all items of the exploitative leadership measure had excellent psychometric properties, we dropped several items of the other two measures (i.e., abusive supervision, organization-directed destructive leadership) because they did not represent the underlying



construct well (i.e., factor loadings were below 0.60 and item reliabilities below 0.40; Hair et al., 2006).

Next, we tested the discriminant validity of our measures. Because of the relatively large number of estimated parameters in the overall model and the small sample size, we created item parcels for all latent leadership constructs (Landis et al., 2000). For exploitative leadership, we formed five parcels based on the five dimensions specified by Schmid et al. (2017) (i.e., egoism, taking credit, exerting pressure, undermining development, and manipulation). For abusive supervision and organizationdirected destructive leadership, we used the factorial algorithm to create parcels (see Matsunaga, 2008). By sequentially including the items with the highest to the lowest factor loadings, while alternating the direction of item selection, three parcels were formed for abusive supervision and two parcels for organizationdirected destructive leadership.

On this basis, we tested a series of theoretically viable factor models. **Table 4** shows that a three-factor model with the three target constructs as latent variables and parcels as indicators obtained the best model fit and was preferable over alternative solutions. These results provide evidence that our measures captured distinct constructs versus common source effects.

### **Hypothesis tests concerning followers' emotional reactions**

To test our hypotheses, we conducted a series of multiple regression analyses. In addition, given the high correlations among the destructive leadership measures, we followed the procedures suggested by Lorenzo-Seva et al. (2010) and applied relative weight analysis. The results of these procedures are depicted in **Tables 5**, **6**.

Abusive supervision was the strongest predictor of overall negative affect (β = 0.50, p < 0.001), followed by exploitative leadership (β = 0.29, p < 0.001), and organization-directed destructive leadership (β = 0.08, ns). Thus, hypothesis 1 was supported. For overall positive affect, abusive supervision (β = −0.27, p < 0.01) and exploitative leadership (β = −0.29, p < 0.05) exerted a similar negative effect, while the effect for organization-directed destructive leadership was not significant (β = 0.09, ns). With regard to the sub-dimensions of negative affect (see **Table 7**), the following pattern was revealed: the upset dimension was best predicted by abusive supervision (β = 0.47, p < 0.001), followed by exploitative leadership (β = 0.33, p < 0.001). The effect for organization-directed destructive leadership was not significant (β = 0.06, ns). In a similar vein, abusive supervision was the strongest predictor for the afraid dimension (β = 0.46, p < 0.001) followed by exploitative leadership (β = 0.23, p < 0.05). Again, organization-directed destructive leadership had no predictive value here (β = 0.09, ns).

The next set of hypotheses refers to different types of turnover intention. For general turnover intention, the results of regression analysis revealed only exploitative leadership as a significant predictor (β = 0.47, p < 0.001). Relative weight analysis, however, showed that the other two leadership forms also explained variance in general turnover intention (see **Table 8**); however, exploitative leadership clearly exerted the strongest effect. While these results do not fully confirm

### TABLE 4 | Measurement models (Study 2).


1χ 2 represents the difference in χ 2 values between the respective model and Model 1 (i.e., the proposed 3-factor model); \*\*\*p < 0.001.

TABLE 5 | Descriptive statistics and correlations (Study 2).


\*\*p < 0.01; \*p< 0.05; Cronbach's alpha appears on the diagonal.

hypothesis 2, relative weights analysis does point to an effect in the expected direction.

With regard to calculative turnover, we found a positive effect for exploitative leadership (β = 0.48, p < 0.001), whereas the effect of abusive supervision was negative (β = −0.39, p < 0.001). Given that the two predictor variables were highly correlated (r = 0.75, p < 0.001), while abusive supervision did not correlate with the outcome variable (r = 0.01, ns), this pattern shows the classic signs of a suppression effect (Tzelgov and Henik, 1991). This means that abusive supervision shares no or only little variance directly with the outcome variable but contributes to the regression equation by removing irrelevant variance from the other predictor variables. This is also reflected in the results of relative weight analysis, showing that exploitative leadership explained the major portion of variance in calculative turnover intentions (see **Table 6**). While hypothesis 2a is again not fully confirmed, taken together, this pattern points to what was predicted.

Interestingly, for immediate turnover, only exploitative leadership was a significant predictor in the regression analysis (β = 0.46, p < 0.001). Again, relative weight analysis revealed that the other two leadership forms also explained variance in immediate turnover intention, yet only to a moderate extent (see **Table 6**). Thus, hypothesis 2b was not supported.

### Brief Discussion

In line with the results found in Study 1, both abusive supervision and exploitative leadership were found to have a negative relationship with positive affect. With regard to negative affect, however, different patterns were found. Abusive supervision was related most strongly to overall negative affect and to the afraid sub-dimension of negative affect. Also, it was more strongly related to the upset sub-dimension, relative to exploitative leader behavior. This is different from what we found in Study 1, where exploitative leadership had an equally strong effect on the upset sub-dimension. Overall, the pattern found in Study 2 supports the notion that abusive supervision is both generally and relative to exploitative leadership more strongly related to negative emotional reactions of followers. Organization-directed destructive leader behavior seems to play a marginal role when it comes to followers' emotional reactions. A potential explanation for this could be that followers perceive such leader behaviors as rather distal—i.e., as actions they can more efficiently distance themselves from.

For turnover intentions, the results are more complex. While all three types of destructive leadership behavior relate to general turnover intention, when examining the relative weights, exploitative leadership has the strongest relationship. However, exploitative leadership had the strongest positive relationship

### TABLE 6 | Effects of destructive leadership on overall negative and positive affect (Study 2).


\*\*\*p < 0.001; \*\*p < 0.0; relative weights reflect the relative contribution to R<sup>2</sup> (percentages).

TABLE 7 | Effects of destructive leadership on negative affect sub-dimensions (Study 2).


\*\*\*p < 0.001; \*\*p < 0.0; relative weights reflect the relative contribution to R<sup>2</sup> (percentages).

with calculative turnover intention—i.e., followers would stay until the next milestone in their career was reached before leaving—whereas abusive supervision had limited impact. This is in line with our hypothesis: because of the stronger selfworth threat, followers would be less likely to have a calculative approach.

However, when looking at immediate turnover intention, a low effect was found for abusive supervision. Whereas this may seem counterintuitive at first, the underlying explanation may be that the decision to leave a job depends on many factors that are situational, and may depend on the individual follower's personality.

### GENERAL DISCUSSION

The main purpose of this article was to investigate if different destructive leadership behaviors may affect followers in a distinct way. Our focus was on three destructive leadership constructs: abusive supervision (Tepper, 2000), exploitative leadership (Schmid et al., 2017), and organizationdirected destructive leadership (Thoroughgood et al., 2012). To answer the question of how far these behaviors would elicit different reactions in followers, we investigated followers' emotions as the first reaction to an interaction with leaders (Dasborough, 2006) and the intention to leave. The results of both a scenario-based experimental study and a field study suggest that exploitative leadership does indeed influence different outcomes compared to leaders behaving in an abusive manner or leaders behaving in a manner that harms the organization. As expected, all three constructs had a positive relationship with negative affect. Yet, with regard to the afraid dimension of the PANAS, higher scores for abusive supervision were found. Organization-directed destructive leader behavior showed marginal relevance with regard to different urgencies

### TABLE 8 | Effects of destructive leadership on turnover intentions (Study 2).


\*\*\*p < 0.001; relative weights reflect the relative contribution to R<sup>2</sup> (percentages).

of turnover intention. However, exploitative leadership and abusive supervision affected calculative and immediate turnover intentions to a similar degree. In what follows, we discuss the theoretical and practical implications of these results in more detail.

### Theoretical Implications

In line with prior research (Schyns and Schilling, 2013), our results confirm that destructive leadership is a critical source of negative affect among followers. Since negative affect has generally been shown to undermine employees' social wellbeing and productivity (Barsade and Gibson, 2007; Elfenbein, 2007; for overviews,see Ashkanasy and Dorris, 2017), it is theoretically and practically useful to understand the different influence leaders may have in this regard. Our results extend existing knowledge by showing that different forms of destructive leader behavior have different effects on both the type and intensity of negative affect. Specifically, our results indicate that, in contrast to exploitative leadership, abusive supervision is more strongly related to anxiety among followers. This anxiety is reflected in increased arousal and feelings of loss of control (Mehrabian, 1997). These higher levels of anxiety may be explained through the more hostile and direct attack on the follower posed by abusive supervision. This hostility—i.e., shouting at followers or ridiculing them in public—is a high threat to the self-worth of the follower (e.g., Schaubhut et al., 2004; Tepper, 2007; Thau and Mitchell, 2010) and may thus results in feelings of submissiveness and anxiousness.

This difference in follower emotional reactions is important, since previous research indicates that negative affect is related to stronger effects in organizations than positive affect, and also to more nuanced effects on followers' behavior (e.g., Baumeister et al., 2001; Dasborough, 2006). As an example, previous research suggests that different emotions relate to how employees' attribute blame (Gooty et al., 2009). Related to our results, this means that anxiety in followers may likely be related to blame being attributed internally, and so followers blaming themselves for leaders' behavior. Thus, abusive supervision may be more likely to result in internal attributions of blame, whereas followers with an exploitative leader may rather attribute blame externally—i.e., blame the leader for taking credit for their work. This attribution will likely set in motion very distinct behavioral dynamics, since whether employees attribute destructive leadership internally or externally has been linked to the occurrence of distinct forms of workplace deviance. Specifically, according to the causal reasoning model of counterproductive work behavior (Martinko et al., 2002), external attributions are more likely to trigger retaliatory behaviors (such as hiding knowledge or sabotage), whereas internal attributions are thought to trigger more self-destructive deviance (such as drug and alcohol abuse; Bamberger and Bacharach, 2006). Thus, the difference in attribution relates to very different follower behaviors (Gooty et al., 2009); previous research has shown, furthermore, that it also relates to decision-making and risktaking (Forgas and George, 2001). Thus, an abusive supervisor, by relating to higher anxiety in followers, may inhibit risktaking behavior which would in the long run impede the innovation and flexibility of teams and ultimately the entire organization.

Schmid et al. Consequences of Different Types of Destructive Leadership

Although our results relate to emotions at the individual follower level, they nevertheless imply that different destructive leader behaviors trigger distinct emotional reactions in followers, and these different emotional reactions may trigger very distinct dynamics in teams and organizations. In fact, emotions in organizations are described as a multilevel phenomenon and Ashkanasy and Dorris (2017) posited that emotions act at different levels ranging from the individual level to the team level and the organizational level. Leadership plays an important role in this multilevel phenomenon, since it enables emotions to spread from the individual to the organization through the process of emotional contagion (Hatfield et al., 1992).

A similar pattern, suggesting distinct dynamics resulting from different destructive leader behaviors, was found for turnover intention. Although our results regarding immediate and calculative turnover intention are not as clear as expected, they do suggest that different leader behaviors, depending on how hostile they are and how directly they attack the follower, may trigger a more or less immediate need to act. Thus, the time frame for different destructive leader behaviors to unfold their destructive effect may vary. In a related line of research, it was shown that narcissists make good first impressions and only over time, when their dark side shows, do perceptions others have of them change for the worse (Paulhus, 1998). Narcissists are seen as charming and confident on first encounter, while their exploitative and manipulative side only shows over time, leading to a delay in negative effect on others. A similar effect can be assumed for exploitative leadership. Schmid et al. (2017) described that exploitative leadership can be seemingly friendly. The hostile behaviors of an abusive leader may be more immediately threatening and harder to tolerate on a daily basis.

However, our results also show a counterintuitive pattern: that is, in Study 2, followers' intentions to immediately leave an abusive leader were low. The underlying explanation may be that the decision to leave a job depends on many factors, and we need to consider situational as well as individual factors. An important factor is the availability of other employment options. Thus, the socioeconomic environment needs to be taken into account. Besides the job market, another important factor is how a follower judges their employability. Victims of abuse are often low in self-esteem (see, for example, Aguilar and Nightingale, 1994) and may not rate their employability very highly; they may thus remain in an (abusive) workplace, although it seems counterintuitive.

As mentioned above, attribution may play an important role in unfolding the destructive leadership dynamic (Gooty et al., 2009). While emotions relate to different attribution patterns, followers' individual attribution style should also play an important role. Different attribution of why the leader is showing certain destructive behaviors will relate to different conclusions and, in consequence, different follower behaviors (see, for example, Peus et al., 2012). An abusive supervisor, showing hostile behaviors, may rather lead to an attribution of hostile intentions, whereas an exploitative leader, taking credit for others' work and manipulating others to advance their career, may be seen as rather overly ambitious. While this will naturally lead to different individual follower behavioral reactions, we can also imagine that it will impact the team dynamics differentially. Whereas a leader that is seen as hostile may prompt a team to rally together and create cohesion, a leader that is exploitative may rather create a focus on individual self-interest in the team (Peus et al., 2012).

Taken together, our results show a very complex pattern of different destructive leader behaviors and point to the importance of understanding nuances in destructive leadership. Since previous research suggests that it may be easier to discourage desired follower behaviors, such as creativity, than to encourage them (e.g., Kark et al., 2018), understanding how the destructive leadership dynamics unfold seems crucial for organizations. With this study, we contribute to the advancement of destructive leadership theory and methodology by providing empirical evidence that followers indeed have different reactions to different destructive leadership behaviors and that these reactions are able to provide unique information in terms of predicting followers' emotions and turnover intentions. This has important implications for the landscape of destructive leadership, since the literature so far has overlooked important insights from a methodological, theoretical, and practical perspective. From the perspective of theory advancement, we may overlook mediators and outcomes that are specific to a certain type of destructive leadership behavior (Herschcovis and Barling, 2010). From a methodological perspective, the fact that the majority of studies examine one type of destructive leadership in relation to an outcome (Schyns and Schilling, 2013) and do not compare the effects of different destructive leadership behaviors may result in under- or over-estimations of the true effects (Herschcovis and Barling, 2010). Related to this, with this being only the second empirical study on exploitative leadership that we are aware of, we also make a further contribution to the construct validity of the new construct of exploitative leadership (Schmid et al., 2017).

# Practical Implications

Knowing that different kinds of destructive leadership impact followers differently has important implications for practice. Practitioners, for the purpose of leadership development and coaching, will be able to understand destructive leadership in a more nuanced manner. This allows for more tailored interventions that take into account the impact that is likely to be expected from a certain type of destructive behavior. Related to this, in our view, the results of our studies generally point to the importance of customizing organizational interventions. This means that first the destructive leader behavior needs to be assessed to understand it in terms of the target of the behavior and the level of hostility. Next, interventions can be chosen for example, personal coaching for the leader can work on the specific harming behaviors. In targeting specific behaviors in a customized way in coaching and training, digital learning methods can be highly beneficial in offering individualized solutions. For instance, apps are used to help leaders apply new behaviors in their daily work and receive instant feedback. With knowledge about the specific behaviors, mechanisms, and effects of different types of destructive leader behavior, this may be a promising avenue for future leader training on the job.

### Limitations and Future Research

Despite the contributions, our studies are not without their limitations. In Study 1, we chose an experimental vignette approach, which naturally has a range of limitations. Scenarios, rather than real experiences, may reflect the perception of participants. Whereas internal validity is high in our scenario study, the generalizability is limited. Nevertheless, it has been shown that scenario experiments tend to score well on common realism (Van Dierendonck et al., 2014). In Study 2, we further conducted a field study to gain an understanding of real organizational effects. However, this was a measurement at one point in time and relied solely on followers' perceptions, thus being prone to common method bias. While self-reports are certainly well-suited to capture followers' emotional reactions and individual attitudes (Conway and Lance, 2010), future research in this field may benefit from using more objective measures, such as physiological reactions (Mauss and Robinson, 2009). A further limitation refers to our use of single-item measures for the different facets of turnover intention, most notably with regard to measurement reliability. Yet, prior research has demonstrated that single-item measures are a reasonable option under certain circumstances (e.g., Fisher et al., 2016). On the one hand, we chose single-item measures to minimize respondent burden while increasing face validity. Moreover, we consider the facets of turnover intention rather concrete and specific, so that a general single item enhances respondents' clarity regarding what is actually being measured (Fuchs and Diamantopoulos, 2009; Fisher et al., 2016). It would certainly be fruitful in future research to test multiple- item measures to capture different facets of turnover intention.

Overall, our research opens up multiple avenues for future research. While we have focused on two types of outcome, there are certainly many more outcomes and important mechanisms that would benefit from a more differentiated view. In our view, the most promising next route would be to investigate mechanisms that can shed further light on how different types of destructive leader behavior influence followers. Organizational justice theory has been studied as an important mechanism for destructive leadership (Tepper et al., 2006) and we can imagine that the different types of organizational justice may work as mechanisms with different types of destructive behavior. Whereas abusive supervision may more strongly relate to perceptions of interpersonal unfairness, exploitative leadership will rather violate concerns of distributive and procedural justice. In a similar vein, our results suggest that negative affect may not even be the primary mechanism through which exploitative leadership affects followers. Rather, in contrast to abusive supervision, with its strong focus on hostility and aggression, exploitative leadership may work more strongly through follower cognition than affect. Future research should test this assumption by considering follower outcomes that are inherently cognitive, such as reciprocity expectations (Bernerth et al., 2007).

Furthermore, qualitative studies would be of great interest to shed light on the differences in perceptions and effects of leaders behaving destructively in either an exploitative, an abusive, or an organization-directed way. Specifically, qualitative interviews are especially suited to examine mechanisms and reasons why followers react in certain ways to destructive leadership.

Of interest, furthermore, would be to investigate different effects of different destructive leadership behaviors in a longterm field study, to capture real and longer-term followerleader interactions. We would argue that destructive behaviors high on hostility, such as abusive supervision, would lead to negative effects on outcomes much faster than destructive leader behaviors lower on hostility, like exploitative leadership. With exploitative leadership, negative effects, such as negative emotional reactions and turnover intentions, may only unfold over time.

Since our study focused on the individual follower perspective, future research needs to provide an understanding of how different destructive leader behaviors impact teams. We see different avenues for this. Peus et al. (2012) posited that the negative perceptions an individual develops of a leader can spread to the team through social and emotional contagion processes and create a shared negative perception of the leader. Thus, an employee who witnesses or becomes aware of the leader treating a colleague in an exploitative or abusive manner can be influenced by this (see also Priesemuth et al., 2014). Schmid et al. (2017) have shown first evidence for team level perceptions of exploitative leadership, but how these perceptions spread differently for different destructive leader behaviors remains to be understood. Moreover, followers may mimic their leader's behaviors (e.g., Yaffe and Kark, 2011). Further research should investigate how the different destructive behaviors may be mimicked and how follower mimicking abusive versus exploitative behaviors may impact teamwork. Related to this, future research should further investigate the role of followership in the destructive leader dynamic (Howell and Shamir, 2005). Followers may show these different destructive behaviors toward their leader; thus different destructive upward leadership behaviors and their outcomes need to be understood.

Moreover, when it comes to better understanding antecedents, leader identity and self-concept have received much attention in the leadership literature recently (e.g., Kark and Shamir, 2013; Mainemelis et al., 2015). It describes three levels of the self—the intrapersonal, the interpersonal, and the collective—and may be an important theory to understand why different destructive leaders behave in the way they do. We can imagine an exploitative leader focusing mainly on the intrapersonal aspect of self, whereas an abusive leader may rather focus on interpersonal aspects of their self. An organizationdirected destructive leader would rather focus on the collective aspect.

In conclusion, a more tailored approach to destructive leadership, whether in research or practice, seems necessary, since all destructive leaders are destructive in their own way.

# AUTHOR CONTRIBUTIONS

All authors contributed extensively to the work presented in this paper. All authors designed the study jointly. ES and AP collected the data, AP conducted the majority of the data analysis. ES prepared the first draft of the manuscript, ES, AP, and CP wrote sections of the manuscript. ES, AP, and CP contributed to manuscript revisions, read and approved the submitted version.

### Schmid et al. Consequences of Different Types of Destructive Leadership

### REFERENCES


issues and application guidelines. Die Betriebswirtschaft 69, 195–210. doi: 10.2139/ssrn.2318105


Isaacson, W. (2011). Steve Jobs. New York, NY: Simon & Schuster.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Schmid, Pircher Verdorfer and Peus. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# APPENDIX

Scenarios used in Study 1

### **Scenario: Follower-Directed Destructive Leadership**∗ Please imagine. . .

You are working on a high priority project under a lot of time pressure. You have put a lot of effort into it and have achieved some very good results. Your boss, however, does not give you any credit for the hard work that was required or acknowledge the successful milestone you just completed, but rather keeps reminding you of the things that went wrong in the project. He makes fun of you in front of the whole project team and calls you a failure in public. When you make suggestions on how to go about the next project steps, he tells you that your ideas are stupid and even asks you not to interact with the other team members, since you are incompetent. He has broken numerous promises he made to you and it is not uncommon for him to lose his temper and shout.

### **Scenario: Self-Interested Destructive Leadership**∗

Please imagine. . .

Your boss assigned you to a high priority project under a lot of time pressure. You have put a lot of effort into it, and were asked by your boss to work weekends and sacrifice training and professional development activities to reach the deadline. Now the project is completed and you are very proud of the outcome.Your boss, charming as ever, seizes the task of presenting the results to the customers, who are very impressed and invite him to present it at a prestigious convention. Your boss happily tells you that he has been invited on that all expenses paid trip. He gets all the fame for the successful project and, upon his return, even the desired promotion for advancing the company's reputation. You, however, do not get any credit for your work in the project.

### **Scenario: Organization-Directed Destructive Leadership**∗ Please imagine. . .

You are working on a high priority project under a lot of time pressure. You have put a lot of effort into it, and have achieved some very good results.

During the course of the project, you realize that your boss is harming the organization. He regularly violates company policy. For example, he asks a colleague of yours, who is also working on the project, to take over private tasks for him, thus delaying her work on the project. Whilst reviewing project documentation, you realize that your boss has forged project results. Moreover, during informal talks with external business partners, he sometimes talks negatively about your organization. You heard him say what a lousy company this is. At a company dinner, you even saw him accept a corrupting gift. Recently, you also suspect him of sometimes coming to work under the influence of alcohol.

∗Please note: The original scenario was in German. This is a translation.

# Does Despotic Leadership Harm Employee Family Life: Exploring the Effects of Emotional Exhaustion and Anxiety

### Shazia Nauman<sup>1</sup> \*, Tasneem Fatima<sup>2</sup> and Inam Ul Haq<sup>3</sup>

<sup>1</sup> Riphah School of Business & Management, Riphah International University, Lahore, Pakistan, <sup>2</sup> Faculty of Management Sciences, International Islamic University, Islamabad, Pakistan, <sup>3</sup> Lahore Business School, University of Lahore, Lahore, Pakistan

Research has not focused on the negative effects of despotic leadership on subordinates' life satisfaction and the interface between work and family. Drawing on the Conservation of Resources theory, this research investigates the mediating effect of emotional exhaustion through which despotic leadership transcends from the workplace to subordinates' personal lives, resulting in work-family conflict and decreased life satisfaction. The research also examines the moderating effect of subordinates' anxiety on the relationship of their perceptions of despotic leadership with work-family conflict and life satisfaction. Three waves of time-lagged data was collected from 224 book sellers who work in publishing houses. We used Hayes' PROCESS to test moderation and SEM to test mediation. The results of the study suggest that despotic leadership is related to work-family conflict via emotional exhaustion, but offer no support for its relationship with life satisfaction. As expected, when subordinates' anxiety increases, the positive relationship between a supervisor's despotism and his or her subordinates' work-family conflict and the negative relationship between despotic leadership and life satisfaction both strengthen. The results suggest that despotic leaders harm their subordinates' non-work lives, and these effects intensify when subordinates have high levels of anxiety. These findings have important implications for service organizations in mitigating the negative effects of despotic leadership by minimizing subordinates' anxiety through coping mechanisms and giving reward and incentives.

### Edited by:

Susanne Braun, Durham University, United Kingdom

### Reviewed by:

Rita Berger, Universitat de Barcelona, Spain Arianna Costantini, University of Verona, Italy

> \*Correspondence: Shazia Nauman shaznaum@yahoo.com

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 06 November 2017 Accepted: 10 April 2018 Published: 03 May 2018

### Citation:

Nauman S, Fatima T and Haq IU (2018) Does Despotic Leadership Harm Employee Family Life: Exploring the Effects of Emotional Exhaustion and Anxiety. Front. Psychol. 9:601. doi: 10.3389/fpsyg.2018.00601 Keywords: despotic leadership, emotional exhaustion, anxiety, work-family conflict, life satisfaction

# INTRODUCTION

Research has highlighted the negative or dark side of leadership (Griffin and Lopez, 2005; Wu and Hu, 2009; Naseer et al., 2016) by revealing destructive aspects of leadership that can have negative effects (Schyns and Hansbrough, 2010) on such factors as absenteeism, turnover, effectiveness (Tepper et al., 2006), emotional exhaustion (Harvey et al., 2007), deviant work behavior (Duffy et al., 2002), job satisfaction (Tepper, 2000; Tepper et al., 2004), stress (Tepper, 2000; Chen et al., 2009), and performance (Aryee et al., 2007). These destructive leadership behaviors have been conceptualized and examined under such labels as petty tyranny (Ashforth, 1994), abusive supervision (Tepper, 2000), destructive leadership (Schyns and Hansbrough, 2010),

and despotic leadership (Aronson, 2001). According to Schyns and Schilling (2013), despotic leadership comprises prominent characteristics of negative leadership types, but there is a lack of research in this area in the management and psychology literatures (Naseer et al., 2016).

De Hoogh and Den Hartog (2008) defined despotic leadership as a leader's tendency to engage in authoritarian and dominant behavior in pursuit of self-interest, self-aggrandizement, and exploitation of their subordinates. According to Schilling (2009), despotic leaders want unquestioned submission from their subordinates and use demanding and controlling mechanisms to manipulate and exploit their subordinates for personal gain, regardless of their subordinates' needs and concerns. Thus, despotic leaders work against their organizations' legitimate interests by indulging in self-serving and morally corrupt behavior (Aronson, 2001). Despotic leaders' unethical and unfair behavior in the workplace negatively impacts subordinates' job performance, organizational citizenship behaviors, and creativity (Naseer et al., 2016). Despite increasing evidence that despotic leadership is harmful to employees, there is a lack of research on its negative effects on employees' life satisfaction and the interface between their work and family lives.

The effects of destructive leadership behavior may not be limited to subordinates, as they may also enfold the organization, customers, employees' families, and even society in general. Research has indicated that such behavior is related to a number of negative outcomes, including lowered job satisfaction, organizational commitment, organizational performance, and increased emotional exhaustion, turnover intentions, work-family conflict, and psychological distress among employees (Richman et al., 1992; Ashforth, 1997; Tepper, 2000; Aasland et al., 2010; Hershcovis and Rafferty, 2012; Schyns and Schilling, 2013). Therefore, the negative aspects of leadership are a matter of grave concern for organizations (Hoobler and Hu, 2013) and further investigation is needed (Schyns and Schilling, 2013; Collins and Jackson, 2015) into what causes destructive leadership and how its negative aspects affect subordinates' behavior and relationships at home. The current study addresses this research gap by investigating the harmful effects of despotic leadership on subordinates' life satisfaction and work-family conflicts.

There are several reasons for focusing on these particular outcome variables. First, as despotic leadership is a social stressor and have harmful effects on the home life of a subordinate, we therefore choose work- family conflict instead of work life balance which is a more positive way of viewing workfamily relationships. For example, according to Clark, work– family balance is "satisfaction and good functioning at work and at home with a minimum of role conflict" (Clark, 2000). In a similar vein, Frone (2003) refers work life balance as an absence of role conflict and presence of facilitation: "low levels of inter-role conflict and high levels of inter-role facilitation represent work–family balance." Greenhaus et al. (2003) found that when individuals invest little time or involvement in their combined work and family roles, balance has little or no implications for an individual's quality of life. Second, despotic leadership is a social stressor which is an antecedent of work family conflict and life satisfaction. Third, despotic leadership creates stressors like emotional exhaustion, which has been linked to employee well-being and quality of life at home (Ernst Kossek and Ozeki, 1998). Fourth, studies that have observed spillover effects from work to home show that job experiences influence the home life of employees even after they leave the workplace (Ilies et al., 2009; Eby et al., 2010; Wagner et al., 2014), which shows that the effects of emotional exhaustion may affect other domains of employees' lives as well. For instance, emotional exhaustion harms the family domain, increases work- family conflict, and decreases life satisfaction (Gali Cinamon and Rich, 2010; Lambert et al., 2010; Carlson et al., 2012; Boekhorst et al., 2017). Here, we argue that these dependent variables are the most suitable for this study, as they directly influence the subordinates' wellbeing and quality of life.

The Conservation of Resources (COR) theory comprises several stress theories (Hobfoll, 1989) and provides insight into the interface between work and family (Witt and Carlson, 2006; Grandey and Cropanzano, 1999). The COR theory suggests that people experience stress from an actual or threatened loss of resources (Hobfoll, 1989). It also envisages that resources are lost as individuals try to manage both work and family roles (Grandey and Cropanzano, 1999). This potential or real loss leads to conflicts in the interface between work and family (Witt and Carlson, 2006). According to Grandey and Cropanzano (1999), negative work stressors hamper subordinates' ability to perform their family roles, which may result in inter-role conflict in the form of work-family conflict and life dissatisfaction.

Using the COR theory as a foundation, we theorized that despotic leadership is the source of social stress and the loss of leadership support reflected in self-serving behavior in the supervisor-subordinate relationship. As despotic leadership is authoritarian, vengeful, unethical, self-serving, and exploitative (De Hoogh and Den Hartog, 2008; Naseer et al., 2016), despotic leadership in a highly collectivist, uncertainty avoidant, and power distant culture like Pakistan would lead to emotional exhaustion (Hofstede, 1983, 2010). The loss of resources that results from emotional exhaustion also leads subordinates to experience decreased levels of life satisfaction and has a negative effect on the work-family interface. In this context, we posit that despotic leadership is directly related to the outcome variables of life satisfaction and work-family conflict and is indirectly related via emotional exhaustion.

According to Spielberger and Sydeman (1994), anxiety is defined as "the tendency to perceive a wide range of situations as dangerous or threatening" and is also specified as a predictor of victimization (Olweus, 1978; Aquino and Thau, 2009). As subordinates differ in their tendency to perceive authoritarian behavior in their leaders, their reactions vary such that the subordinates who have a high degree of anxiety are likely to be more sensitive than those who do not. Drawing from COR theory, subordinates lose the support of a despotic leader, and the leader's self-serving behavior is likely to decrease subordinates' life satisfaction and affect their work-family interface negatively.

This loss intensifies among subordinates who have high levels of anxiety. Tepper (2007) also proposes that future research identifies personality moderators of destructive leadership. Since the perceivers' personality affects their reactions to despotic leadership, and trait anxiety moderates the relationship between such leadership, life satisfaction, and work-family conflict, we contend that the interactive relationship between perceived despotic leadership and subordinates' anxiety has a detrimental effect on the subordinates' life satisfaction and work-family conflict.

Therefore, the purpose of this research is to examine the extent to which despotic leadership transcends the work boundary to affect employees' life satisfaction and the interface between their work and family lives. Consequently, the current research investigates how despotic leadership creates emotional exhaustion, which influences employees' life satisfaction and work-family conflict. The moderating role of anxiety on the relationship between despotic leadership and the outcome variables has been probed as well. Since the dark side of leadership is more obvious in a highly collectivist and power distant culture (Luthans et al., 1998) as subordinates in high power-distant and collectivistic cultures are expected to obey what their supervisors order without questioning and accept power inequalities, we thus see Pakistani employees as ideal for this study (Naseer et al., 2016).

# LITERATURE REVIEW AND HYPOTHESES

# Despotic Leadership and Emotional Exhaustion

Emotional exhaustion occurs when emotional demands exceed an individual's ability to deal with interpersonal interactions at work (Maslach et al., 2001). There is growing evidence that aggressive leadership leads to harmful outcomes for subordinates, including anxiety, depression (Tepper, 2000), and burnout (Tepper, 2000; Harvey et al., 2007; Aryee et al., 2008; Wu and Hu, 2009). Despotic leaders, who are autocratic, inconsiderate, and exploitative, create stress among their subordinates, resulting in burnout (Ashforth and Lee, 1997; Den Hartog and De Hoogh, 2009; Schilling, 2009; Fontaine et al., 2010). Therefore, we hypothesize:

H1: Subordinates' perceptions of despotic leadership are positively related to their emotional exhaustion.

# Despotic Leadership and Work-Family Conflict

Despotic leaders are unethical and authoritarian, use an unethical code of conduct, and have little regard for others' interests (Naseer et al., 2016). In pursuing their self-interests, they can be domineering, controlling, vengeful, and exploitative (Bass, 1990; Howell and Avolio, 1992; Aronson, 2001). The harmful consequences of despotic leadership highlight the importance of understanding the effects of this kind of leadership on subordinates' lives. Work-family conflict has been defined as "a form of inter-role conflict in which the role pressures from the work and family domains are mutually incompatible in some respect" (Greenhaus and Beutell, 1985, p. 77), suggesting that "participation in the work (family) role is made more difficult by virtue of participation in the family (work) role" (Van Steenbergen et al., 2009; Kalliath et al., 2012). We examine work-family conflict since the emphasis of the study is on assessing despotic leadership as a social stressor that encircles the work and family life of the subordinates.

Work family conflict has been conceptualized into three types: time-based, strain-based, and behavior-based. Time-based conflict arises when the time committed to one role makes it difficult to participate in another role, such as when one has no time to participate in family functions or children's school events because of time spent at work. An example of strainbased conflict is being too tired from work to do home chores or help one's spouse. Behavior-based conflict occurs when one's emotional exhaustion from work leads to coming home in a bad mood and fighting with one's spouse.

According to Hoobler and Brass (2006), subordinates carry workplace aggression home in the form of behaviors that undermine their families. In a similar vein, Demsky et al. (2014) found a positive relationship between workplace aggression and work- family conflict. According to (Westman, 2001), despotic leadership increases tension in subordinates' marital relationships, weakening the family structure (Carlson et al., 2011). Despotic leaders demand unquestioned compliance and obedience from their subordinates (Schilling, 2009), are selfcentered, have low ethical standards (De Hoogh and Den Hartog, 2008), and exploit their subordinates for personal gain (Naseer et al., 2016). Therefore, we argue that despotic leadership is a workplace stressor that leaves the subordinates drained and emotionally exhausted, dramatically stressing their personal lives. Therefore, we hypothesize that:

H2a: Subordinates' perceptions of despotic leadership are directly and positively related to their work-family conflict.

# Despotic Leadership and Life Satisfaction

Life satisfaction is a critical indicator of an individual's overall well-being from evaluating his or her life (Karatepe and Baddar, 2006; Aryee et al., 2007; Erdogan et al., 2012); life satisfaction is widely accepted as a vital factor in an individual's quality of life (Pavot and Diener, 1993). There is ample evidence to support the strong relationship between experience at work and an individual's overall perspective of his or her life (Rain et al., 1991). Considering the significance of life satisfaction in measuring an individual's wellbeing, a few studies have examined the harmful effects of negative leadership on life satisfaction and have suggested that poor and unfair treatment by others in the form of abusive supervision (Tepper, 2000) and workplace bullying (Dzuka and Dalbert, 2007; Moore et al., 2012) is negatively related to life satisfaction. Following this line of discussion, we posit that, when a leader is manipulative, exploitive, and vengeful, subordinates' sense of personal control to cope up with such

pressures declines. As a result, subordinates feel emotionally exhausted and are likely to have low level of life satisfaction. Therefore, we hypothesize that:

H2b: Subordinates' perception of despotic leadership is directly and negatively related to their life satisfaction.

### Emotional Exhaustion as a Mediator

According to Westman et al. (2004), burnout is a significant predictor of work-family conflict. Emotional exhaustion, one of the core factors in burnout (Johnson and Spector, 2007), is most clearly linked to depletion of resources, as described by COR theory. Drawing from COR theory, we posit that despotic leadership may lead subordinates to deplete their personal and emotional resources and become exhausted, an effect that is likely to increase over time as the frequency of interaction with the supervisor increases (Grandey et al., 2004). Emotionally exhausted subordinates may have little energy left for family chores or family enriching activities, leading to work-family conflict (Gali Cinamon and Rich, 2010; Carlson et al., 2012). Using COR theory as a foundation, we observed that subordinates' experience of despotic leadership results in depletion of subordinates' energy, increasing emotional exhaustion and work-to-family conflict and diminishing life satisfaction. Therefore, we hypothesize:

H3: Subordinates' emotional exhaustion is (a) positively related to work-family conflict and (b) negatively related to life satisfaction.

H4: Subordinates' perceptions of despotic leadership are (a) positively related to work family conflict and (b) negatively related to life satisfaction via emotional exhaustion.

### Anxiety as a Moderator

Personality has been identified as an antecedent of work-family conflict (Michel et al., 2011), and work-family researchers have called for an examination of personality variables (Friede and Ryan, 2005; Michel and Clark, 2009; Michel et al., 2011) and job outcomes (Ceschi et al., 2016) in that context. Personality has been treated as both a mediator and a moderator between antecedents and work-family conflict (Wayne et al., 2004; Ceschi et al., 2016). According to Stoeva et al. (2002), negative affect mediates the relationship between stress and work-family conflict, and negativity moderates the relationship such that the relationship between stress and work-family conflict is stronger for individuals with high negative affect than it is for those with low negative affect.

Kant et al. (2013) suggested that leaders' negative behaviors are linked to subordinates' anxiety. Anxious subordinates perceive others negatively and are likely to increase their expressions of criticism and disapproval (Forgas and Vargas, 1998; Story and Repetti, 2006). Despotic leadership refers to aggressive behavior toward subordinates and to the exploitation that creates fear and stress among subordinates regarding their position in the organization (De Hoogh and Den Hartog, 2008). Those who face such issues at work carry the resulting emotional instability back home, where they often retreat from the family (Schulz and Martire, 2004; Story and Repetti, 2006) and are unavailable to help or support their families.

We argue that anxiety is high among subordinates who perceive their leaders' behavior as exploitative and unfair (Kant et al., 2013) and that this dyadic relationship eventually affects their personal domains. Therefore, subordinates who have a high degree of anxiety are more likely to experience work-family conflict and diminished life satisfaction than are those who have less anxiety.

H5: Subordinates anxiety moderates the relationship between their perception of despotic leadership and (a) their work-family conflict and (b) their life satisfaction.

# Proposed Research Model

**Figure 1** presents a model of our hypothesized relationships.

# MATERIALS AND METHODS

# Sample and Data Collection Procedure

To reduce common method variance, the three-wave data was collected (Podsakoff et al., 2003). The respondents were salespeople working in book-publishing houses. The publishing houses were approached through personal contacts to identify respondents, and surveys were distributed by the researchers to the respondents in person who agreed to participate in the survey voluntarily. Prior to their participation in the surveys, the participants completed consent forms that explained the purpose of the study and assured complete confidentiality. It was communicated that all the responses would be accessible to the researchers only, no individual level information would be made public and only aggregate information would be shared. These precautions helped us to deal with social desirability and made the respondents feel confident. The late respondents and non-respondents were contacted in follow-up to increase the response rate. The study was approved by the Riphah International University Ethical Research Committee.

The data collection was completed by means of three penand-pencil surveys fielded on site, one of which measured despotic leadership and anxiety, the second of which measured emotional exhaustion, and the third measured life satisfaction and work-family conflict (time-, strain-, and behavior-based conflict). After completing the surveys, participants placed them in sealed envelopes and returned them to the contact person. The researchers collected the filled responses from the contact person. No monetary reward or other incentive was offered to participants, and participation was voluntary.

Questionnaires were initially distributed to 400 salespersons, and 327 completed surveys were returned. The second survey was offered only to the 327 who returned the first survey, and 255 completed surveys were returned. These 255 received their third survey, and 245 were returned. The surveys were fielded 3–4 weeks apart. The final sample size after discarding incomplete questionnaires and matching three-time data was 224, for a final response rate of those who completed all three

surveys of 56 percent. As schools and most organizations in Pakistan use English, all survey questions were in English. All respondents were males, and all had reached at least the graduate level to ensure their solid understanding of English language.

### Measures

All measures were adopted from extant studies that had tested them in a variety of cultures, countries, and work settings. The use of established standardized scales to measure the study's variables reduces the likelihood of instrumentation errors (Luthans and Youssef, 2007). To establish convergent and discriminant validity, confirmatory factor analyses were conducted for all variables.

### Despotic Leadership

We used a six-item scale developed by De Hoogh and Den Hartog (2008) to measure despotic leadership. The items included "My supervisor is punitive and has no pity or compassion," My supervisor is in charge and does not tolerate disagreement or questions," and "My supervisor gives orders." Responses were given on a 5-point Likert scale, anchored at 1 for strongly disagree and 5 for strongly agree. The Cronbach's alpha for this scale was 0.80. Convergent validity was also established because all items loaded in a range of 0.45 to 0.74 with average variance extracted (AVE) = 0.50.

### Emotional Exhaustion

To assess employees' emotional exhaustion, we used a nine- item scale developed by Pines and Aronson (1988). Responses were rated on a 5-point scale, where 1 = never and 5 = very often. Examples of items are "I feel emotionally exhausted" and "I feel that I can't take it anymore." The Cronbach's alpha for this scale was 0.86. Convergent validity was also established because all items loaded in a range of 0.48 to 0.78 with AVE = 0.52.

### Job-Related Anxiety

Following Wagner et al. (2014), we used a four-item measure of anxiety drawn from (Mackinnon et al., 1999). Respondents indicated the extent to which they generally feel facts of anxiety (e.g., "nervous," "distressed") in their jobs using a 5-item Likert scale, where 1 = very slightly or not at all to 5 = very much. The Cronbach's alpha for this scale was 0.76. Convergent validity was also established because all items loaded in a range of 0.63 to 0.83 with AVE = 0.61.

### Work-Family Conflict

Work-family conflict was measured using the nine-item scale developed by Carlson and Kacmar (2000). We compared the three-factor model of work family conflict (time-based, strainbased, and behavior-based) with one overall-factor model. The overall-factor model produced better results than the three-factor model did, and the three dimensions were highly correlated, so we used the overall-, one-factor model. Responses were measured on 5-point Likert scale, anchored at 1 for strongly disagree and 5 for strongly agree. The Cronbach's alpha for this scale was 0.86. Convergent validity was also established because all items loaded in a range of 0.43 to 0.84 with AVE = 0.54.

### Life Satisfaction

We used a 5-item scale from (Diener et al., 1985) to measure life satisfaction. The five items included "In most ways, my life is close to my ideal" and "I am satisfied with my life." One item was dropped because of low factor loading. The Cronbach's alpha for the remaining four items was 0.75. Convergent validity was also established because all items loaded in a range of 0.51 to 0.80 with AVE = 0.56.

# RESULTS

Structure equation modeling (SEM) (Bollen, 1989) using AMOS 16 was employed to test the hypotheses and followed the twostep analytical strategy suggested by Anderson and Gerbing (1988). The first step involved a confirmatory factor analysis to develop an acceptable measurement model that defined the observed variables in terms of "true" latent variables (endogenous or exogenous) and a measurement-error term. At this stage, each latent variable was allowed to correlate freely with every other latent variable. In step two, we modified the measurement model to represent the hypothesized theoretical framework. This strategy provided an analytical method with which to identify a best-fit measurement model and an estimation of the proposed hypotheses. To test our mediation hypotheses, we also used the bootstrap technique using the confidence-interval method. Bootstrapping is a resampling method that involves creating a sampling distribution to estimate standard errors and to create the confidence intervals. Considered important for mediation analysis, bootstrapping is used to confirm the mediation effect because of its accuracy in computing confidence intervals for the mediation effect when the mediation effect is non-zero. It can be applied when the assumptions of large sample size and multivariate normality may not hold (Cheung and Lau, 2008).

**Table 1** shows the descriptive statistics, bi-variate correlations, reliability estimates, and AVEs. We performed a series of confirmatory factor analyses to establish the discriminant validity of the variables, tapped through the same source and time. In light of (Anderson and Gerbing, 1988, 1992) suggestions, we compared a two-factor unconstrained model with a single-factor constrained model for every possible pairing of variables from the same source. The results of these confirmatory factor analyses are presented in **Table 2**, which shows that, in every comparison, the unconstrained multiple-factor model provided a better fit than the singlefactor model did. All fit indices are within the range of a good model fit (Kline, 2005; Steiger, 2007; Tabachnik and Fidell, 2007).

After getting adequate model fit results for the measurement models, we estimated the path models using SEM to test the hypotheses. **Table 3** shows the results of SEM analyses, and **Figure 2** presents the standardized path coefficients of the best fitting model. We tested three structural models to determine which provided the best fit to data. The first model included indirect paths between despotic leadership and two outcomes work-family conflict and life satisfaction through emotional exhaustion. The second model included direct and indirect paths between despotic leadership and the same two outcomes through emotional exhaustion and revealed a direct path between despotic leadership and the two outcome variables. The results indicated that the indirect-path model between despotic leadership and the two outcomes through emotional exhaustion provided the best results for model fit indices (χ2 = 549.89, df = 324, χ2/df = 1.69, CFI = 0.92, NFI = 0.88, GFI = 0.90, TLI = 0.90, RMR = 0.19, and RMSEA = 0.05), as shown in **Table 3**.

The direct-path model provides the results for H1, which predicts that despotic leadership is positively related to workfamily conflict and negatively related to life satisfaction. The results provide support for both work-family conflict (β = 0.55, p < 0.001) and life satisfaction (β = −0.27, p < 0.001). In support of H2, despotic leadership is positively related to emotional exhaustion (β = 0.53, p < 0.001). Hypotheses 3a and 3b, that emotional exhaustion has a positive relationship with work-family conflict (β = 0.25, p < 0.01) and a negative relationship with life satisfaction (β = −0.15, p < 0.05), are also supported. Hypotheses 4a and 4b predict a mediating role of emotional exhaustion between despotic leadership and both outcome variables. We applied bootstrapping using a biascorrected confidence interval method to analyze the indirect effects. The results support the indirect effect of work-family conflict (indirect effect = 0.13, CI 95%, [0.04,0.27], p ≤ 0.01) but not life satisfaction (indirect effect = −0.08, CI95%, [−0.20,0.01], p > 0.05). All these direct and indirect effects are given in **Table 4**.

Hypotheses 5a and 5b propose a moderating role of anxiety between despotic leadership and work-family conflict and life satisfaction. Our moderation analysis employed Hayes' 13 macro PROCESS, and mean-centering was done for the independent and moderation variables (Aiken and West, 1991). The results indicate support for both variables, work-family conflict (β = 0.09, p < 0.05, CI [0.01,0.19], 1R <sup>2</sup> = 0.015) and life satisfaction (β = −0.23, p < 0.001, CI [−0.38,−0.08], 1R <sup>2</sup> = 0.037). The slope test indicates that the change in beta is in the same direction as proposed, at ±1 standard deviation of the moderator's mean value. The interaction plots shown in **Figures 3**, **4**


N = 224. <sup>∗</sup>Correlation is significant at the 0.05 level (two-tailed). ∗∗Correlation is significant at the 0.01 level (two-tailed). DL, Despotic Leadership; EE, Emotional Exhaustion; TA, Trait Anxiety; LS, Life Satisfaction; WFC, Work family conflict.

### TABLE 2 | Model Fit Indices for CFAs.

fpsyg-09-00601 May 3, 2018 Time: 11:47 # 7


N = 224. T1, time 1; T2, time 2. DL, Despotic Leadership; EE, Emotional Exhaustion; TA, Trait Anxiety; LS, Life Satisfaction; WFC, Work family conflict. Best model fits are given in bold.

TABLE 3 | Comparison of alternative path models.


N = 224. Age is controlled in all models. DL, Despotic Leadership; EE, Emotional Exhaustion; Outcomes, life satisfaction and work family conflict.

show that the interaction for work-family conflict is stronger when anxiety is high (i.e., β = 0.42, p < 0.001) than when it is low (β = 0.23, p < 0.001). Similarly, the interaction for life satisfaction is negative when anxiety is high (β = −0.40, p < 0.001), whereas it becomes insignificant when anxiety is low (β = −0.05, p > 0.05) as given in **Table 5**. These results are in line with our proposed moderation hypotheses.

### TABLE 4 | Standardized direct path coefficients of the hypothesized model.


### Bootstrap Results for Indirect Effects


N = 224. Bootstrap sample size = 5,000. Age is controlled in all models. ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, <sup>∗</sup>p ≤ 0.05. T1, time 1; T2, time 2. DL, Despotic Leadership; EE, Emotional Exhaustion; LS, Life Satisfaction; WFC, Work family conflict; LL, lower limit; CI, confidence interval; UL, upper limit.

### TABLE 5 | Moderation analysis.


N = 224. Unstandardized regression coefficients are reported. DL, Despotic Leadership; TA, Trait Anxiety; Bootstrap sample size = 5,000. LL, lower limit; CI, confidence interval; UL, upper limit. <sup>∗</sup>p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001.

# DISCUSSION

By integrating despotic leadership, anxiety, work-family conflict, and life satisfaction with the CORs theory, we find evidence of an indirect effect of despotic leadership on work-family conflict and life satisfaction via emotional exhaustion. Our findings indicate that emotional exhaustion partially mediates the relationship of despotic leadership with work-family conflict and life satisfaction. We also show that highly anxious employees are more prone respond negatively to despotic leadership, increasing work-family conflict and decreasing life satisfaction. These findings, which are in line with previous research (Story and Repetti, 2006; Carlson et al., 2011, 2012; Kant et al., 2013), suggest that despotic leaders have harmful effects on their subordinates' home lives and that these effects intensify when subordinates are anxious. These findings are in line with previous research. Given these findings, this study contributes to the literature on dark side of leadership, employee wellbeing, and CORs theory. This study also revealed a significant correlation between age of the subordinates and perceived despotic leadership and life satisfaction. The demographic variables are supposed to be controlled in studies intended to measure employee attitudes and behaviors (Riordan et al., 2003). Therefore, we controlled age in the mediation model.

### Theoretical Implications

This study makes theoretical contributions to both the dark side of leadership literature and the work-family literature. We extend both the despotic leadership and work-family literatures by investigating the relationships between despotic leadership and subordinates' work-family conflict and life satisfaction in Pakistan's cultural setting, which is ranked high in collectivism, uncertainty avoidance, and power distance (Hofstede, 1983). Our research shows that the negative effects of despotic leadership

extend beyond the work domain to affect subordinates' nonwork lives. We also provide insights into how despotic leadership relates to the family domain via emotional exhaustion. The interactions between despotic leaders and their subordinates leave subordinates emotionally exhausted and prone to workfamily conflict and low life satisfaction. Finally, we identify a boundary condition drawn by anxiety on the relationship between despotic leadership and the family domain.

# Practical Implications

An important practical implication is that despotic leadership has detrimental effects on the home lives and life satisfaction of subordinates who work in the service sector. An organization that fails to identify leaders who have despotic tendencies and an overarching desire for power risks having emotionally exhausted and dissatisfied employees. The ideal is for organizations to avoid appointing such leaders in the first place, but steps can also be taken to reduce subordinates' emotional exhaustion by giving them easy access to the human resources department, where they can give confidential feedback about the despotic supervisor. Confidentiality is essential as despotic leaders are likely to manipulate and harm subordinates who give such feedback. When feedback is provided about despotic leaders, grievances should be addressed by means of appropriate investigation. Checks and balances can help to prevent despotic leadership (Padilla et al., 2007). As booksellers' work is incentive-based, reward and incentives like company recognition and jobpromotion opportunities can help to increase subordinates' selfesteem (Ceschi et al., 2017) and reduce emotional exhaustion. Interventions like psychological training can also reduce emotional exhaustion and work disengagement (Costantini et al., 2017).

Another practical implication is that the harmful effects of despotic leadership on subordinates can be attenuated if HR fosters positive organizational climate for reducing despotic leadership and provides support to the subordinates who suffered. Engaging employees in such recovery activities as relaxation, personal control, psychological detachment, and

exercise can help them reduce the anxiety that leads to worklife conflict and life dissatisfaction (Erfurt et al., 1992; Sonnentag et al., 2008, 2010).

# Strengths, Limitations and Suggestions for Future Research

A major limitation to the present study is that all participants were male, and all were drawn from two publishing houses. Future research should examine both genders and additional occupations to determine the extent to which the findings can be generalized.

Another limitation is that all of the data was collected through self-reports. Studies that rely on self-reports can suffer from common method variance (Podsakoff et al., 2003), However, the study's research design minimizes such concerns, as the data for the independent variable and the moderator, that for the mediator, and that for work-family conflict and life satisfaction were collected at different times that helped to avoid common method variance (Podsakoff et al., 2012). A study on addressing the issues of common method variance by Johnson et al. (2011) found that 3-week gap between predictors and criterion variables decreased correlations between constructs by 43 percent. Moreover, nature of all variables require self-reports of the subordinates like rating the perceived despotic leadership. The time-lagged design also addresses reverse causality between variables in mediation models. Moreover, the correlation size from low to moderate reveals that there is no issue of common method variance among all study variables. To improve the results' accuracy, all of the data were collected from the same employees and matched time lagged responses. As participants responded to the survey voluntarily and at their convenience, we have considered the possibility of a self-selection bias. However, as the response rate across three waves of data collection was comparatively high, we argue that a strong influence of a selfselection bias on the subsequent results is unlikely.

The strength of the study is that it was conducted in Pakistan, which provides an ideal context in which to examine the negative aspects of despotic leadership (Naseer et al., 2016).

Subordinates in high power-distant and collectivistic cultures are expected to do what they are told to do by their supervisors and accept power inequalities. However, despotic leaders goes beyond controlling and self-serving behavior and are engaged in exploitative and unethical acts which can drain subordinates resources and heighten emotional exhaustion, reducing life satisfaction and increasing work-family conflict. Subordinates in certainty-avoidant cultures like that of Pakistan also prefer to stay in their jobs—even those they dislike—so they seldom do anything about the inter-role conflict that results in work-family conflict and life dissatisfaction and thus suffer in their home lives, especially if they have high levels of anxiety.

As the data for the current study was collected at three points in time, future researchers may consider measuring the daily effects of despotic leadership. Future research could also examine other factors that may influence the day-to-day variance in the effects of despotic leadership, such as lack of sleep. There is also room to clarify this interface further by examining additional moderators of such relationships, such as emotional intelligence, organizational justice, faith and social support.

### CONCLUSION

This study extends research on the dark side of leadership by showing that despotic leadership has negative effects on work-family conflict and life satisfaction. Despotic leadership is related to work-family conflict via emotional exhaustion, and the interaction of anxiety with despotic leadership has

### REFERENCES


negative effects on work-family conflict and life satisfaction. Thus, despotic leadership socially undermines subordinates. This study should stimulate research on how to control and attenuate the harmful impacts of despotic leadership on employees in service organizations in order to enhance their wellbeing.

### ETHICS STATEMENT

Research Ethics Committee at Riphah International University has reviewed the aforementioned research proposal and considers the procedures, as described by the applicant, to conform to the University's ethical standards and RIU guidelines. The committee has granted clearance from 03/02/2017 to 01/02/2018.

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct, and intellectual contribution to the work, and approved it for publication.

# FUNDING

The project was funded by ORIC, Riphah International University, Pakistan.


endoplasmic reticulum stress and mitochondria-dependent pathways in human tongue cancer SCC-4 cells. Oral Oncol. 45, 273–283. doi: 10.1016/j. oraloncology.2008.05.012


evaluation of factorial validity and invariance across demographic variables in a community sample. Pers. Individ. Dif. 27, 405–416. doi: 10.1016/S0191- 8869(98)00251-7


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Nauman, Fatima and Haq. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Is It Me or You?—How Reactions to Abusive Supervision Are Shaped by Leader Behavior and Follower Perceptions

### Birgit Schyns <sup>1</sup> \*, Jörg Felfe<sup>2</sup> and Jan Schilling<sup>3</sup>

<sup>1</sup> Department People and Organisations, Center for Leadership and Effective Organizations, Neoma Business School, Reims, France, <sup>2</sup> Department of Psychology, Helmut-Schmidt University, Hamburg, Germany, <sup>3</sup> Department of Economics and Social Sciences: Work and Organizational Psychology, University of Applied Administrative Sciences, Hannover, Germany

Edited by: Ronit Kark, Bar-Ilan University, Israel

### Reviewed by:

Steffen Robert Giessner, Erasmus University Rotterdam, Netherlands Laura Guillén, European School of Management and Technology, Germany

> \*Correspondence: Birgit Schyns birgit.schyns@neoma-bs.fr

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 19 September 2017 Accepted: 09 July 2018 Published: 27 July 2018

### Citation:

Schyns B, Felfe J and Schilling J (2018) Is It Me or You?—How Reactions to Abusive Supervision Are Shaped by Leader Behavior and Follower Perceptions. Front. Psychol. 9:1309. doi: 10.3389/fpsyg.2018.01309 There is a growing interest in understanding how follower reactions toward abusive leadership are shaped by followers' perceptions and attributions. Our studies add to the understanding of the process happening between different levels of leaders' abusive behavior (from constructive leadership as control, laissez-faire, mild to strong abusive) and follower reactions. Specifically, we focus on the role of perception of abusive supervision as a mediator and attribution as a moderator of the relationship between leader abusive behavior and follower reactions. Follower reactions are defined in terms of exit, voice, loyalty, and neglect. Two studies using a two point experimental design and vignettes and a cross-sectional field study were conducted. Perception partly mediates the relationship between leader behavior and reactions (Study 1 and 2). Different attributions (intention, control) moderate the relationship between the perception of abusive supervision and reactions in Study 2 and 3. In Study 2, attribution of intentionality of the leader behavior served as a moderator of the relationship between abusive supervision and loyalty, turnover, and voice. Attribution of intentionality reduced the relationship between perception of abusive supervision and reactions. Attribution of intentionality only strengthened negative reactions when milder abusive leadership was perceived. These results were not supported in Study 3. However, in Study 3, attribution to the supervisor' control served as moderator for loyalty and voice. A stronger relationship between the perception of abusive supervision and reactions emerged for high vs. for low attribution to the supervisor. The differences in results between the studies reflect that in Study 1 and 2 abusive behavior was manipulated and in Study 3 the perception of abusive supervision of actual leaders was assessed. Our findings show that avoidance of abusive supervision should be taken seriously and followers' perception and suffering is not only due to subjective judgment but reflects actual differences in behavior. The relationships are stronger in the field study, because, in practice, abusive behaviors might be more ambiguous. The research presented here can help leaders to better understand their own and the followers' role in the perception of and reaction to abusive supervision.

Keywords: abusive supervision, perception, attribution, withdrawal, voice, commitment

'One of the hardest tasks of leadership is understanding that you are not what you are, but what you're perceived to be by others.'

### Edward L. Flom

The last years have seen a growing interest in the topic of destructive leadership (Schyns and Schilling, 2013; Mackey et al., 2017) with the number of studies investigating the core construct of abusive supervision (Tepper, 2000) having significantly grown over the last decade (Martinko et al., 2013; Tepper et al., 2017). Schyns and Schilling (2013) define destructive leadership as "a process in which over a longer period of time the activities, experiences, and/or relationships of an individual or the members of a group are repeatedly influenced by their supervisor in a way that is perceived as hostile and/or obstructive" (p. 141).

Research into destructive leadership acknowledges that supervisors often do more than simply fail to exhibit constructive behavior toward their followers. Instead, recent meta-analyses demonstrate that strong negative relationships exist between destructive leadership and attitudes toward the supervisor, well-being, job satisfaction, and job-related attitudes (e.g., job engagement), while there are strong positive correlations between destructive leadership and counterproductive work behavior, negative affectivity, and perceived organizational injustice (Schyns and Schilling, 2013; Mackey et al., 2017). This makes the study of abusive supervision a priority. However, it sometimes appears that some followers suffer more under abusive supervisors than others (Brees et al., 2016; Tepper et al., 2017), which has led to an ongoing discussion regarding the distinction between actual abusive supervision and follower perception (Martinko et al., 2013; Tepper et al., 2017). The research presented here aims to add to our understanding of the role of perception in abusive supervision and its negative outcomes. Specifically, we investigate the role of follower perceptions in the relationship between actual behavior and follower behavioral reactions toward abusive supervision. We additionally examine the role of attribution in terms of further explaining different reactions following from abusive perception. In this model, we take into account follower characteristics that are typically associated with perception biases as control variables.

In order to capture reactions, we employ the exit, voice, loyalty, and neglect (EVLN) model (Withey and Cooper, 1989). The EVLN-model was introduced to describe reactions toward dissatisfaction at work. It includes reactions on different levels of severity and is therefore well-suited to examine reactions toward different types of leadership. While there are numerous studies on the outcomes of destructive form of leadership like abusive supervision, we suggest that the EVLN model is particularly helpful to systematize behavioral intentions as reactions toward abusive supervision due to the incremental approach it provides. This is an important first step toward understanding how abusive supervision fosters follower behavior in a more systematic way.

# INTRODUCTION

Research into destructive leadership mainly uses follower ratings of their leaders when assessing leader behavior (Tepper et al., 2017). The most used concept, and the one we use here, abusive supervision, is explicitly defined as a perception, namely: "Subordinates' perceptions of the extent to which their supervisors engage in the sustained display of hostile verbal and non-verbal behaviors, excluding physical contact" (Tepper, 2000, p. 178). However, first, using this approach to the measurement of abusive supervision does not provide sufficient information as to whether or not a leader really behaves differently toward their followers or whether or not followers perceive the same behavior in a different way. Second, even if a leader is perceived similarly across followers, their perceptions might not accurately reflect this behavior. Third, we argue that if abusive behavior is not perceived as abusive, the follower will not show negative reactions to these actual behaviors. Naturally, the opposite applies as well that if a behavior is not (or not intended to be) abusive but perceived as such, negative reactions will ensue. Therefore, it is important to study actual leader behavior and how it is related to outcomes via perception, rather than relying on perceptual measurements only. In order to address some of these issues and to be able to draw conclusions about actual behavior and its outcomes, we use an experimental approach, followed by a field study. We argue that even studies using multiple sources such as ratings from different followers of the same leader cannot really disentangle perception from actual behavior. This is in line with Tepper et al. (2017) who make the case that agreement between subordinates does not reflect objectivity, as leaders are likely to vary their behavior toward their followers. Therefore, low intra class correlations (ICC) of leadership ratings may be either indicative of perceptual biases of the followers or of leaders actually behaving differently to different followers.

In addition, previous studies cannot rule out the problem of reverse causality. For example, follower stress is related to the perception of abusive supervision (e.g., Tepper, 2000; Chen and Kao, 2009; for an overview: Schyns and Schilling, 2013). However, there is a possibility that leaders might react negatively to stressed followers so that follower stress influences actual leader behavior, or that the relationship is circular. Similarly, followers' poor performance or negative affectivity may lead to negative reactions by the leader which is perceived as abusive (Wang et al., 2015).

Therefore, to better understand the effects of abusive leadership and the validity of the perception of abusive supervision, it is necessary to investigate abusive supervision in a way which allows us to systematically control actual leadership behavior so that differences in follower perception of and reactions to identical behavior can be attributed to follower characteristics with confidence (see for example Martinko et al., 2012). By using an experimental vignette design (Aguinis and Bradley, 2014), we address these issues. Specifically, we use detailed descriptions of leader behaviors. This way, we can be certain (a) that systematic differences in the perception of different behavior are due to actual behavior and any differences in the perception of (the same) behavior is due to rater effects, and (b) that the relationship between leader behavior and reactions is causal as the leader behavior cannot be influenced by follower behavior or reactions. We are also testing (c) in how far abusive supervision is purely an unspecific perception of any non-positive leadership behavior by including another nonpositive leadership style (i.e., laissez-faire). At the same time, by focusing on the EVLN-model relating to outcomes, we investigate effects of leader behavior on outcomes of different severity.

In our studies, we specifically focus on abusive supervision for several reasons. First, abusive supervision is defined as follower perception, thus acknowledging that the same behavior might be perceived differently by followers and can lead to different reactions. Second, we want to add to the discussion regarding the relationship between actual abusive supervision and follower perception/ attribution (Martinko et al., 2013; Tepper et al., 2017) and follower related outcomes. In their comprehensive overview, Martinko et al. (2013, see also Brees et al., 2016) convincingly argue that it is important to distinguish between perceived and actual abusive supervision as we cannot be sure if perceived leader abuse is a valid proxy for actual behavior. In our study, we examine whether or not the perception of abusive supervision mediates the relationship between actual leader behavior and reactions to abusive supervisor behavior, also taking into account rater characteristics commonly found to influence perceptions (Brees et al., 2016).

We therefore contribute to the existing body of knowledge on abusive supervision by examining the degree to which follower reactions are influenced by actual leader behavior through perceptions. We also add to the question of causality by increasing our knowledge about the most likely direction of relationship between leader behavior and follower reactions. Moreover, by comparing different leadership behaviors (i.e., strong abuse, mild abuse, and laissez-faire leadership), we contribute to the knowledge of the impact of different intensities of negative leadership (Schilling, 2009). In addition, and in line with previous literature suggesting that attribution might be relevant in follower perceptions and reactions to destructive leadership (Martinko et al., 2013), we include attributions to further clarify the process between leader behavior/ follower perceptions and follower reactions. Specifically, we include attribution of intentionality and attribution to the supervisor' control in our study. We argue here that attributions can increase the effect of perceptions of abusive supervision on follower reactions. If abusive leadership is perceived as intentional or as under the control of the supervisor (as opposed to, e.g., results out of incompetence or pressure from outside), followers should react more strongly to the abusive behavior. **Figure 1** displays the research model for our studies. In the following we will first draw upon the path between abusive leadership and follower reactions, then address the mediating role of perception and the moderating role of attribution, and finally outline the potential bias of followers' characteristics on perception. Abusive supervision is something that organizations should avoid, and knowing more about the reactions toward abusive supervision can help them do so.

# ABUSIVE LEADERSHIP AND FOLLOWER REACTIONS

Very often findings of relationships between follower ratings of abusive supervision and outcomes are interpreted as if those follower ratings are a direct and perfect assessment of actual leader behavior (see Martinko et al., 2013, for a critique). However, follower ratings are influenced by rater characteristics such as personality, implicit leadership theories, or affect (for an overview see Hansbrough et al., 2015). Consequently, there is a need to better understand how leader behavior leads to follower reactions via perceptions.

There is, obviously, a myriad of possible reactions to abusive supervisions, and prior research studied a wide variety of outcomes (Schyns and Schilling, 2013). Reactions can happen on different levels (cognitive, emotional, and behavioral) and various coping strategies might come into play when followers are exposed to abusive leadership. We base our choice of reactions on the exit, voice, loyalty, and neglect (EVLN)- Model (Withey and Cooper, 1989), which addresses possible behavioral reactions of employees when facing dissatisfactory work situations. That is, we suppose that exit, voice, loyalty, and neglect should all be related to supervisory abuse. This framework is particularly suited to our research as the EVLN framework describes reactions differing in severity. This is helpful for our purposes as we are interested in different levels of leadership which can lead to different levels of reactions.

The most serious reaction would be considering quitting the job (exit). Previous research has shown that abusive leadership increases turnover intentions (Schyns and Schilling, 2013). Although there is a difference between intention and actual behavior, frequently thinking about quitting reduces engagement and enhances the probability of quitting as soon there is an appropriate opportunity.

A more active, however, indirect coping strategy is actively complaining or reducing voice behavior (voice). In comparison to an increasing number of studies focused on aggressive and retaliatory responses of followers (Martinko et al., 2013), actively complaining as a more constructive behavioral alternative did not receive much interest in prior research, so that the current study enters new territory in this respect (Study 1). We also included (in Study 2 and 3), the more traditional assessment of voice (i.e., speaking up), assuming that followers subjected to abusive supervision will show less voice.

While the aforementioned strategies clearly indicate that followers do not accept the perceived behavior, it is also possible that followers lower their attitudes toward their leaders. Indeed, in Schyns and Schilling's (2013) meta-analysis, follower attitudes toward their leaders were the strongest outcome of destructive leadership. Thus, we assess lack of acceptance as loyalty (Study 1). In Study 2, we used a more traditional assessment of loyalty, that is, supervisor commitment, which we assume will decline with abusive supervision.

Finally, neglect is another passive strategy relevant in the study of abusive supervision. Followers may pretend to ignore the situation but in truth brood angrily (Study 1) over the unjust

treatment and hope for opportunities to take revenge on the leader or they might passively withdraw from the relationship with the leader (Study 2 and 3).

In the literature around destructive leadership, there is a discussion around how abusive laissez-faire leadership is perceived and how negative it is for follower outcomes. While Schyns and Schilling (2013) explicitly excluded laissez-faire from their meta-analysis arguing that doing nothing is not destructive enough to be a part of destructive leadership, others have argued that laissez-faire is destructive due to its considerable negative consequences (Skogstad et al., 2007, 2014). In line with Schyns and Schilling (2013), we expect that followers react less negatively to non-positive leadership behaviors such as laissez-faire than when exposed to strong abusive leader behavior. At the same time, based on previous research regarding the outcomes of laissez-faire (Skogstad et al., 2007, 2014), we assume that laissezfaire will still instill negative follower reactions. Consequently, with respect to reactions toward the different leadership styles, we assume that reactions are stronger for strong abusive leader behavior than for mild abusive leader behavior, laissez-faire, or constructive leadership. In this respect, the present study offers the opportunity to investigate different forms of negative/ destructive leadership. Experimentally manipulating these types of behavior is especially useful with regard to severe destructive leadership which is difficult to investigate in a field study due to its low frequency.

Hypothesis 1: Displayed abusive supervision behaviors are positively related to reactions of (a) exit, and (b) neglect, and negatively related to (c) voice and (d) loyalty. The strongest reaction will emerge for strong abusive leader behavior.

# PERCEPTION AS MEDIATOR BETWEEN LEADER BEHAVIOR AND FOLLOWER REACTIONS

Given our argument that perception is the most important predictor of reactions toward abusive supervision (Martinko et al., 2013), in the sense that if abusive behavior is not perceived as such, there will also not be a reaction toward this behavior, we contend that perception will mediate the relationship between abusive leader behavior and reactions. As Martinko et al. (2013) argued, it is important to distinguish between abusive supervisor behavior and its perception and to investigate its relationship in more depth. As outlined in the introduction, it is important to clarify if the perception of abusive supervision which is typically measured in survey research is a valid proxy for actual behavior. Schyns and Schilling (2013) underline that hostility and obstructiveness of destructive leadership are and can only be subjective evaluations. As Tepper (2000, p. 178) put it "the same individual could view a supervisor's behavior as abusive in one context and as non-abusive in another context, and two subordinates could differ in their evaluations of the same supervisor's behavior." It seems highly plausible that differences in the perception of whether or not one has been exposed to abusive supervision should be of major importance for subsequent follower reactions. We therefore assume that those evaluations are a necessary prerequisite for any reaction by the follower. Thus we propose the following hypothesis.

Hypothesis 2: The relationship between displayed abusive supervision behaviors and reactions (a) exit, (b) voice, (c) neglect, and (d) loyalty is mediated by perceptions of abusive supervision.

# ATTRIBUTION OF ABUSIVE SUPERVISOR BEHAVIOR AS A MODERATOR IN THE RELATIONSHIP BETWEEN PERCEPTIONS OF ABUSIVE SUPERVISION AND FOLLOWER REACTIONS

We agree with Wang et al. (2015) that it is important to understand the reasons why some individuals react more strongly to perceptions of abusive supervision than others. Attributions have been shown to be important factors in predicting workplace outcomes and reactions (see Harvey et al., 2014, for a recent meta-analysis). We argue that attributions of intentionality and control are particularly relevant here as they will determine the members' perceptions of the leadership behavior (Dasborough and Ashkanasy, 2002).

As Ferris et al. (1995) point out, the way we perceive others' motives for behavior has a distinctive impact on our interpretation and reaction following that behavior. Specifically, we assume that attributing intentionality to perceived abusive behavior will increase the negative effect that abusive supervision has on reactions. While one could argue that abusive behavior is mostly seen as intentional, we think that there can be many reasons to be abusive and that raters will acknowledge these reasons when thinking about intentionality. For example, leaders might push on the pressure they receive from their leader, they might be incompetent, or they might be following norms of their industry or organization. Thus, the extent to which raters attribute intentionality to perceived abusive behaviors can differ. We argue that if raters assume that supervisors use abusive behaviors on purpose rather than maybe due to circumstances, they should react stronger to the abuse. Therefore, we also examine the role of attributions in explaining reactions toward abusive supervision (Study 2 and 3).

Empirically, Lyu et al. (2016) found that attribution style influences the relationship between perceptions of abusive supervision and reactions. They reasoned that "individuals with low hostile attribution bias may attribute their supervisor's hostility as unintentional" (p. 72). More generally, attribution theory (Kelley, 1967; Weiner, 2018) states that causes for behavior can be attributed to person vs. situational reasons. In terms of abusive supervision, we assume that attributions to the person are relevant for reactions. One reason for this assumption is that when the reason for the behavior as situated within the supervisor (rather than the situation) then the behavior is likely to be repeated. Comparing internal, external, and relational attributions for supervisor abuse, Burton et al. (2014) found that attribution to the supervisor was most strongly related to injustice perceptions and related indirectly through injustice to aggression toward the supervisor and lower OCB. Thus, theory and earlier findings argue that attribution to the supervisor is relevant for the reactions of followers to abusive supervision. Therefore, we assume here that the reactions toward abusive supervision will be stronger if followers see the reason for the behavior to be situated in the supervisor. In summary, we propose:

Hypothesis 3: The mediation of perceptions of abusive supervision of the relationship between displayed abusive supervision behaviors and reactions (a) exit, (b) voice, (c) neglect, and (d) loyalty is moderated by attributed intention/control, such that the relationships will be stronger when the leader's behavior is attributed to the supervisor or his/her intention/control.

# FOLLOWER CHARACTERISTICS AS CONTROL VARIABLES

Based on Martinko et al.'s review (2013) as well as previous research on followers' personality on the perception and acceptance of transformational leadership (Felfe and Schyns, 2010), we argue that follower characteristics influence the perception of abusive supervision. We particularly assume that perceivers who are high in trait negative affectivity, hostile attribution style, anxiety, and irritation (stress) are more likely to perceive leader behavior as abusive. In Mackey et al.'s (2017) meta-analysis, negative affectivity was the strongest antecedent of the perception of abusive supervision and therefore we included negative affectivity in our study in order to control for this potential bias. Brees et al. (2016; see also Martinko et al., 2013) highlight the role of hostile attribution style, which we also take forward as a follower characteristic. Hoobler and Brass (2006) define hostile attribution style based on Adams and John (1997) as "an extra-punitive mentality where individuals tend to project blame onto others." Because they interpret others' behaviors in a negative way, individuals high in hostile attribution style are likely to report more abusive supervision than individuals low in hostile attribution style. In addition, we include anxiety (as a trait) as a control variable. Mawritz et al. (2014) were able to show that leader anxiety as a state acts as a mediator between leader's hindrance stress (due to exceedingly difficult goals) and perceived abusive supervision. Trait anxiety describes the rather stable tendency of a person to respond to threatening situations with more intense feelings of tension and apprehension and heightened autonomic nervous system activity (Spielberger et al., 1971). We assume that individuals who are more anxious interpret leader behavior more negatively as they are more likely to experience situations as threatening their self-esteem.

We also include follower stress in our analyses as, in addition to the effects of personality on the perception of abusive supervision, we expect that followers will be less tolerant and show stronger reactions to abusive supervision when their experienced strain is high. The more followers feel stressed, the more likely they are to perceive abusive supervision (e.g., Tepper, 2000; Chen and Kao, 2009) and react more strongly toward abusive behavior due to their lack of resources to selfregulate (conservation of resource theory, Hobfoll, 1989, and ego-depletion, Baumeister et al., 1998).

# PRE-STUDY: TEST OF VIGNETTES

Similar to previous research in leadership (e.g., Butterfield and Powell, 1981; Rush et al., 1981; Felfe and Schyns, 2006), we used vignettes to describe leader behavior. As the vignettes were used in this study for the first time, we conducted a pre-study to check if they indeed reflect different leadership behavior as we intended them to do. The text of the vignettes is displayed as Supplementary Material. First, we examined in how far the vignettes differ in terms of participants' perceptions of our main concept abusive supervision. We additionally tested our vignettes in terms of how much the described leader is liked by the raters, assuming that the more abusive the leaders are, the less they are liked. We base this assumption on Schyns and Schilling's (2013) finding that abusive supervision is negatively related to follower attitudes toward the leader. In addition, Xu et al. (2012) found a strongly negative relationship between abusive supervision and Leader-Member Exchange, which includes elements of liking. Finally, we investigated to what extent raters would rate the described leaders as more or less leader-like. Again, we assume the more abusive the leader was described, the less leader-like they would be rated. However, also laissez-faire leaders should be rated as not very leader-like as they avoid leading.

# Sample and Procedure

We collected a sample of 223 full-time employed participants who currently have a supervisor via an online panel service (Qualtrics)<sup>1</sup> . In order to ensure that participants read the descriptions carefully, we asked them several multiple choice questions as attention checks ("What kind of meeting is it?," "What is interrupting the meeting?," "What is the main topic of the meeting?," and "The employee is a..." man/woman). For each attention check question, we provided three choices (e.g., for question 1, "Face to face business meeting," "Team meeting," or "Project kick-off meeting"). Overall, 16 participants did not pass these attention checks, that is, did not answer the questions correctly, leaving a sample of N = 207.

The gender was equally distributed (N = 104 men, N = 103 women). The average age was 25 years old (SD = 10 years) and participants had on average worked for their supervisor for 6 months (SD = 42 months). We used a between-participants design. The group sizes were as follows: constructive N = 47, laissez-faire N = 60, mild abusive N = 46, and strong abusive N = 54.

### Instruments

Perception of Abusive Supervision was assessed using Tepper's (2000) 15 item instrument asking the participants to rate the displayed leader's behavior. The scale ranges from 1 = I cannot remember him/her ever using this behavior to 5 = He/she uses this behavior very often. The reliability was α = 0.96. Liking was assessed using Engle and Lord's (1997) 5 items instrument. The scale ranges from 1 = strongly disagree to 5 = strongly agree. The reliability was α = 0.96. Generalized Leadership Impression (GLI; Cronshaw and Lord's (1987) was used to assess leader-likeness. The instrument consists of 5 items. The scale ranges from 1 to 5 with different anchors. A high value indicates a high leadership impression. The reliability was α= 0.91.

### Results and Discussion

**Table 1** displays the mean values for the four vignettes regarding abusive supervision, liking, and generalized leadership impression. In terms of the perception of abusive supervision, the differences were as expected with constructive leadership being perceived as the least abusive. All mean differences were significant (p < 0.01), except for the difference between mild and strong abusive supervision. The ranking of the means was in the expected direction, with constructive leader behavior having the lowest mean and strong abusive leader behavior showing the highest mean.

In terms of liking, there was also a non-significant difference between mild and strong abusive supervision. Here, also the difference between laissez-faire and mild abusive was not significant. The ranking of the means was in the expected direction, with constructive leader behavior having the highest mean and strong abusive leader behavior showing the lowest TABLE 1 | Mean values for abusive supervision, liking, and generalized leadership impression for the four vignettes (pre-study).


Standard deviations in brackets. Pre-study—Abusive Supervision F(3, 203) = 37.83, p < 0.001; Liking F(3, 203) = 44.87, p < 0.001; GLI F(3, 203) = 43.74, p < 0.001; Study 1—Abusive Supervision F(3, 306) = 93.22, p < 0.001; Liking F(3, 306) = 147.62, p < 0.001; GLI F(3, 306) = 134.58, p < 0.001; Going across rows, mean values with different letters are significantly different from each other (minimum p < 0.05). "a" stands for the highest mean value, b for the second highest, etc. (post-hoc Bonferoni test).

mean. For the impression of the leaders as leader-like, constructive was significantly different from all other styles. That is, only the leader showing constructive behavior was considered leader-like. Apart from the leadership impression, all outcomes were in the expected ranking order from constructive via laissezfaire and mild abusive to strong abusive. We assume that leaders who show laissez-faire are equally considered unlike leaders as abusive leaders. With a larger sample size, the results would have most likely become significant. Also, some values were already low, leading to a floor effect and thus a lower likelihood of finding significant differences. However, we take the correct ranking of the vignettes in terms of our criterion variables as an indicator that our vignettes are a useful means to represent different leadership styles<sup>2</sup> . Hence, based on these results, we assume that we can use the different vignettes as representing varying degrees of abusive behavior (from non-abusive: constructive to strong abusive).

# STUDY 1

### Design and Procedure

We used an online provider (Qualtrics) to collect the sample. The study reported here was part of a larger study. In order to separate measurements, we asked participants to fill out two questionnaires. At time 1, we asked them to indicate stable characteristics, that is, negative affectivity, and hostile attribution style. At time 2 (about a week later), we asked participants about their stress (irritation). We also gave them the descriptions of the

<sup>1</sup> Since all data reported here were collected using panel providers, no informed consent forms were used. Participants in panel studies can chose to take part in studies for a small recompensation. The first page of each study contained information about the study as well as that data would be treated confidentially and used for scientific purposes. Ethical consent from the first author's institutions was gained via self-assessment as no particular ethical issues were detected that would merit the involvement of an ethics committee. This is standard practice at both institutions.

<sup>2</sup>We conducted the same manipulation check in Study 1 with comparable results (see **Table 1**). The reliability for liking was α = 0.97. For Generalized Leadership Impression, the reliability was α = 0.91.

leaders to read (between participants design: constructive N = 71, laissez-faire N = 76, mild abusive N = 82, and strong abusive N = 81). Only participants who answered both questionnaires and correctly responded to the attention check questions (see pre-study) were maintained in the sample to test the hypotheses. Ninety participants did not pass the attention checks. After reading the vignette, participants rated the described leader on the abusive supervision scale and indicated their anticipated reactions to such a leadership behavior.

### Participants

The sample consisted of 310 full-time working employees, 162 men, 148 women, with an average of age 45 (SD = 11). The average work experience was 22.6 years (SD = 12.13). We did not ask about managerial experience as we were interested in our participants' views as followers.

### Instruments

### Follower Characteristics as Control Variables: Negative Affectivity

Follower characteristics as control variables: Negative affectivity (T1) was assessed using 10 items relating to negative affectivity of the PANAS (Watson et al., 1988). The reliability was α = 0.91 (1 = very slightly / not at all to 5 = extremely). Hostile attribution style (T1) was assessed using an instrument by Hoobler and Brass (2006). The reliability for six items was α = 0.76 (1 = strongly disagree to 5 = strongly agree). Stress was assessed (at T2) using the irritation instrument by Mohr et al. (2006). The reliability for the 8 items was α = 0.90 (1 = strongly disagree to 7 = strongly agree).

### Leader Behavior

Leader behavior was displayed as described in the pre-study.

### Perception of Abusive Supervision

We used Tepper's (2000) 15-item instrument to assess the perception of abusive supervision of the displayed leader behavior at T2. The reliability was α = 0.96.

### Reactions to Abusive Supervision

We created an instrument to assess the reactions to abusive supervision. We were particularly interested in the acceptance of abusive supervision (loyalty), complaining about the supervisor (voice), in how far people wanted to quit (exit) after being exposed to an abusive supervisory style, and in how far they anticipated showing an anger reaction (neglect) to abusive supervision (based on the EVLN-model). We used two items to assess each of those reactions. For the first item of each reaction, the response was 1 = certainly not to 6 = certainly; for the second item of each reaction, it was 1 = very poor way of dealing with it, 4 = very appropriate way of dealing with it. For Quit, we used: "I think about leaving and changing my job." Anger was assessed with "I swallow my anger and wait for a chance to pay my supervisor back by some means or other." For complain, we used "I'm going to complain to a higher authority, because this is not the way you should treat people." Finally, accept was assessed with "I will do as I am told, because my supervisor is right and it is my fault." An exploratory factor analysis supported the four factor structure.

### Results

We examined in how far the vignettes differed in terms of perceived abusive supervision and reactions using ANOVAs (see **Table 2**). The results show that the perception of abusive supervision differed between the vignettes in the expected direction. With regard to H1, for the reactions "quitting" and "anger" the differences between laissez-faire, mild abusive, and strong abusive supervision were not significant but all are significantly different from constructive leadership in the expected direction, namely that for constructive leadership these reactions were significantly lower compared to all other conditions. For the reaction "complaining," all differences were significant apart from the differences between laissez-faire and mild abusive supervision. For "acceptance" differences between laissez-faire, mild abusive, and strong abusive supervision were not significant. Here, the differences between constructive and laissez-faire as well as mild and strong abusive were significant. The difference between mild abuse and laissezfaire was significant. The difference between mild and strong abusive was significant at p < .05. These results lend support to Hypothesis 1. Intercorrelations are shown in **Table 3**.

In order to test H2, we conducted mediation analyses using the PROCESS macro (Preacher and Hayes, 2008). Based on our pre-test of the vignettes, we used the vignettes as a continuous predictor, representing a continuum from not abusive at all (constructive) to strong abusive behavior. We used negative affectivity and hostile attribution style to control for the influence of personality and stress to control for resource depletion that might also affect the perception of abusive leadership<sup>3</sup> . Apart from the mediation of perception of abusive supervision on the relationship between abusive supervision behavior and anger reactions, all other mediation effects were supported, lending partial support to our H2 (see **Figure 2**). With respect to

TABLE 2 | Mean values for reactions for the four vignettes (Study 1).


Standard deviations in brackets. Quitting F(3, 306) = 34.74, p < 0.001; Anger F(3, 306) = 5.92, p < 0.005; Complain F(3, 306) = 89.28, p < 0.001; Accept F(3, 306) = 32.47, p < 0.001; Going across rows, mean values with different letters are significantly different from each other (minimum p < 0.05). "a" stands for the highest mean value, "b" for the second highest, etc. (post-hoc Bonferoni test).

<sup>3</sup>Testing the model without control variables does not change the results.


TABLE 3 | Intercorrelations for abusive supervision, control variables, and reactions (Study 1).

\*p < 0.05 (2-tailed), \*\*p < 0.01 (2-tailed).

the control variables, we did not find any relationships with perception. The control variables were, however, related to some of the reactions: for quitting irritation was significant, for anger, irritation and hostile attribution style, and for accepting, negative affectivity.

### Discussion Study 1

The aim of our first study was to examine in how far different leader behaviors are related to different follower reactions, mediated by the perception of abusive supervision. As expected, we found the strongest effect of strong abusive behaviors on reactions. There has been some discussion around the negative effects of laissez-faire leadership (Skogstad et al., 2014) and whether or not laissez-faire might be even worse than abusive supervision. Our results do not support this consideration. Although mild abuse and laissez-faire did not differ on all outcome variables, they did differ in terms of the perception of how abusive the described leader is.

We also found that perception of abusive supervision partially mediates the relationship between behavior and most reactions. We could show here that in the process of abusive supervision, perception is relevant and can influence the strength of the reaction to leader behavior. At the same time, it is clear that leader behavior remains a strong influence on reactions. Perception appears to be a valid proxy of actual behavior; however perception does not capture the entire effect of behavior. Moreover, some follower characteristics (irritation, hostile attribution style, and negative affectivity) were related followers' reaction but not to perception. This also means that follower reactions cannot be simply put down to sensitivities of the follower but that leader behavior is a crucial factor in reactions toward negative leader behavior.

### STUDY 2

### Overview

In Study 2, we extended Study 1 and addressed limitations. First, in Study 1, reactions in terms of EVLN were measured with a self-constructed questionnaire. In Study 2, we use more established questionnaires to assess reactions. In addition, we also asked participants to rate their attribution of intentionality of the described leader behavior. We assumed that attribution of intentionality will moderate the relationship between perception of leadership and reactions. In line with our theoretical assumptions regarding influences of follower characteristics on ratings of abusive supervision, we added trait anxiety to control for the influence of follower characteristics. We assume that anxious individuals interpret leader behavior more negatively and react accordingly.

# Design and Procedure

We again used an online provider (respondi) to collect the sample. As in Study 1, we asked participants to fill out two questionnaires. At time 1, we asked them to indicate stable characteristics, that is, negative affectivity, hostile attribution style, and trait anxiety. At time 2 (about 2 days later), we first asked participants about their experienced stress (irritation). We then gave them the descriptions of the leaders to read (between participants design: constructive N = 60, laissez-faire N = 57, mild abusive N = 56, and strong abusive N = 61). Participants rated the described leader on the abusive supervision scale (perceived leader behavior), answered four items regarding the attribution of the described behavior (intentional) and finally, indicated their anticipated reactions to such a leadership behavior.

# Participants

As before, we only included participants who filled in the questionnaires at both times and passed the attention checks (see pre-study for the questions). Sixty-three failed our attention checks. The final sample consisted of N = 234 participants. Of those 141 were men, 93 women. Most participants were between 45 and 54 years old (30%).

### Measures

### Follower Characteristics as Control Variables

Negative affectivity (T1, α = 0.90), hostile attribution style (T1, α = 0.75), and stress (T2, α = 0.92) were assessed using the same instruments as in Study 1. Anxiety (T1) was assessed with four items. One item stemmed from the neuroticism subscale of the Big Six (Ashton et al., 2004) and three were taken from the BFI (John and Srivastava, 1999) (1 = strongly disagree to 5 = strongly agree; α = 0.89).

### Leader Behavior (T2)

Leader behavior (T2) was displayed with the vignettes described in the pre-study and Study 1.

### Perception of Abusive Supervision (T2)

We again used Tepper's (2000) abusive supervision scale (α = 0.97).

### Attribution of Intentionality (T2)

We used one item to assess attribution of the described leader behavior. Participants were asked to indicate on a scale from 1 (strongly disagree) to 7 (strongly agree) in how far they agree with the following statement. "The leader behavior: . . . is clearly intentional on part of the supervisor."

### Reactions to Abusive Supervision (T2)

As in Study 1, we based our assessment of reactions to leadership on the EVLN-model (Withey and Cooper, 1989). We asked people to indicate "If you were in this situation, please rate what would be most likely for you to do or think in this situation" on different instruments. We assessed Exit using three items relating to turnover intention (Hom et al., 1984; Mitchell et al., 2001). A sample item reads "Would you intend to leave the organization in the next 12 months?" (α = 0.96; 1 = certainly not to 6 = certainly). Voice was assessed using Liang et al. (2012) ten item measure (α = 0.93; 1 = never to 5 = frequently). In line with Liang et al. (2012), we differentiated between promotive (α = 0.97) and prohibitive voice (α = 0.88). For loyalty, we used the 5 items supervisor commitment scale (α = 0.96; 1 = never to 5 = frequently; Felfe et al., 2006). For neglect, we used a three items measurement of withdrawal (α = 0.95; "I would withdraw from this supervisor"; 1 = never to 5 = frequently) based on Aquino et al. (2006).

### Results

As in Study 1, we calculated ANOVAs to examine the differences between the vignettes concerning abusive supervision perceptions as well as the reactions. **Table 4** shows the results. All mean values ranked in the expected order. With respect to the reactions (H1), again the ranking of the mean values was as expected, though some differences between the means were not significant. Overall Fs were significant apart from prohibitive voice. This lends support to H1.

Prior to conducting our mediation and moderation analyses, we correlated the rater characteristics assessed at T1 (negative affectivity, anxiety, and hostile attribution style) as well as stress assessed at T2 with abusive supervision assessed at T2 in order to examine rater biases in abusive supervision ratings independent of the leader behavior described to the participants. As can be seen in **Table 5**, the correlations between the perception of abusive supervision and negative affectivity, anxiety and irritation were positive and significant. There was no significant correlation between hostile attribution style and abusive supervision. We also conducted a hierarchical regression analysis with all control variables predicting abusive supervision

TABLE 4 | ANOVA results for the differences between the vignettes on perceptions and reactions (Study 2).


Abusive supervision F(3, 230) = 20.99, p < 0.001; turnover intention F(3, 230) = 31.24, p < 0.001; supervisor commitment F(3, 230) = 22.03, p < 0.001; withdrawal F(3, 230) = 30.44, p < 0.001; prohibitive voice F(3, 230) = 1.39, n.s., and promotive voice F(3, 230) = 17.75, p < 0.001. Going across columns, mean values with different letters are significantly different from each other (minimum p < 0.05). "a" stands for the highest mean value, "b" for the second highest, etc.

(first negative affectivity, second hostile attribution style, third anxiety, and last irritation). While the R 2 change was significant for negative affectivity, none of the controls was significant when all were entered into the regression. Therefore, only negative affectivity was included as a control in our moderated mediation analyses.

While there was no significant correlation between abusive supervision and attributed intentionality (r = −.03), intentionality was positively related to all outcomes apart from prohibitive voice. Perception of abusive supervision was related to all outcomes as expected, again apart from prohibitive voice.

In the following step, we conducted mediation analyses using Preacher and Hayes (2008) bootstrapping method to examine if we can replicate the results of Study 1 with the improved assessments for the outcome variables. As before, we used the vignettes as a continuous predictor. In terms of H 2, **Figure 3** depicts that all indirect effects are significant apart from prohibitive voice. The direct effects were substantially reduced when perception of abusive supervision was included as a mediator, again apart from prohibitive voice. In terms of the control variables, negative affectivity was significant for all but supervisor commitment at least on the p < 0.05 level. **Figure 3** shows the mediation results with control variable (negative affectivity). This lends support to our H2.

Finally, we tested a moderated mediation model (H3), in which attribution of intention moderates the relationship between the mediator perception of abusive supervision and the reaction variables. As intention in our model is only relevant when it relates to negative behavior, we only used the two groups that read the negative leadership behavior description, that is, mild and strong abusive leadership. This reduced our sample size to N = 117.

For exit, the direct effect of the vignettes was .45 (p < 0.05). The interaction between perception of abusive supervision and intention was significant (−0.15, p < 0.05). For withdrawal, the direct effect of the vignettes was significant at .45 (p < 0.01). The interaction between perception of abusive supervision and intention was significant (−0.09, p < 0.10). For supervisor commitment, the direct effect of the vignettes was −0.04 (n.s.). The interaction between perception of abusive supervision and intention was significant (0.10, p < 0.05). For prohibitive voice, the direct effect of the vignettes was −.09 (n.s.). The interaction between perception of abusive supervision and intention was significant (0.21, p < 0.01). For promotive voice, the direct effect of the vignettes was significant (−0.62, p < 0.01). The interaction between perception of abusive supervision and intention was significant (0.13, p < 0.10). In summary, apart from withdrawal and promotive voice (both interactions p < 0.10) as independent variables, all moderated mediation effects were significant at least at p < 0.05 (see **Figures 4A–E**).

We plotted the interactions for turnover intention, supervisor commitment, prohibitive voice (see **Figures 5**–**7**) 4 . For high

<sup>4</sup> In order to plot the interaction, we conducted moderated regressions using the vignettes as a control variable.

### TABLE 5 | Means, standard deviations and, intercorrelations (Study 2).


\*p < 0.05 (2-tailed), \*\*p < 0.01 (2-tailed).

attribution of intentionality, the relationship between abusive supervision and turnover intention was less negative than for low intentionality, but overall, turnover intention was on a higher level. The interaction thus shows that at high abusive supervision, turnover intention is high, independent of intentionality and that intentionality is relevant at lower levels of abusive supervision (note that the data used here only reflects the two abusive vignettes). Though not plotted here, the pattern was the same for withdrawal. For high attribution of intentionality, the relationship between abusive supervision and supervisor commitment was near zero, whereas for low attribution of intentionality, the relationship between abusive supervision and supervisor commitment was negative. Again at high levels of abusive supervision, attribution of intentionality makes little difference to the reaction of the participant in terms of supervisor commitment. Whereas at low levels of abusive supervision, attribution was relevant for supervisor commitment, this was not the case at high levels of abusive supervision. Looking at prohibitive voice as an outcome, the relationship between abusive supervision and prohibitive voice changed from positive (for high attribution of intentionality) to negative (for low attribution of intentionality). Thus, voice reduced under low intentionality with abusive supervision and increased under high intentionality with abusive supervision. Though not plotted here, the pattern was the same for promotive voice. This lends support to H3.

### Discussion Study 2

In Study 2, we replicated (most of) the results of Study 1 in terms of the mediation effect of the perception of abusive supervision on the relationship between abusive leader behaviors and reactions, again in an experimental design with improved measures for reactions, including a differentiation between prohibitive and promotive voice. We also added attribution of intentionality as a moderator of this mediation relationship and found that attribution of intentionality is less relevant for turnover and supervisor commitment when high abusive supervision is perceived than when abuse is perceived as lower. In other words, with high abusive supervision, there is a high level of turnover intention and a low level of commitment while attribution of intention does not make a difference. In terms of voice, the relationship between abusive supervision and voice turned positive for high attribution of intentionality. We wonder if this is a limitation of our design, where participants were asked to indicate how they would react to an abusive supervisor. Participants might think that they would use more proactive behaviors to counteract abusive (as those behaviors are not necessarily directed directly to the supervisor). However, this result is unlikely to replicate in our field Study 3 where we ask for actual reactions. Here, voice might go down when intentionality is attributed to abusive behavior due to fear of retaliation or being subjected to more abuse when the follower makes him/herself more visible through the use of voice.

So far, all our studies used experimental designs to control for leader behavior and allow for a better control of perception effects as well as drawing conclusions about causality of the effects. In Study 3, we examined if the results for attribution replicate in a field study. We also assume here that rater characteristics will have a stronger impact on the perception of abusive supervision as actual behavior is more ambiguous in the field and thus we expect more rater bias than in our experimental studies.

# STUDY 3

### Overview

In Study 3, we wanted to examine our model using the perception of abusive supervision of actual leaders in a cross-sectional field study. While this design does not allow drawing conclusions about causality, it is aimed to add more external validity to our experimental results. At the same time, we expect that actual leader behavior is more ambiguous and that, consequently, we will find a stronger effect of the control variables (follower characteristics; specifically, negative affectivity, hostile attribution style, trait anxiety, and irritation) on the perception of abusive supervision in the field study, comparable to previous studies (see Martinko et al., 2013, for an overview). As mentioned in the introduction, we also assume that we will find overall low levels of perceived abusive supervision as especially strong abuse is seldom. Hence, the field study complements our experiments by adding ecological validity to our results.

In order to further explore the moderating effect of attribution, we added a second measurement of attribution, namely attribution to the supervisor's control in addition to our intentionality assessment. Similarly to attribution to intentionality, we assumed that attribution to the supervisor's control would increase the strength of the relationship between perceptions of abusive supervision and reactions.

# Design and Procedure

We again used an online provider (respondi) to collect the sample. However, this time, we did not ask the participants to read vignettes and evaluate a described leader but to evaluate their own leader. First, we asked the participants to indicate a few stable characteristics, namely, trait negative affectivity, hostile attribution style, irritation, and anxiety. Second, participants rated their leader's abusive supervision and their attributions for this behavior. They also indicated their intention to quit, commitment to their supervisor, voice behavior, and withdrawal from the relationship with their supervisor.

# Participants

We used the same inclusion criteria as in Study 2 to keep the samples comparable. Specifically, participants were employed, between 18 and 65 years old, and had 3 or more months of work

experience. The sample size after applying quality checks (e.g., lack of standard deviations, N = 56 did not pass our quality checks) was N = 313. The majority (55.2%) was between 35 and 54 years old. Of the participants, 182 were male (58.1%) and 131 were female (41.9%). The majority of the participants had 10 or more years of work experience (75.4%). About a third (31.6%) had A-levels, 28.4% had a graduate degree, 25.9% GCSEs and 14.1% had a postgraduate degree. The participants came from various industries such as health care, government, or retail. On average, they worked with their supervisor since 4 years (SD = 4.54).

### Measures

Unless otherwise stated, the scale ranges are the same as in Study 2. The instruments for Negative affectivity (α = 0.92), Hostile attribution style (α = 0.80), Trait anxiety (α = 0.88) and Irritation (1 = strongly disagree to 5 = strongly agree; α = 0.89) were the same as in Study 2. We also used the same instrument for Perception of abusive supervision (α = 0.96). However, the items now referred to the participants' actual leader.

We used two measurements of Attribution. First, we used the one item measurement used in Study 2 referring to intentionality but now relating to their actual supervisor. Second, we adapted a three-item instrument by McAuley et al. (1992) to assess in how the participants attributed to their supervisor on a semantic differential ranging from 1 to 9. A sample questions reads "Is it a behavior. . . Over which he/she has control / over which he/she has no control." The reliability was α = 0.92.

Reactions to abusive supervision were assessed with the same instruments as in Study 2 but now relating to their actual experience: Exit/Turnover intention (α = 0.96), voice (overall α = 0.92, promotive α = 0.96 and prohibitive α = 0.88), loyalty/supervisor commitment (α = 0.93), and neglect/withdrawal (α = 0.94).

# Preliminary Results: Comparison Between Studies

First, we looked at the means and standard deviations of Study 1, 2, and 3 to see if our studies are comparable in terms of sample characteristics relevant to this study. Negative affectivity, hostile attribution style, anxiety (Study 2 and 3), and irritation were comparable between Study 1, 2, and Study 3, indicating that the samples are equivalent in terms of participants' characteristics. However, perceived abusive supervision was clearly lower in Study 3 (M = 1.62, SD = 0.85) than in Study 1 (M = 2.82, SD = 1.11) and Study 2 (M = 2.60, SD = 1.14), reflecting that in our experimental studies, half of our sample was confronted with abusive supervision whereas the phenomenon is rare in the field study, comparable to previous studies. We also found higher levels of negative reactions in Study 2 compared to Study 3, reflecting the manipulation of abusive supervision. It is also noticeable that stable characteristics (apart from hostile attribution style) and irritation were more strongly related to the perception of abusive supervision in the field Study 3 than in Study 1 and 2, reflecting the ambiguity of actual leader-follower interactions.

reactions (including negative affectivity as control, only abusive vignettes) Study 2. †p < 0.10 (2-tailed), \*p < 0.05 (2-tailed), \*\*p < 0.01 (2-tailed).

All our follower characteristics, that is, irritation (r= 0.42), anxiety (r = 0.28), hostile attribution style (r = 0.30), and negative affectivity (r = 0.51) were positively and significantly related to perception of abusive supervision perceptions (see **Table 6**). However, in a multiple regression analyses, only irritation, anxiety, and negative affectivity remained significant predictors of abusive supervision perceptions, and were thus taken forward as control variables.

### Results

Perceived abusive supervision was positively related to attribution to the supervisor's control (r= 0.22) and to intentionality (r = −0.33). Perceptions of abusive supervision were related to outcomes as expected: Turnover intention (r = 0.40), supervisor commitment (r = −0.44), and withdrawal (r= 0.65) but not to prohibitive voice (r = 0.08) or promotive voice (r = −0.10), lending partial support to H1.

Before conducting moderated regression analyses, we centered perceived abusive supervision and the moderators (intentionality, and attribution to the supervisor) and then calculated the interaction terms. As depicted in **Table 7**, none of the interactions between abusive supervision and attribution of intentionality became significant, thus H3 was not supported for attribution to intentionality.

The results for the moderation effect of attribution to the supervisor's control on the relationship between perceived abusive supervision and reactions showed some significant results: For both aspects of voice as well as for supervisor commitment, the interactions became significant (see **Table 8**).

We plotted the interactions to further examine the moderation effects. **Figures 8**–**10** depict the interaction effects. The relationship between perceived abusive supervision and supervisor commitment was negative for both high and low attribution pf control to the supervisor but the relationship was stronger for high attribution, indicating that attribution makes the negative effect of perceived abusive supervision on supervisor commitment stronger.

The relationship between perceived abusive supervision and prohibitive voice was almost zero for high attribution but positive for low attribution, meaning that participants perceiving high abusive supervision but who do not attribute this behavior to their supervisor feel less inhibited to speak up. A similar pattern emerges for promotive voice: Here the relationship between perceived abusive supervision and promotive voice was also positive for low attribution but it was negative for high attribution, indicating that participants perceiving high abusive supervision and hold their supervisor responsible for his/her behavior, are less likely to be proactive in their voice behavior. Thus, the results lend support to H3 for attribution to the supervisor's control.

### GENERAL DISCUSSION

The aim of the study was to deepen our understanding of destructive leadership, especially with regard to how followers react to abusive supervisor behaviors and which role perceptions and attributions play in this process. Using an experimental approach, we examined in how far followers react differently to different levels of abusive leadership and how this relationship is mediated by perceptions of abusive leadership, controlling for individual variables which have been previously identified as causing rater biases (e.g., Hansbrough et al., 2015; Brees et al., 2016). It was expected that the effects of differences in behavior were stronger in the experimental setting whereas subjective biases should be stronger in the field context, where behavior is less clear and more ambiguous. Experimental vignette designs complement field research as they enhance experimental realism and also allow researchers to manipulate and control independent variables, thereby simultaneously enhancing both internal and external validity (Aguinis and Bradley, 2014). Moreover, we wanted to clarify if perceivers are able to distinguish between different levels of leadership behavior in order to provide evidence for the validity of different concepts of negative leadership behavior (Schilling, 2009; Schyns and Schilling, 2013).

We report results of overall four studies: one pre-study, two experimental studies, and one field study. Our first three studies used vignettes of leader behavior. In all those studies, we found


\*p < 0.05 (2-tailed), \*\*p < 0.01 (2-tailed).



\*p < 0.05 (2-tailed), \*\*p < 0.01 (2-tailed).

the expected differences in follower perceptions (though not all significant) between different levels of abusive leader behavior from not at all abusive (constructive) to strong abusive. Hence, we can conclude that actual leader behavior is a strong predictor of perceived abusive supervision. However, in the pre-study and in Study 2 those differences were not significant between laissezfaire and mild abuse. There is a discussion in the literature that laissez-faire leadership is perceived as abusive as abusive supervision (Skogstad et al., 2014). In our study, we directly compared the perceptions of laissez-faire to different levels of abusive supervision, and our results indicate that, when leader behavior is systematically varied, the perceptions of abusive supervision are comparable for laissez-faire and mild abusive supervision, also in terms of liking and generalized leadership impressions. Hence, our results indicate that laissez-faire is comparable to mild abusive behavior but that strong abusive behavior is worse in terms of leadership perceptions.

In Study 1, we included reactions toward abusive supervisor behavior. Here we found the expected differences depending on the described leader behavior. Constructive leadership always elicited the most positive reaction; laissez-faire and mild abuse were comparable in terms of reactions. For quite, anger, and accept, strong abuse also did not differ from laissez-faire and mild abuse. All the negative forms of leadership seem unacceptable to our participants. As expected, we found that perceptions of abusive supervision are related to those reactions. However our mediation analysis showed that perception does not capture the entire effect of behavior and some follower characteristics (irritation, hostile attribution style, and negative affectivity) are related to followers' reactions. This also means that follower reactions cannot be simply reduced to sensitivities of the followers. Instead leader behavior is a crucial factor in reactions toward negative leader behavior. The measurements for reactions were improved in Study 2. Here, we found again that most reactions differed as expected depending on the described leader behavior (apart from the effect for prohibitive voice). However, looking at the mean differences for the vignettes, again, laissezfaire did not differ from mild abuse. Overall, it seems that laissezfaire and mild abuse are perceived similarly and provoke similar reactions. This is interesting and adds to the discussion of how abusive laissez-faire leadership is perceived (Schyns and Schilling, 2013; Skogstad et al., 2014). Schyns and Schilling (2013) argue that there is a clear qualitative difference between non-leadership and active supervisor hostility. However, it seems that laissezfaire leadership in terms of the perception and reactions of followers is more negative than this conceptual distinction may lead us to expect.

Our results may point to an explanation for the result that laissez-faire and abusive supervision show similar relationships with important variables in the workplace (cp. Skogstad et al., 2014). It seems safe to assume—like in our own field study that abusive supervision in real work settings will mainly come in mild forms which seems to provoke reactions and consequences which are similar to laissez-faire. Strong abusive supervision (as presented in one of our vignettes) is likely a rather seldom phenomenon so that field studies will find it difficult to capture its consequences.

The correlations between perceptions of abusive supervision and reactions were all significant, again apart from prohibitive voice. The latter result was similar in Study 3 (field study), where perception of abusive supervision was related to reactions apart from the two aspects of voice. Overall, our results replicate previous studies showing that abusive supervision is related to negative outcomes such a lower supervisor commitment and higher turnover/withdrawal from the supervisor (Martinko et al., 2013; Schyns and Schilling, 2013; Zhang and Liao, 2015). Even



\*p < 0.05 (2-tailed), \*\*p < 0.01 (2-tailed).

more, in our studies, due to the experimental designs, we could show that this effect is based on actual leader behavior.

As mentioned before, we examined and confirmed most of the postulated mediating effects of the perception of abusive supervision on the relationship between leader behavior and reactions. In Study 1, we found both direct and indirect effects of leader behavior on follower reactions, apart from anger where the indirect effect was not significant. We could replicate the same pattern of results for the mediation analyses in Study 2, apart from the effect for prohibitive voice. We thus found that abusive leader behavior is related to outcomes and that this relationship is partly mediated by perceptions of abusive supervision, emphasizing that both behavior and perception are relevant for reactions to outcomes which indicates evidence for the importance of this distinction (Martinko et al., 2013).

In terms of the reactions, it seems that voice is ambivalent in relation to abusive supervision. It likely contains two aspects that are differently influenced by abusive supervision, namely, complain (e.g., about the abuse) in study 1 and speaking up in study 2. While followers of abusive supervisors complain more, for example to higher authorities, they are less likely to speak up to the supervisor, as they lack of trust and may fear negative

consequences. This also is reflected in our partly different results for promotive and prohibitive voice. In both Study 2 and 3, abusive supervision was not related to prohibitive voice, which is related to pointing out problems in the workplace. As abusive supervision is likely to be part of the problem, different effects could occur: Some employees might speak out against abusive supervision to address the issue while most feel hindered to do so due to fear of retaliation from the supervisor (cp. Schyns and Schilling, 2013). In Study 1, participants said that they would complain more under abusive supervision but in Study 2 and 3, voice was negatively (promotive, Study 2) or unrelated to abusive supervision (Study 3). These mixed results are in line with prior research which found interactions between interactions between person-centered and situational factors play an important role in predicting voice behavior (e.g., LePine and Van Dyne, 1998; Wei et al., 2015). Future research should try to disentangle who the voice is addressed to (the supervisor, peers, or higher up leaders or HR) and what the voice is about (the supervisor vs. other aspects of work) to further clarify these relationships.

We also investigated the role of attributions in this mediating relationship (Study 2). Specifically, we investigated if and to what degree attributions moderated the relationship between the perception of abusive leadership and reactions (Study 2 and 3). In Study 2, we found that attribution to intentionality moderated the relationship between perception of abusive supervision and reactions (Study 2). Here attribution of intentionality reduced the relationship between perception of abusive supervision and reactions, in so far that intention played no role in reactions for strong abusive supervision but it strengthened negative reactions when milder abuse was perceived. This is interesting as it shows that once abusive supervision is strong, it does not seem to matter in how far that behavior is shown intentionally; the reactions remain strong. However, where abuse is milder (and, thus, likely more ambiguous to interpret), the way perceivers attribute the behavior is more relevant in determining their reactions (cp. Liu et al., 2012). This is also potentially relevant in the field context, where behavior is likely to be more ambiguous and very strong abuse is (thankfully) rather a rare phenomenon.

Interestingly, attribution to intentionality was unrelated to abusive supervision in Study 2 but related to the perception of abusive supervision in the field Study 3, likely to reflect ongoing dynamics in the relationships between followers and supervisors in the field. However, intentionality was related to most outcomes in both studies, indicating that attributions can influence outcomes directly (cp. Martinko et al., 2013).

We further investigated the moderating role of attributions in Study 3. In addition to attribution of intentionality, we also included attribution to control of the supervisor. We could not replicate the results of Study 2 regarding the moderating effect of attribution of intentionality in our field Study 3. This might be the case because we did not differentiate between types of intention. So, for example, followers might perceive that a leader is intentionally abusive to achieve a certain goal or intentionally abusive to hurt followers. Tepper (2007) differentiates two types of attributions for abusive supervision, namely, harm of others or achieving an objective (e.g., performance). Liu et al. (2012) argue that attribution to performance promotion motives moderate the relationship between abusive supervision and creativity in a different way from attribution to injury initiating motives. Indeed, both attributions were differently related to the perception of abusive supervision. Liu et al. also found the suggested different moderation effects: While the relationship between abusive supervision and creativity was negative for both high and low performance attribution, it was most negative for low performance attribution. For injury attribution, the effect was higher for high injury attribution. Thus, future research should differentiate between performance and injury attribution to examine in how far they differentially influence on the relationship between abusive supervision and reactions.

While our results for the attribution of intentionality failed to become significant, we found moderating effects of attribution to control of the supervisor in Study 3. For strong attribution to the supervisor the relationship between abusive supervision and reactions was stronger than for weaker attributions. Thus, participants who did not hold their supervisor responsible for his/her abusive behavior showed more voice and were slightly more committed to this supervisor. Here, it would be interesting for future research to differentiate further what other attributions might be relevant. For example, we would assume that the effects of abusive supervision on reactions might be mitigated by attributions toward circumstances (external and unstable attributions: e.g., leader stress, time or task pressure; cp. Weiner, 1986; Martinko et al., 2011). It would also be interesting to examine in how far such excuses would hold up over time. That is, even when abusive supervision is attributed toward circumstances outside the control of the supervisor, followers might eventually still react more negatively as the power of the excuse runs out.

We found some interesting results with respect to the discussion around the effects of follower characteristics on the perception of abusive supervision (e.g., Martinko et al., 2013; Brees et al., 2016). In Study 1, we found no effect of participant characteristics on the perception of abusive supervision, contrary to our expectations based on the literature regarding perception biases. However, this shows clear differences in leader behavior displayed in the descriptions as they leave little room for rater biases. In Study 2, there was a slight effect of negative affectivity effect, but the strongest effect emerged in Study 3, lending support to the assumption that actual leader behavior leaves more room for interpretation based on perceiver characteristics than our described supervisors. This also means that when leaders behave unambiguously, perception effects are likely a lot lower than when behavior is ambiguous. In that sense, our vignettes can be described as strong situations according to Mischel (1977), meaning that fewer effects of personality on perception can be expected.

We also contribute to our understanding of the influence of follower characteristics on the perception of abusive supervision vs. on the behavior of abusive supervisors. For example, Wang et al. (2015) argue that supervisors might treat followers high in neuroticism in an abusive way based on the victim precipitation approach. Brees et al. (2016), in contrast, argue for perception effects. Due to the experimental design of our first two studies, we can show that where behavior is clearly positive or negative, few rater effects occur. Future research needs to further disentangle the relationship between follower characteristics and abusive supervision. Specifically, it would be interesting to study circular effects over time where follower characteristics influence leader behavior which then lead to changes in follower characteristics over the period of abuse.

# LIMITATIONS

Experimental studies are often criticized for their lack of external validity (Aguinis and Bradley, 2014). However, in order to show causal effects, experiments are invaluable. In terms of the moderating effects of attribution, we conducted a field study to examine in how far our results replicate. One strength of our studies reported here is that the differences in the means and standard deviations between Study 2 and Study 3 lend support to the validity of our measurement as well as our manipulation of abusive supervision. However, the correlations were not affected by those mean differences (bottom or ceiling effects). In addition, in all three experimental studies, the mean values for abusive supervisions and differences between the vignettes were comparable, indicating the validity of our manipulations. We also found no differences in terms of the participants' characteristics in the three studies but found the expected differences between the experimental and field studies in terms of perceptions of abusive supervision and reactions. This lends support to our approach of using experimental studies and combining them with a field study.

In our experimental studies, we asked our participants to anticipate how they would react to the leader behavior. Again, this could be criticized as being artificial. However, our results were mainly replicated in the field study, showing that we can draw conclusions from experimental studies to the field. Indeed, one might argue that the fact that our participants showed reactions after a limited exposure to leader behaviors constitutes a more conservative test of reactions toward abusive supervision. The exception here was voice where it seems that participants in the experimental studies might over-estimate their engagement in voice when exposed to abusive supervision. What could be interesting for future research is to assess physiological stress measures after the exposure to abusive supervision in an experiment. This would go beyond anticipated reactions as we investigated them here.

While we pre-tested our vignettes, we only used abusive supervision, liking, and leadership impressions as manipulation checks. Ideally, we would have included a measurement for constructive and, more importantly, laissez-faire leadership. The latter would have been useful to check in more detail in how far laissez-faire and mild abusive differ from each other.

Our studies employed the description of only one situation. This is a clear limitation as results relating to a variety of situations would have lend more confidence to our results. However, our vignette is quite detailed and describes a typical leader follower interaction. Future studies could employ a different situation to add confidence to the generalizability of our results across situations. However, they have to take into account either the issue of a larger sample size (between participant design) or risk participant fatigue (within participant design).

Future research might also consider using videos instead of written vignettes. While they are easier to present, written vignettes are likely to be seen as less realistic than videos. However, when constructing videos, researchers have to be careful not to vary appearance as faces of leaders already lead to leadership impressions (Antonakis and Dalgas, 2009; Trichas et al., 2017). They also have to carefully manipulate tone of voice or facial expressions to best express the different leadership styles.

### PRACTICAL IMPLICATIONS

In terms of practical implications, we can conclude that leader behavior is important and that negative leader behavior needs to be addressed by organizations. At the same time, leaders should be made aware that in practice, their behavior might come across as ambivalent and could be subject to rater effects and thereby lead to negative effects in their followers. For example, we found

# REFERENCES


that laissez-faire is perceived as similar and reacted to in a similar as mild abusive supervision. This points to the usefulness of integrating negative behaviors, such as abusive supervision, into 360 degree feedback and to carefully disentangle interactions both for the benefit of the leader and the follower.

# CONCLUSION

The aim of this study was to examine the differentiated effects of abusive leader behavior on follower reaction as mediated by the perception of abusive supervision. According to our results both are important and, similar to constructive leadership, there are effects of raters on the perception of abusive supervision, specifically in the field where behavior differences may be less clear than in an experiment. Perception of abusive supervision mediates the relationship between leader behavior and reactions, lending support to the relevance of perceptions in leadership research and the necessity to take into account perception effects when assessing leadership and its outcomes. We also found that attributions can influence the strength of reactions, lending support to the notion that some people suffer more under abusive supervisors than others.

# AUTHOR CONTRIBUTIONS

All authors contributed to the design of the studies. BS did the majority of the writing up and the analyses, which JF supported and checked. JS was involved in all the writing and thinking process.

# FUNDING

Funded in part by a seed money award from Neoma Business School as well as Durham University Business School.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg. 2018.01309/full#supplementary-material

from psycholexical studies in seven languages. J. Pers. Soc. Psychol. 86, 356–366. doi: 10.1037/0022-3514.86.2.356


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Schyns, Felfe and Schilling. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Narcissistic Leaders and Their Victims: Followers Low on Self-Esteem and Low on Core Self-Evaluations Suffer Most

Barbara Nevicka<sup>1</sup> \*, Annebel H. B. De Hoogh<sup>2</sup> , Deanne N. Den Hartog<sup>2</sup> and Frank D. Belschak<sup>2</sup>

<sup>1</sup> Department of Psychology, University of Amsterdam, Amsterdam, Netherlands, <sup>2</sup> Amsterdam Business School, University of Amsterdam, Amsterdam, Netherlands

Edited by:

Barbara Wisse, University of Groningen, Netherlands

### Reviewed by:

Laura Venz, Universität Mannheim, Germany Stacey Sanders, University of Groningen, Netherlands

> \*Correspondence: Barbara Nevicka b.nevicka@uva.nl

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 29 September 2017 Accepted: 14 March 2018 Published: 29 March 2018

### Citation:

Nevicka B, De Hoogh AHB, Den Hartog DN and Belschak FD (2018) Narcissistic Leaders and Their Victims: Followers Low on Self-Esteem and Low on Core Self-Evaluations Suffer Most. Front. Psychol. 9:422. doi: 10.3389/fpsyg.2018.00422 Narcissistic leaders are self-absorbed and hold beliefs of entitlement and superiority. Their aggressive tendencies in the face of criticism and inclinations to validate their selfworth by derogating others may lead others to perceive them as being abusive. Here, we test the relationship between leader narcissism and followers' perceptions of abusive supervision. Drawing upon research related to the behavioral plasticity hypothesis, we propose that followers with low self-esteem will perceive narcissistic leaders as more abusive than those with high self-esteem. Followers low on self-esteem are more insecure, more in need of approval from their supervisor and are more likely to interpret the haughty, derogatory attitude of narcissistic leaders as abusive. Such followers also make for 'easier targets' and thus may actually suffer more abusive behavior from their narcissistic leaders. In a first multi-source study of 85 leaders and 128 followers, we found support for the moderating role of follower self-esteem in the relationship between leader narcissism and perceived abusive supervision: Narcissistic leaders were rated as more abusive by followers who were low on self-esteem, but not those higher on self-esteem. In a second multi-source field study among 177 leader-follower dyads, we tested a moderated mediation model and showed that this finding also holds for the broader concept of follower core self-evaluations as a moderator. Abusive supervision, in turn, was related to lower follower performance and followers experiencing more burnout symptoms. Thus, followers low on self-esteem or low on core self-evaluations seem to suffer most from narcissistic leaders as they perceive them to be abusive and, in turn, these followers show reduced performance and more burnout symptoms when working for such leaders. This research thus identifies an important moderator that might help reconcile previous inconsistent findings regarding perceptions of narcissistic leaders.

Keywords: leader narcissism, abusive supervision, follower self-esteem, follower core self-evaluations, performance, exhaustion

# INTRODUCTION

fpsyg-09-00422 March 27, 2018 Time: 17:35 # 2

Narcissism, a personality trait characterized by grandiose and overly positive self-views, is not only rising in Western individualistic countries (Twenge et al., 2008; Twenge and Foster, 2010), but also appears to be societally valued as evidenced by narcissists' emergence as leaders (Brunell et al., 2008; Nevicka et al., 2011a; Grijalva et al., 2015a). The reason for this is that narcissistic individuals possess many characteristics that people associate with a prototypical leader (e.g., confidence, extraversion, dominance; Smith and Foti, 1998; Judge et al., 2002; Kellett et al., 2006; Paunonen et al., 2006). Furthermore, narcissists' charm, humor, enthusiasm and often attractive charismatic vision (Galvin et al., 2010; Goncalo et al., 2010) engender positive first impressions (Back et al., 2010), which can facilitate successful appraisal in selection contexts and help narcissists rise to power.

The problem with narcissists' rise to power, however, is that narcissists also have many negative interpersonal characteristics, such as a lack of empathy, exploitativeness, a sense of entitlement, antagonism and egocentrism (Sedikides and Campbell, 2017), which could lead them to abuse their power and adversely impact those they lead. For instance, narcissists are known to aggress against and derogate others when their ego is threatened (Bushman and Baumeister, 1998), and even sometimes aggress without provocation (Martinez et al., 2008; Lobbestael et al., 2014; Park and Colvin, 2015). Furthermore, they externalize blame while accepting credit for others' success (Stucke, 2003), they are exceedingly critical of others and expect perfection (Stoeber et al., 2015), and they show unethical behavior (Soyer et al., 1999; Penney and Spector, 2002; Watts et al., 2013). Scholars have theorized that narcissists' tendencies to act in a self-interested and dominant manner might predispose them to engage in abusive or destructive behavior as leaders (Tepper, 2007; Krasikova et al., 2013; Martinko et al., 2013). Interestingly, a recent study in an organizational setting found no direct relationship between leader narcissism and abusive supervision (Wisse and Sleebos, 2016), defined as sustained display of hostile verbal and non-verbal behaviors, excluding physical contact (Tepper, 2000). We propose that the extent to which narcissistic leaders are perceived as abusive depends on followers' personality. What is interpreted as abusive behavior often substantially varies between individual perceivers (Tepper, 2000). Thus, we propose that while some followers may perceive narcissistic individuals in a leadership role as abusive, others may not.

In line with this proposition of differential perceptions of narcissistic leaders by different followers, findings regarding followers' general perceptions of narcissistic leaders are mixed. Some studies show that followers had favorable perceptions of narcissistic leaders (Judge et al., 2006; Nevicka et al., 2011b; Owens et al., 2015), while others show followers having negative perceptions (Judge et al., 2006; Martin et al., 2016) and a recent meta-analysis showed no linear relationship between narcissism of leaders and perceptions of leader effectiveness (Grijalva et al., 2015a). These inconsistent findings suggest that moderators may play an important role in followers' perceptions of narcissistic leaders. For instance, prior research shows that perceptions of narcissistic individuals in peer groups vary according to the length of acquaintance because the passage of time exposes narcissists' negative characteristics. Thus, shortterm acquaintances tend to evaluate narcissistic peers more positively, whereas over time with longer acquaintance these positive perceptions diminish (Carlson et al., 2011; Leckelt et al., 2015; Ong et al., 2016).

In a similar vein, followers with certain personality traits might be more sensitive to the toxic characteristics of narcissistic leaders, while others may be better able to cope with such leaders. Therefore, the current research set out to answer the important question of which followers would be most likely negatively impacted by narcissistic leaders? Specifically, we expect that narcissistic leaders will be perceived as abusive especially by followers with low self-esteem. By focusing on followers' self-esteem as an important moderator, we thus help reconcile inconsistent findings regarding followers' perceptions of narcissistic leaders.

# Leader Narcissism and Follower Self-Esteem

Self-esteem − the appraisal of a person's self-worth (Leary and Baumeister, 2000) – has been theorized to be a personality trait which increases individuals' susceptibility to leaders' toxicity (Padilla et al., 2007; Thoroughgood et al., 2012). This suggests that self-esteem may moderate how followers perceive destructive leaders such as narcissistic leaders. We propose two main theoretical reasons why followers with low self-esteem (rather than high self-esteem) would perceive leaders as more abusive the more narcissistic they are, namely because of followers': (1) greater sensitivity to narcissistic leaders' negative characteristics and (2) greater likelihood to actually encounter narcissists' abusive behavior.

Firstly, behavioral plasticity hypothesis contends that selfesteem moderates the extent to which individuals react to external cues (Brockner, 1988). Because they are uncertain of the appropriateness of their attitudes and behavior, individuals with low self-esteem are more sensitive and reactive to external social cues. In the organizational context, a leader would constitute an important contextual cue as the leader provides direction, evaluates the employee and has the power to reward or punish. Low self-esteem followers are therefore likely to be more perceptive of external cues such as their leader's traits, than followers high on self-esteem (Elangovan and Xie, 1999; Avey et al., 2011).

In addition, low self-esteem individuals are more likely to interpret leaders' toxic characteristics as stressful and threatening and they would be less able to cope with them (Smith and Petty, 1995). This does not mean that high self-esteem individuals would be completely oblivious to the toxic side of narcissistic individuals. Rather, they would be better equipped to deal with such leaders because of their better coping strategies in general, rely less on their leaders for direction and support, and would generally discern the negative characteristics of narcissistic leaders as less threatening to them (Leary and Baumeister,

2000) and thus as less abusive. For example, a follower low on self-esteem might see a leader taking all the credit for the follower's success as unfair and abusive, while someone high on self-esteem might interpret this as a signal that they did well and expect that the leader will eventually reciprocate and thus might not always interpret this behavior as being abusive.

Secondly, because of their insecurities about their abilities, low self-esteem individuals, also dubbed as 'lost souls' (West and Sweeting, 1997), look toward their leaders for approval and validation and they especially seek charismatic high-power individuals who can help them increase their own self-esteem and offer them direction and clarity (Hayes, 2014; Padilla et al., 2007; Thoroughgood et al., 2012). Narcissistic leaders who tend to exude visionary charisma and come across as confident and dominant (Galvin et al., 2010) would nicely fit that template. This stronger dependence on their (narcissistic) leaders, however, also makes low self-esteem followers more vulnerable to actually becoming victims of abusive behavior. Prior research indeed shows that individuals with low self-esteem are less able to defend themselves against aggression (Matthiesen and Einarsen, 2001) and are more likely to become targets of workplace bullying (e.g., Harvey and Keashly, 2003; Bowling and Beehr, 2006; Aquino and Thau, 2009). They are also more likely to avoid confrontation and to conform to social norms (Leary and Baumeister, 2000), have poorer conflict resolution skills (Zapf, 1999), and are susceptible to manipulation especially from authoritarian figures (Gudjonsson and Sigurdsson, 2003; Aquino and Thau, 2009).

Additionally, individuals with low self-esteem might even accept derogatory or aggressive behavior toward them because of their own low perceptions of their self-worth (Padilla et al., 2007; Thoroughgood et al., 2012) and because negative feedback is more consistent with their cognitive structures and expectations (Shrauger, 1975). For instance, low self-esteem individuals are less likely to retaliate against abusive leaders (Tepper, 2007) than those with high self-esteem. Thus, low self-esteem followers' high need for approval, their tendency to conform to social norms, their dependence on their leader for clarity, direction and validation, their reticence to challenge authority figures and their low self-worth all make followers with low self-esteem "easy targets" for narcissistic leaders' abuse. Abusive leadership often entails displaced aggression especially toward "safe" targets who are unwilling or unable to defend themselves (Tepper, 2007). Given narcissists' proclivity to aggress against innocent others when provoked (Martinez et al., 2008), their tendencies for proactive aggression, which constitutes an instrumental use of aggression to exploit others for personal gain (Lobbestael et al., 2014), and their preference for confident others over less confident individuals (e.g., Burton et al., 2017), narcissistic leaders would be more likely to show negative of hostile behavior toward followers with low rather than high self-esteem. Consequently, since followers with low self-esteem are more likely to be affected by narcissistic leaders' negative characteristics, and also more likely to become chosen as targets of abuse by narcissistic leaders, we expect the following:

Hypothesis 1: Follower self-esteem moderates the relationship between leader narcissism and perceived abusive supervision, such that leader narcissism will be positively associated with perceived abusive supervision for followers with low self-esteem, but not for followers with high self-esteem.

We will test this hypothesis in Study 1, a multi-source empirical field study. In Study 2, we will test the same hypothesis using the broader construct of followers' core self-evaluations, while also examining the consequences of abusive supervision for followers. We will return to this after discussing Study 1 and its results in detail.

The research presented here will make several contributions. Firstly, in focusing on the role of follower personality (i.e., follower self-esteem) in followers' perceptions of narcissistic leaders, it proposes an important moderator to reconcile previously inconsistent findings. Secondly, this research focuses on which followers are especially vulnerable to suffer from the toxic side of narcissistic leaders and who are thus most likely to perceive these leaders as abusive. We thereby further extend the literature on susceptible followers and destructive leaders in general (Padilla et al., 2007; Thoroughgood et al., 2012).

# METHOD – STUDY 1

# Sample and Procedure

We used a multi-source field study to test the proposed research model. The sample consisted of 128 followers matched with 85 leaders who worked in different organizations and across different industries (e.g., hospitality, healthcare, and business). Leaders were first approached through Business School graduate student contacts. If they agreed to participate they were then sent a survey link to complete the survey online. The leaders were asked to nominate up to three followers and to provide their email addresses, after which the followers were then forwarded a separate survey link. Surveys could be completed either in English (74% of respondents) or in Dutch (26% of respondents).

The voluntary nature of participation and confidentially was stressed in the accompanying letter for each respondent. The study was carried out in accordance with the recommendations of the Ethics Review Board of the Faculty of Social and Behavioural Sciences of the University of Amsterdam, who approved the protocol for the study. All subjects gave written informed consent in accordance with the Declaration of Helsinki.

The questionnaires were completed anonymously. A unique code was used to match the surveys. To increase the response rate, participants were sent several reminders and leader-follower pairs were offered a small incentive — three pairs would be randomly selected to win a voucher worth 40 euros. Out of 128 leaders who were sent the survey links, 97 completed the survey (response rate 75.8%). In total leaders nominated 203 followers, out of which 128 completed the survey (response rate 63.1%). Leaders (Mage = 38.38 years, SD = 11.09; 69.4% men) had an average tenure of 5.53 (SD = 1.90) years and had 34.38 (SD = 74.60) followers on average. There were on average

1.51 followers per leader in this sample (observed range 1–3). Followers (Mage = 35.48 years, SD = 12.28; 39.8% men) had an average tenure of 5.79 (SD = 7.70) years and had worked with their leader for 2.53 (SD = 2.83) years.

### Measures

Leaders filled in the Narcissism personality inventory. Followers filled in the self-esteem personality questionnaire and rated the abusive supervision of the leader.

### Leader Narcissism

Leaders filled in the 16-item version of the Narcissistic Personality Inventory (NPI-16; Ames et al., 2006). This measure is based on the original 40-item NPI (Raskin and Hall, 1979, 1981), was shown to be psychometrically sound (Ames et al., 2006) and is frequently used to measure narcissism in normal populations (e.g., Peterson et al., 2012; Owens et al., 2015). The scale has a forced choice format, with example items including "I think I am a special person" (narcissistic option = 1) vs. "I am no better or worse than most people" (non-narcissistic option = 0). Especially when items are dichotomous, coefficient alpha can underestimate the reliability of the scale (Raykov et al., 2010). Following recommendation by Widaman et al. (2011) and in line with prior research using NPI-16 with the forced choice variant (e.g., Orth and Luciano, 2015; Orth et al., 2016), we therefore calculated coefficient omega (McDonald, 1999). Coefficient omega of the scale was 0.64. Removing two items increased the reliability coefficient to 0.67 and we used the remaining 14 items in the analyses. The NPI score was computed as the sum of the items, with a higher score indicating higher narcissism.

### Follower Self-Esteem

Follower self-esteem was measured using the 10-item Rosenberg Scale (Rosenberg, 1965). Example items include: "I feel that I have a number of good qualities" and "I certainly feel useless at times" (reverse item). Responses were given on a four-point Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). Coefficient alpha of the scale was 0.83.

### Abusive Supervision

Abusive supervision was measured using the 5-item shortened version (Mitchell and Ambrose, 2007) of Tepper's (2000) Abusive Supervision measure. Followers indicated their agreement with each item. Examples of items include: "My supervisor ridicules me" and "My supervisor tells me my thoughts and feelings are stupid." Items were rated on a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Coefficient alpha of the scale was 0.91.

### Control Variables

It included followers' tenure with the leader and gender of the leader and the follower. The negative effects of narcissism may increase over time (Paulhus, 1998), men score higher on narcissism than women (Grijalva et al., 2015b) and followers' gender is found to be related to perceived victimization, with females reporting more abuse (Aquino and Bradfield, 2000).

# Results – Study 1

### Confirmatory Factor Analysis

We conducted confirmatory factor analyses to determine whether the data conformed to the assumption that each of the proposed latent variables represents a separate construct. Fitting a measurement model with a large number of indicators (and items) can adversely affect model fit (Hall et al., 1999; Judge et al., 2002). To control for inflated measurement errors caused by multiple items for the latent variable, we divided the items for the personality constructs self-esteem (10) and narcissism (16) into parcels of 3 to 4 items to serve as indicators of the factors using random heterogeneous assignment (Little et al., 2002; Cole et al., 2016). This led to a total of three parcels for CSEs and four parcels for narcissism. The individual scale items were used as indicators of the abusive supervision construct (five items). In addition to the Chi-square statistic, we investigated the Root Mean Square Error of Approximation (RMSEA; acceptable fit: 0.05–0.08, good fit: 0–0.05), the Standardized Root Mean Square Residual (SRMR; acceptable fit: 0.05–0.10, good fit: 0–0.05) and the Comparative Fit Index (CFI; acceptable fit: 0.90–97, good fit: 0.97–1) (see Bentler, 1990; Hu and Bentler, 1999; Schermelleh-Engel et al., 2003; Marsh et al., 2004; Chen et al., 2008).

The CFA supported the proposed 3-factor measurement model, [χ 2 (51, N = 128) = 72.11, p = 0.03; RMSEA = 0.05; SRMR = 0.07; CFI = 0.96]. Two of the possible alternative models, one in which the items of self-esteem and abusive supervision were merged into an overall factor, and one in which all items loaded on 1 factor, did not converge. A final alternative model, in which the items of follower self-esteem and leader narcissism were merged into an overall factor [χ 2 (53, N = 128) = 104.52, p < 0.001, RMSEA = 0.07; SRMR = 0.10; CFI = 0.90; 1χ 2 (2) = 32.41, p < 0.001], exhibited significantly poorer fit.

### Hypothesis Testing

**Table 1** presents means, standard deviations and bivariate correlations of the variables. Given the hierarchical structure of our data, with followers (level 1) nested in leaders (level 2), we tested our hypotheses using a random coefficient model. Leader narcissism and follower self-esteem were grand-mean centered. The total variance explained by the models was calculated using


N = 128 followers (level 1 data) matched with N = 85 leaders (level 2 data). Tenure in years. Men are coded as 1, women are coded as 2. <sup>∗</sup>p < 0.05, ∗∗p < 0.01.



N = 128 followers (level 1 data) matched with N = 85 leaders (level 2 data). <sup>∗</sup>p < 0.05, ∗∗p < 0.01.

the conditional R 2 (Snijders and Bosker, 1994; Nakagawa and Schielzeth, 2013).

To test the hypothesis, the control variables, leader narcissism, follower self-esteem and their interaction were entered into the random coefficient model. The results of this analysis are presented in **Table 2**. The results showed no main effect of leader narcissism on abusive supervision (B = 0.04, t(81.06) = 1.54, p = 0.127, 95%CI[−0.01, 0.10]), but did show a negative relationship between follower self-esteem and abusive supervision (B = −0.57, t(120.80) = −3.42, p = 0.001, r = 0.30, 95%CI[−0.90, −0.24]). As expected, there was a significant interaction found between leader narcissism and follower selfesteem (B = −0.14, t(112.87) = −2.38, p = 0.019, r = 0.22, 95%CI[−0.27, −0.02]), which accounted for 3% of the variance in abusive supervision. Subsequent analyses of simple slopes (Aiken and West, 1991) showed that for followers with low self-esteem (1 SD below the mean) the relationship between leader's narcissism and abusive leadership was positive (B = 0.10, t(116.84) = 2.71, p = 0.008, r = 0.24, 95%CI[0.03, 0.17]). For followers with high self-esteem (1 SD above the mean) this relationship was not significant (B = −0.02, t(108.30) = −0.51, p = 0.608, 95%CI[−0.09, 0.05]). See **Figure 1**. Thus, Hypothesis 1 received support, followers with low self-esteem perceived narcissistic leaders as more abusive, those with high self-esteem did not.

### STUDY 2

In a second multi-source study we aim to provide a conceptual replication of Study 1 and test whether the stronger relationship of leader narcissism with abusive supervision also occurs for followers who are low on the higher order self-esteem related construct of core self-evaluations (CSEs). In this way, we aim to not only show the robustness of our findings in Study 1, but also to broaden the scope of the research to include a more comprehensive conceptualization of who the potentially most vulnerable followers are (e.g., Padilla et al., 2007; Thoroughgood et al., 2012). In addition, we test the relationship of perceived abusive supervision with followers' outcomes in order to examine whether vulnerable followers also suffer more negative consequences under narcissistic leaders.

# Leader Narcissism and Follower Core Self-Evaluations

Core self-evaluations is a more general higher order construct which, in addition to self-esteem, comprises of self-efficacy, locus of control and emotional stability and refers to "basic conclusions or bottom-line evaluations that individuals hold about themselves" (Judge and Bono, 2001, p. 81). Individuals with more positive CSEs like themselves and think of themselves as capable, worthy, and competent in dealing with issues in different contexts (Judge et al., 2003). Conversely, individuals with more negative CSEs dislike themselves and are not confident in their capabilities, competence, or worthiness. Having lower or more negative self-evaluations, similarly as with followers with low self-esteem, makes such followers more susceptible to suffer from abusive or destructive leaders (Luthans et al., 1998; Padilla et al., 2007; Thoroughgood et al., 2012). For instance, having low expectations of one's ability to perform well (i.e., low self-efficacy)

Nevicka et al. Narcissistic Leaders and Their Victims

increases the followers' dependence on their leaders because these individuals are more likely to feel they need the leaders to provide them with clarity and direction (Thoroughgood et al., 2012). Similarly, having a belief that outcomes are the result of external events (i.e., external locus of control) instead of one's own actions makes individuals easier to manipulate, and also makes them more likely to seek out powerful others who can take care of them and to whom they can defer responsibility (Padilla et al., 2007). Thus, we expect that we can extend the construct of vulnerable followers from low self-esteem to include those individuals who have more general negative views regarding not only their selfworth, but also their competencies and feelings of control over outcomes (i.e., those followers with low CSEs). This leads us to the following hypothesis:

Hypothesis 2: Follower CSE moderates the relationship between leader narcissism and abusive supervision, such that leader narcissism will be positively associated with perceived abusive supervision for followers with low CSEs, but not for followers with high CSEs.

# Consequences of Abusive Supervision

Abusive supervision has been shown to have many detrimental consequences for followers, such as psychological distress (e.g., strain, emotional exhaustion, and depression), lower family wellbeing, and higher turnover intentions (Tepper, 2000, 2007; Aryee et al., 2008; Wu and Hu, 2009; Carlson et al., 2012; Schyns and Schilling, 2013). Abusive supervision has also been linked to lower follower job performance, both with respect to reduced core task performance as well as reduced organizational citizenship behavior (Zellars et al., 2002; Aryee et al., 2007; Harris et al., 2007; Tepper, 2007; Xu et al., 2012; Schyns and Schilling, 2013).

In Study 2 we include the consequences of abusive supervision and test whether perceptions of abusive supervision relate to distress and job performance. Specifically, we focus on followers' self-rated emotional exhaustion, as being reflective of their experienced psychological distress, as well as their task performance as rated by their leaders. Given the argumentation presented above we expect that leader narcissism, through greater perceived abusive supervision, will be associated with greater emotional exhaustion and worse performance, especially for followers with low CSEs. Combining the arguments presented above in the development of Hypothesis 2 we thus propose a moderated mediation model and argue that leader narcissism has an indirect negative effect on follower performance and emotional exhaustion, via perceptions of abusive supervision, and that this indirect effect is contingent on followers' CSEs.

Hypothesis 3: Leader narcissism is related to follower task performance via a conditional indirect effect, such that the negative indirect effect via abusive supervision on performance is stronger when follower CSE is low rather than high.

Hypothesis 4: Leader narcissism is related to follower exhaustion via a conditional indirect effect, such that the positive indirect effect on exhaustion via abusive

supervision is stronger when follower CSE is low rather than high.

To sum up, we propose, in replication of the findings of Study 1 that because of their greater reliance on external cues and dependence on narcissistic leaders, followers with low general CSEs will be more likely to perceive narcissistic leaders as abusive. Furthermore, as a consequence, low CSE followers are more likely to suffer negative outcomes in terms of psychological distress as well as lower performance as a result of leader narcissism. **Figure 2** presents the full proposed model.

# Method – Study 2

### Sample and Procedure

We performed a multi-source field study to test the proposed moderated mediation research model. The sample consisted of 176 unique leader-follower dyads working in a wide range of jobs (lawyers, salespersons, account managers) in different organizations (e.g., health care, government, insurance) in the Netherlands. These contacts were approached through Business School graduate student contacts. Survey packets were sent to both the supervisor and the employee and the voluntary nature of participation and confidentially was stressed in the accompanying letter for each respondent. The study was carried out in accordance with the recommendations of the Economics and Business Ethics Committee, University of Amsterdam, who approved the protocol for the study. All subjects gave written informed consent in accordance with the Declaration of Helsinki.

The questionnaires were completed anonymously. Individual surveys could be returned directly to the researchers and a unique code was used to match the surveys. In total, 179 of the contacted supervisors and 186 of the employees returned fully filled out questionnaires, resulting in a response rate of 69% for complete dyads. Most leaders (Mean age 42.35 years, Mean tenure 9.00 years) were male (58.5%), and most followers (Mean age 33.84 years, Mean tenure 5.79 years) were female (56.3%).

### Measures

Unless otherwise indicated, all items were rated on a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Leaders filled in the Narcissism personality inventory and rated followers' task performance. Followers filled in the CSE personality questionnaire, rated the abusive supervision of the leader and indicated their feelings of exhaustion.

### **Leader narcissism**

fpsyg-09-00422 March 27, 2018 Time: 17:35 # 7

Similarly as in Study 1, leaders filled in the 16-item version of the Narcissistic Personality Inventory (NPI-16; Ames et al., 2006). Consistent with recent research that suggests that a Likert response format to the NPI results in stronger reliabilities (Miller et al., 2017) and our experiences in Study 1, in Study 2 we replaced the forced-choice response by a seven point Likert format (cf. Penney and Spector, 2002; Moon et al., 2016), with 1 = strongly disagree through 7 = strongly agree. High NPI scores indicate higher levels of narcissism. A sample item of a narcissistic response is "I am apt to show off if I get a chance." Coefficient alpha of the scale was 0.88.

### **Follower performance**

Leaders also provided ratings for the focal follower's performance using four items from Pearce and Porter (1986, see also Ashford and Black, 1996). Leaders were asked to report how the follower was rated relative to others on a percentage basis at their last actual performance evaluation (e.g., 60th percentile, 70th percentile). A sample item is "The achievement of work goals." Coefficient alpha of the scale was 0.85.

### **Follower core self-evaluations (CSEs)**

We measured followers' CSEs with the 12- item scale developed and validated by Judge et al. (2003). The scale measures positive feelings about the self in terms of self-esteem, generalized selfefficacy, emotional stability, and locus of control. Examples of items are: "Overall, I am satisfied with myself " and "I am capable of coping with most of my problems." The coefficient alpha in this study was 0.78.

### **Abusive supervision**

Abusive supervision was measured using the 5-item shortened version (Mitchell and Ambrose, 2007) of Tepper's (2000). Abusive Supervision measure. Followers indicated their agreement with each item. Examples of items are: "ridicules me" and "tells me my thoughts and feelings are stupid." Coefficient alpha of the scale was 0.92.

### **Follower exhaustion**

Followers' emotional exhaustion was assessed with the Dutch version (Schaufeli and van Dierendonck, 2000) of the Exhaustion scale of the Maslach Burnout Inventory–General Survey (Schaufeli et al., 1996). A sample item is "I feel mentally exhausted by my work" Coefficient alpha of the scale was 0.84.

### **Control variables**

Control variables were the same as in Study 1, namely followers' tenure with the leader and gender of the leader and the follower.

### Results – Study 2

### Confirmatory Factor Analysis

We again conducted confirmatory factor analyses to determine whether the data conformed to the assumption that each of the proposed latent variables represents a separate construct. To control for inflated measurement errors caused by multiple items for the latent variable, we divided the items for the personality constructs CSEs (12) and narcissism (16) into parcels of four items to serve as indicators of the factors using random heterogeneous assignment (Little et al., 2002; Cole et al., 2016). This lead to a total of three parcels for CSEs and four parcels for narcissism. The individual scale items were used as indicators of the abusive supervision (five items), performance (four items), and exhaustion factors (five items).

The CFA supported the proposed 5-factor measurement model, [χ 2 (179, N = 176) = 349.39, p < 0.001; RMSEA = 0.07; SRMR = 0.07; CFI = 0.92]. Two alternative models, one in which the items of follower performance and exhaustion were merged into an overall factor (χ 2 (183, N = 176) = 714.79, p < 0.001, RMSEA = 0.13; SRMR = 0.16; CFI = 0.76; 1χ 2 (4) = 365.4, p < 0.001], and one in which the items of follower exhaustion and abusive supervision were merged into an overall factor [χ 2 (183, N = 176) = 656.88, p < 0.001, RMSEA = 0.12; SRMR = 0.132; CFI = 0.79; 1χ 2 (4) = 307.49, p < 0.001] exhibited significantly poorer fit. We also compared the proposed 5-factor measurement model with a two-factor model, which had the items of leader narcissism and follower performance (all rated by the leader) loading on the same factor and the items rated by the follower (CSE, abusive supervision and exhaustion) loading on a separate factor. Again, the 5-factor measurement model showed a significantly better fit over the alternative model [χ 2 (188, N = 176) = 1290.89, p < 0.001, RMSEA = 0.18; SRMR = 0.18; CFI = 0.51; 1χ 2 (9) = 941.50, p < 0.001).

### Hypothesis Testing

**Table 3** presents means, standard deviations and bivariate correlations of the variables. To test the hypotheses relating to our moderated mediation model, we follow the procedure outlined by Preacher et al. (2007). Specifically, we use the MODMED macro (Model 7, Preacher and Hayes, 2004), which provides results relevant for our hypotheses in three steps. Leader narcissism and follower CSE were centered at the mean for all analyses. Before employing the MODMED macro to test our hypotheses, we ran a regression analysis including the controls and leader narcissism and follower CSE in order to test for main effects (see **Table 4**, Model 1). The results showed no significant main effect of leader narcissism on abusive supervision [B = 0.18, t(170) = 1.88, p = 0.062, 95%CI[−0.01,0.36]), but did show a negative relationship between follower CSEs and abusive supervision (B = −0.25, t(170) = −2.31, p = 0.022, r = 0.17, 95%CI[−0.47, −0.04]).

To test Hypothesis 2, the first step of the MODMED analysis examines the effect of the interaction between leader narcissism and follower CSEs on abusive supervision. Results are presented in **Table 4** (Model 2) and reveal a significant interaction between leader narcissism and CSEs of the follower (B = −0.33, t(169) = −2.59, p = 0.011, r = 0.20, 95%CI[−0.58, −0.08]) that accounts for 3% of the variance in abusive supervision. We assessed the nature of this significant interaction by plotting values representing plus and minus 1 standard deviation from the means for leader narcissism and follower CSEs. As shown in **Figure 3** and supported by a simple slopes test (Aiken and

### TABLE 3 | Means, standard deviations, correlations (Study 2).

fpsyg-09-00422 March 27, 2018 Time: 17:35 # 8


N = 176 dyads. Tenure in years. Men are coded as 1, women are coded as 2. CSE, core self-evaluation. <sup>∗</sup>p < 0.05, ∗∗p < 0.01.

TABLE 4 | Estimated coefficients of main effects and moderation on abusive supervision (Study 2).


N = 176 dyads. CSE, core self-evaluation. <sup>∗</sup>p < 0.05, ∗∗p < 0.01.

West, 1991), leader narcissism was positively related to abusive supervision when follower CSEs are low (1 SD below the mean, B = 0.45, t(169) = 3.21, p = 0.002, r = 0.24, 95%CI[0.17, 0.72]) and this relationship weakened and became non-significant for followers high on CSEs (1 SD above the mean, B = −0.05, t(169) = −0.37, p = 0.713, 95%CI[−0.29, 0.20]), supporting Hypothesis 2.

To test the moderated mediation model as formalized in Hypothesis 3 and 4, the second step of the MODMED procedure (**Table 5**) examines the impact of abusive supervision on follower task performance and exhaustion, while controlling for leader narcissism. As expected, abusive supervision was negatively related to follower task performance (B = −0.30, t(170) = −3.72, p < 0.001, r = 0.27, 95%CI[−0.46, −0.14]) and positively

related to exhaustion (B = 0.28, t(170) = 3.58, p < 0.001, r = 0.26, 95%CI[0.13, 0.43]). The third step of the MODMED procedure examines the significance of the conditional indirect effect of leader narcissism on task performance and exhaustion through abusive supervision as a function of follower CSEs. The proposed model receives support if the conditional indirect effect of leader narcissism on task performance and exhaustion, via abusive supervision differs in strength across low and high levels of follower CSEs. We indeed found such support as the index of moderated mediation is significant (Hayes, 2014), meaning that the indirect relationship of leader narcissism with task performance and exhaustion through abusive supervision was found to be a function of follower CSEs (performance: Index = 0.10; Bias and accelerated 95% CI[0.02, 0.20]; exhaustion: Index = −0.09; Bias and accelerated 95% CI[−0.22, −0.01]). Specifically, there was a negative effect of leader narcissism on follower task performance (B = −0.13; Bias and accelerated 95% CI[−0.28, −0.03]) and a positive effect on follower exhaustion (B = 0.12; Bias and accelerated 95% CI[0.03, 0.28]) via abusive supervision when follower CSEs were low, and no significant effect of leader narcissism on follower task performance (B = 0.01; Bias and accelerated 95% CI[−0.06, 0.07]) and exhaustion



N = 176 dyads. <sup>∗</sup>p < 0.05, ∗∗p < 0.01.

(B = −0.01; Bias and accelerated 95% CI[−0.09, 0.04]) via abusive supervision when follower CSEs were high (see also **Table 6**).

Thus, as predicted, when follower CSEs are low, leader narcissism is positively related to perceived abusive supervision, and abusive supervision in turn is negatively related to follower task performance and positively to follower exhaustion. When follower CSEs are high, the positive relationship with abusive supervision becomes insignificant and there is no longer an indirect effect through abusive supervision on task performance and exhaustion for leader narcissism.

### DISCUSSION

By focusing on follower self-esteem and follower CSEs, we sought to reconcile the inconsistent findings regarding followers' perceptions of narcissistic leaders and at the same time identify followers who are more or less vulnerable to narcissistic leaders. Despite the fact that narcissistic leaders have many negative characteristics that may predispose them to being abusive toward their followers (e.g., lack of empathy, sense of entitlement, exploitativeness, and aggressive tendencies), using two multisource field studies we consistently found that narcissistic leaders were only perceived as abusive by followers with low selfesteem (Study 1), and followers who were lower on the higher order construct of CSEs (Study 2). Moreover, when these vulnerable followers perceived more abusive leader behavior when working under leaders high on the trait of narcissism, they also showed poorer functioning at work. They reported having higher psychological distress, as reflected in their greater emotional exhaustion, and they were rated by their leaders as having lower task performance (Study 2). Followers with high self-esteem or high CSEs, seemed to be less negatively affected by narcissistic leaders. They did not perceive narcissistic leaders as more abusive, nor did they, as a result of this, show worse functioning at work.

Our research extends prior work in several ways. Firstly, we show that follower personality plays a critical role in determining how followers perceive and experience narcissistic leaders. This provides one explanation as to why prior research has tended to find inconsistencies when looking at followers' evaluations of their narcissistic leaders, with followers sometimes perceiving narcissistic leaders positively or neutrally (Judge et al., 2006; Nevicka et al., 2011b; Owens et al., 2015) and sometimes negatively (Judge et al., 2006; Martin et al., 2016). Additionally, we contribute to literature on leader narcissism which has sought to ascertain what kind of impact narcissistic leaders have on those that they lead (Campbell and Campbell, 2009; Judge et al., 2009; Sedikides and Campbell, 2017), particularly given that they have a paradoxical mixture of positive and negative characteristics. By examining followers' perceptions of abusive supervision, we show that whether or not narcissistic leaders affect their followers negatively depends at least in part on followers' personality traits. In focusing on followers' self-esteem and their CSEs we show that people's fundamental appraisals regarding their own selfworth, competence, capabilities and the extent to which they feel in control of their lives (Chang et al., 2012), influence whether they are affected by the toxic side of narcissistic leaders. Those low on self-esteem and CSEs seem to be more vulnerable likely in both needing more direction, while also perceiving narcissistic leaders as more threatening. Our findings also suggest that individuals with higher self-esteem and high CSEs are better able to cope with the toxic side of narcissistic leaders and perceive them as less threatening, than those low in selfesteem and CSEs. As such, a person's positive self-appraisals may provide them with a buffer in dealing with narcissists' negative side.

Secondly, our findings further inform research on susceptible followers and the initiation and persistence of destructive leadership styles in organizations (Padilla et al., 2007; Thoroughgood et al., 2012). We show that certain personality traits make followers particularly vulnerable to perceiving and/or encountering leader abuse when working with destructive leaders such as narcissistic leaders. We not only demonstrate that followers with low self-esteem and more negative CSEs perceive more abusive behavior when working under narcissistic leaders, but also that as a result of this, narcissistic leaders have significant negative ramifications on such followers' daily functioning at work, both in terms of their psychological distress as well as their work performance.

Finally, our research can help inform literature on abusive supervision and workplace victimization in general (Tepper, 2007; Martinko et al., 2013) by identifying how dispositional leader-level and follower-level characteristics interact to influence followers' experience of abusive supervisory behavior. For instance, prior research on abusive supervision found that leaders with lower emotional intelligence (Xiaqi et al., 2012), as well as higher Machiavellianism and higher psychopathy (Kiazad et al., 2010; Wisse and Sleebos, 2016) were perceived as

TABLE 6 | Bootstrapping results for test of conditional indirect effects on follower task performance and exhaustion at specific values of the moderator (CSE): Mean and ±+1 standard deviation (Study 2).


Results are based on 5000 bootstrap samples. Conditional indirect effects are two-tailed. CI, confidence interval; CSE, core self-evaluation. <sup>∗</sup>p < 0.05.

more abusive. The results of our studies show that narcissism is an important addition to the list of characteristics which may make leaders more predisposed toward abusive behaviors, however, in the case of narcissistic leaders this only holds provided that these leaders are coupled with followers who see themselves as low in self-worth and competence. Thus, our findings suggest that the negative impact of narcissistic leaders is only manifested when there are vulnerable 'targets' available.

# Strengths, Limitations, and Future Research

The main strength of our research lies in the replication of findings across two heterogeneous samples as well as an extension of our moderator from self-esteem to the more general higher order construct of CSEs. This consistent pattern of findings is noteworthy given the acknowledged difficulty in detecting moderation within field settings (McClelland and Judd, 1993). Furthermore, given that the samples were drawn from diverse workplace settings and industries, this lends strength to the generalizability and robustness of the results.

Another methodological strength was the use of multi-source measurement which reduces concerns regarding common source variance with respect to inflating the main effects found between predictors and the dependent variables as well as regarding the attenuation of the interaction effects (Siemsen et al., 2010).

Although our two studies show consistent results and enhance our understanding regarding the interplay of leader and follower dispositional characteristics on perceived abusive leadership by followers, they are not without limitations. Firstly, while our theory provides a strong indication as to the direction of the proposed relationships, the cross-sectional nature of our data prevents assertions of causality. For example, an alternative explanation to our findings in Study 2 could be that followers with low self-esteem and low CSEs receive abusive supervision from narcissistic leaders because they are perceived to be performing less well than followers with high self-esteem and high CSEs. Narcissists are overly critical of others and demand perfectionism (Stoeber et al., 2015), thus, insofar as narcissistic leaders feel that the performance of their followers is reflective of their own success, they may indeed wish to punish low performing followers. Future studies could employ a longitudinal design and measure follower job performance over time to examine how lower or higher follower job performance ratings or evaluations subsequently influence different followers' perceptions of abuse from narcissistic leaders.

Secondly, because abusive supervision as it was measured in our studies concerns followers' subjective perceptions regarding a leader's mistreatment, it may not reflect the actual levels of mistreatment. Thus, narcissistic leaders could be actually behaving more abusively toward those with low self-esteem and low CSEs, or these followers may simply be more attuned to potential victimization (Aquino and Thau, 2009) and as such experience narcissists' dominance, lack of empathy and egocentrism as abusive. Nonetheless, researchers have argued that follower perceptions are critical to include in measures of abusive supervision because leader behavior can only have an effect on followers if it is also perceived by them (Schyns and Schilling, 2013). Future research could further disentangle perceptions of abuse and actual abuse by asking leaders to report on their abusive behavior in relation to specific followers. Another possibility would be to use an experimental paradigm in which actual abusive behavior is observed and contrasted with perceptions of abusive behavior. Given our argumentation that narcissistic leaders would perceive followers with low self-esteem and low CSEs as easy targets, we would expect leader narcissism to be positively related to more actual reported abusive behavior toward these vulnerable followers.

Thirdly, in order to obtain multiple followers, in Study 1 the leaders were asked to nominate followers who would fill out the questionnaire. This might have inadvertently led to a selection bias as leaders might have chosen only those followers with whom they had a good relationship. However, given the consistent findings across both of our studies, it does not appear that this potential bias overrode or influenced the found results.

Finally, because the focus of our research was solely on narcissistic leaders, we did not measure the other two Dark Triad traits (Machiavellianism and psychopathy; Paulhus and Williams, 2002). Therefore, it was not possible to control for these constructs to isolate the unique effects of leader narcissism. Future research should consider measuring all three of the dark triad traits simultaneously to examine whether our findings generalize uniformly or differentially to the other two dark triad traits.

Our research focused on the impact of leader narcissism on followers' emotional exhaustion and task performance (Study 2). Future studies could test our model using other important outcome variables that are known to be affected by abusive supervision, such as followers' job satisfaction, organizational commitment, job stress, vitality, turnover intentions and organizational citizenship behavior (Tepper, 2007; Martinko et al., 2013; Schyns and Schilling, 2013; Mackey et al., 2015). Prior research has shown that due to the sense of injustice that they feel, followers tend to retaliate in response to abusive supervision in the form of deviant behavior such as aggression, theft and sabotage (e.g., Tepper et al., 2008, 2009; Liu et al., 2010; Burton and Hoobler, 2011). Thus, another interesting avenue of research would be to examine whether or not vulnerable followers (i.e., those with low self-esteem and low CSEs) would show such retaliation toward narcissistic leaders. Because individuals with low self-esteem are in general reluctant to engage in confrontation (Gudjonsson and Sigurdsson, 2003) particularly with authority figures, and because their feelings of low self-worth may lead them to believe that abusive behavior is justified (Padilla et al., 2007; Thoroughgood et al., 2012), such vulnerable individuals might be less likely to retaliate against narcissistic leaders, at least in an overt manner. Moreover, prior research shows that it is individuals with high (unstable) self-esteem who are more likely to react aggressively to threats to their ego rather than those with low self-esteem (Bushman and Baumeister, 1998; Bushman et al., 2009). The reason is that low self-esteem individuals tend to be more cautious and risk-averse in their responses, which makes them unlikely to react aggressively (Baumeister et al., 2000).

A final fruitful direction for future research would be to more closely examine the formation of dependence between susceptible followers and destructive leaders, such as narcissistic leaders. Because of their strong need for affirmation, desire for clarity, direction and higher self-esteem, the so-called 'lost souls' seek out charismatic and powerful leaders and thereby make themselves vulnerable to abuse by such leaders (Hayes, 2014; Padilla et al., 2007; Thoroughgood et al., 2012). Their high psychological need for such leaders also makes it likely that they will become dependent on such leaders. We know for instance that followers' personal identification with their transformational leaders (i.e., extent to which an individual's belief about the leader is selfreferential) fosters greater dependence on those leaders rather than empowerment (Kark et al., 2003). It would be interesting to examine whether followers with low self-esteem and negative CSEs show greater dependence on (narcissistic) leaders, and what effects this has, for example in terms of stifling employee voice.

### Practical Implications and Conclusion

This research has several practical implications for organizations. First, given the negative impact of narcissistic leaders on vulnerable followers, organizations could consider obtaining narcissism ratings of job applicants and restricting narcissists' entry to leadership functions, or getting rid of narcissistic leaders altogether. In light of the current findings, avoiding narcissistic individuals in leadership positions might appear to be an attractive alternative, however, narcissistic individuals also have positive characteristics which could make them useful for organizations in certain contexts (Sedikides and Campbell, 2017). For example, narcissists promote bold visions and are charismatic, they tend to persist in the face of failure, and they are good in crisis management (Galvin et al., 2010; Watts et al., 2013). Thus, a more fruitful alternative might be for organizations to attempt to find the best fit between managers and their subordinates. For example, if project teams are being formed then organizations should consider allocating subordinates with lower self-esteem or negative CSEs to project leaders who are lower on narcissism. Additionally, because vulnerable followers are more likely to perceive abusive behavior from narcissistic leaders and might be reluctant to speak out about this, it is important for such employees to be provided with support networks and means of voicing their concerns and feelings. Organizations could, for example, provide these employees with support groups, or a mentor that they could safely talk to when in need. From a preventative perspective, trainings on increasing self-confidence, self-esteem, and self-efficacy could be initiated to help these employees become more resilient to narcissistic leaders.

To conclude, we show that despite having many negative characteristics such as egocentrism, aggression, exploitativeness and lack of empathy, narcissistic leaders do not indiscriminately negatively affect all people they lead. In fact, the toxic effects of narcissistic leaders in terms of perceived abusive supervision, seem to be only experienced by vulnerable followers who have low self-esteem or low core self-evaluations. This research thus helps shed light on the consequences of narcissistic leaders for those they lead and identify which followers are more or less susceptible to experiencing the dark side of these leaders.

# AUTHOR CONTRIBUTIONS

BN, ADH, DDH, and FB conceived and developed the project, contributed to the interpretation of the results. Data collection was coordinated and conducted by BN and ADH. BN and ADH performed the data analyses. BN drafted the manuscript. ADH, DDH, and FB provided the critical revisions. All authors agreed to all aspects of the work and approved the final version of the manuscript.

# REFERENCES

fpsyg-09-00422 March 27, 2018 Time: 17:35 # 12




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer SS and handling Editor declared their shared affiliation.

Copyright © 2018 Nevicka, De Hoogh, Den Hartog and Belschak. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# No Regard for Those Who Need It: The Moderating Role of Follower Self-Esteem in the Relationship Between Leader Psychopathy and Leader Self-Serving Behavior

Dick P. H. Barelds<sup>1</sup> \*, Barbara Wisse1,2 \*, Stacey Sanders<sup>1</sup> and L. Maxim Laurijssen<sup>3</sup>

<sup>1</sup> Department of Organizational Psychology, University of Groningen, Groningen, Netherlands, <sup>2</sup> Durham University Business School, Durham University, Durham, United Kingdom, <sup>3</sup> Department of Economics and Business, University of Groningen, Groningen, Netherlands

### Edited by:

Astrid Schütz, University of Bamberg, Germany

### Reviewed by:

Armin Pircher Verdorfer, Technische Universität München, Germany Iris Katharina Koch, University of Bamberg, Germany

### \*Correspondence:

Dick P. H. Barelds d.p.h.barelds@rug.nl Barbara Wisse b.m.wisse@rug.nl

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 01 September 2017 Accepted: 04 July 2018 Published: 24 July 2018

### Citation:

Barelds DPH, Wisse B, Sanders S and Laurijssen LM (2018) No Regard for Those Who Need It: The Moderating Role of Follower Self-Esteem in the Relationship Between Leader Psychopathy and Leader Self-Serving Behavior. Front. Psychol. 9:1281. doi: 10.3389/fpsyg.2018.01281 Recent instances of corporate misconduct and examples of blatant leader self-serving behavior have rekindled interest in leader personality traits as antecedents of negative leader behavior. The current research builds upon that work, and examines the relationship between leader psychopathy and leader self-serving behavior. Moreover, we investigate whether follower self-esteem affects the occurrence of self-serving behavior in leaders with psychopathic tendencies. We predict that self-serving behaviors by psychopathic leaders are more likely to occur in the interaction with followers low in selfesteem. We first conducted an experimental study (N = 156), in which we manipulated follower self-esteem, measured leader psychopathy, and assessed their combined effect on leader self-serving behavior using an ultimatum game. We then conducted a multisource field study (N = 124 leader–follower dyads) using questionnaires to assess leader psychopathy, follower self-esteem, and perceived leader self-serving behavior. Across both studies, we found that leader psychopathy was positively related to their selfserving behavior, but only when followers had low rather than high self-esteem. As expected, our studies showed that the degree to which (perceived) psychopathic traits of leaders are reflected in their behavior depends on the characteristics of their followers. Apparently, the behavioral expression of negative leader traits is not only a matter of the trait strength, but instead is the result of the interplay between leader and follower in a certain context.

Keywords: self-serving behavior, psychopathy, self-esteem, leadership, followership, Dark Triad

# INTRODUCTION

Although leaders are expected to take group and subordinate interests in consideration when making decisions (Northouse, 2004), some leaders clearly fail to do so. Indeed, recent media accounts have made blatantly clear that some leaders act self-servingly and disregard the needs of others. A recent example is Martin Shkreli who hiked up the price of popular AIDS medicine Daraprim by 5,000% – from \$13.50 to \$750 – and who was arrested after being accused of running

a "Ponzi scheme" in order to pay for personal debts. Not only does such self-serving behavior lead to angry and shocked responses from the general public, research has pointed out that it also undermines the effectiveness and functioning of organizations and the people working in them. Indeed, compared to group or employee focused leaders, self-serving leaders contribute negatively to organizational performance and employee functioning (e.g., Carmeli and Sheaffer, 2009; Mayer et al., 2009; Peterson et al., 2012; Kalshoven et al., 2013; Williams, 2014). In order to mitigate the apparent surplus of self-serving behavior in some leaders, it may be helpful to understand how we can account for it.

In this article, we put forward the hypothesis that leader psychopathy is positively related to their self-serving behavior and disregard for other peoples' interest. Recent research has shown that psychopathic traits (the tendency to be manipulative, callous, egocentric, and a lack of empathy) may explain a fair amount of destructive leadership, unethical behavior, conflict, immoral decision making, and other types of corporate misconduct (cf. Boddy, 2011; Wu and LeBreton, 2011; Smith and Lilienfeld, 2013; Spain et al., 2014). Given the self-interested and uncaring features of the trait, we expect leader psychopathy to also be related to the extent to which leaders use their position to satisfy their self-serving needs at the expense of their followers.

Importantly, however, Trait Activation Theory stipulates that while a trait is unlikely to change, the nature and frequency of its expression can be altered by the context in which people operate (Christiansen and Tett, 2008; also see Padilla et al., 2007; Krasikova et al., 2013; McCabe and Fleeson, 2016). Logically, the same should hold for leader psychopathy. Because leaders always operate in a context where followers are present, we investigate whether follower characteristics affect the occurrence of self-serving behavior in leaders with psychopathic tendencies. We will focus on follower self-esteem, or follower overall self-evaluation, because it has been found to greatly affect their functioning in organizations (Bowling et al., 2010; Kuster et al., 2013). Those with low self-esteem are likely to be compliant and more susceptible to the illtreatment of others (cf., Bowling and Beehr, 2006). Moreover, psychopaths seem to have a certain prowess in picking up on the vulnerability of others (Wheeler et al., 2009). We expect, therefore, leader psychopathy to be more strongly positively related to leader self-serving behavior when followers have low self-esteem, and that follower high self-esteem can act as a buffer against self-serving tendencies of leaders with psychopathic traits.

All in all, this research aims to increase our understanding of when leaders are more likely to make self-serving decisions. We point to the interplay of notable leader and follower individual difference variables as an important precursor for such behaviors. In doing so, we aim to highlight that the social context in which a leader operates may affect the likelihood that leader traits will be reflected in their self-serving behavior. Finally, our study may provide more insight into factors that could prevent employees from becoming the victims of those willing to exploit others, and help those who are already targeted to cope more effectively with the self-serving behaviors of their leaders.

# Psychopathy in Leaders and Self-Serving Decision-Making

Psychopathy is part of the Dark Triad, a constellation of three personality traits: Machiavellianism, psychopathy, and narcissism (Paulhus and Williams, 2002). All three refer to short-term, selfserving, exploitive social strategies that are positively associated with disagreeableness, and the use of dishonest and manipulative behaviors (e.g., Jonason and Webster, 2010). Psychopathy, however, is often considered to be the darkest trait of these three, because it is unique in its coldness (Paulhus, 2014). Moreover, whereas all three Dark Triad traits are characterized by selfishness, those high in psychopathy are also characterized by recklessness (Jones, 2014; Jones and Paulhus, 2017), and the display of antisocial behaviors. In addition, individuals high in psychopathy do not respond well to punishment (e.g., Jones, 2014). As a result, they are likely to harm others for their own gain (Jones, 2014), even in high-risk situations (e.g., Jones and Paulhus, 2017). In line with this, Laurijssen and Sanders (2016) reported those high in psychopathy to have a hostile attitude toward coworkers, and to be characterized by greed. Given the fact that psychopathy appears to be the Dark Triad trait that is most likely to be related to self-serving behavior, the focus of the present study will therefore be on psychopathy, instead of all three Dark triad traits (cf. Jones, 2014).

The Dual Process model (Levenson et al., 1995; Fowles and Dindo, 2009) distinguishes between two forms of psychopathy: primary "emotionally stable" psychopathy, and secondary "reactive" psychopathy. Individuals with higher levels of primary psychopathy are characterized by their manipulativeness, glibness, egocentricity, callousness, and a general lack of empathy and guilt, whereas those with higher levels of secondary psychopathy display impulsive behavior, a need for stimulation, a parasitic lifestyle, and anti-social felonious tendencies. Notably, whereas individuals who score high on both primary and secondary psychopathy often end up incarcerated, the ones that only score high on primary psychopathy fare relatively well in society (Fowles and Dindo, 2009). Probably as a result of this, and the fact that primary psychopathy seems to capture most of the core of the psychopathy concept (cf. Lykken, 1995; Murphy and Vess, 2003), studies in organizational psychology, including the present one, focus on primary psychopathy.

Employees scoring high on primary psychopathy often obtain relatively high ranked positions in organizations (e.g., Babiak and Hare, 2006; Lilienfeld et al., 2012; Howe et al., 2014; Brooks and Fritzon, 2016). It has been argued that the so-called successful psychopath, or the corporate psychopath owns his/her success to the employment of effective communication styles, strategic thinking, impression management skills, and charisma (e.g., Babiak and Hare, 2006; Babiak et al., 2010; Harms et al., 2011; Smith and Lilienfeld, 2013). So, although psychopathic traits have some beneficial effects in the work context (beneficial for the person with the psychopathic traits), they are also known to have some damaging ones (damaging for the organization and its employees).

Specifically, psychopathic leaders seem to have lower objective performance levels (Babiak et al., 2010; O'Boyle et al., 2012), and are more likely to engage in risky and/or unethical decision-making (Stevens et al., 2012; Jones, 2014). Moreover, psychopathy has been positively related to counterproductive work behavior (CWB) (O'Boyle et al., 2012), white-collar crime (Ragatz et al., 2012), corporate misbehavior (Clarke, 2005), bullying, and abusive supervision (Boddy, 2011; Laurijssen et al., 2016, unpublished). In addition, leader psychopathy has been negatively related to individual consideration (Westerlaken and Woods, 2013), and employee well-being and satisfaction (Mathieu et al., 2014).

So far, there has not been any notable research attention to the relationship between psychopathy and self-serving leader behavior. Leader self-serving behavior reflects both acts aimed at securing higher monetary benefits for oneself, as well as making self-serving causal attributions, such as taking unwarranted credit for a group accomplishment or by denying responsibility for failure when it comes to group projects (cf. Weary Bradley, 1978; Rus et al., 2010). It may therefore be distinguished from detrimental leadership behaviors such as abusive supervision (cf. Tepper, 2000), or common types of CWBs (cf. Robinson and Bennett, 1995; Spector et al., 2006). Abusive supervision may be defined as "subordinates' perceptions of the extent to which their supervisors engage in the sustained display of hostile verbal and non-verbal behaviors, excluding physical contact" (Tepper, 2000, p. 178). CWB on the other hand is an umbrella term that may be defined as "any intentional behavior on the part of an organization member viewed by the organization as contrary to its legitimate interests" (Sackett and DeVore, 2001, p. 145). As such, CWB has been proposed to be at the top of a hierarchy, with lower level group factors such as organizational and interpersonal CWB (e.g., Bennett and Robinson, 2000), and more specific behaviors (such as theft, drug or alcohol use, poor attendance, etc.) below these group factors (Sackett and DeVore, 2001). Although measures for these concepts might include one or two items referring to self-interested behaviors (e.g., Schmid et al., 2017), these concepts do not explicitly include self-serving behavior in their definitions.

Given what we know about psychopathy, we can expect a positive relationship between psychopathy and leader self-serving behavior. Those with psychopathic traits are considered to be egotistic and manipulative (Jonason and Webster, 2010), and it has been argued that psychopathic traits may facilitate the effective and unremorseful exploitation of others for personal gain due to a lack of empathic concern (Jonason and Krause, 2013; Jonason et al., 2013). Leaders with psychopathic traits may likewise engage in self-serving behavior at the costs of others, especially because the leader role often comes with power, and power increases the likelihood that people will behave according to their traits (cf., Williams, 2014). Moreover, those scoring high on psychopathy often perceive their workplace as competitive (Jonason et al., 2015), arguably as a function of their competitive orientation (Ten Brinke et al., 2015), which may further enhance the likelihood that resources will be claimed for personal benefit at the expense of others. Therefore, we expect that:

Hypothesis 1: Leader psychopathy will be positively related to leader self-serving behavior.

Yet, the extent to which negative leader traits are manifested in their behavior is not only a matter of the strength of the trait (cf. Padilla et al., 2007; Krasikova et al., 2013). Leaders do not operate in a vacuum, instead the leadership role is highly social in nature and followers are part and parcel of the leadership process (Uhl-Bien et al., 2014). To understand and predict leader behavior one needs to consider the leader as well has her or his followers in their particular context and take their interaction into account (Padilla et al., 2007). In this paper, we focus on one characteristic of followers that has been coined as potentially important when studying the consequences of leader psychopathic traits: selfesteem (Thoroughgood et al., 2012).

It has been argued that some people are chronically more likely to fall victim to all sorts of negative interpersonal behaviors, including being the ones that receive the short of the end of the stick when it comes to the division of resources (Zapf and Einarsen, 2003). Knowledge about what characterizes target followers of the harmful behaviors of leaders scoring high on psychopathy can help identify individuals who may be in need of help, now or in the future. Moreover, it could be used to develop interventions aimed at (1) preventing vulnerable individuals becoming the victims of those eager to exploit others, and (2) helping those who are already targeted to cope more adequately with the self-serving behaviors of their leaders.

# The Role of Follower Self-Esteem

Self-esteem is a personal evaluation reflecting what people think of themselves as individuals. As such, it refers to an individual's overall self-evaluation of his/her competencies (Rosenberg, 1965). Self-esteem also has an affective component: people with high self-esteem like who and what they are, and people with low self-esteem do not (Pelham and Swann, 1989). People with low self-esteem are often attracted to others they believe can provide them direction, the possibility to be a more 'desirable' or 'better' person, an increased sense of self-worth, and/or a sense of belonging (Howell and Shamir, 2005; Padilla et al., 2007; Thoroughgood et al., 2012). As such, they often turn to their leaders and develop a strong desire to emulate and garner approval from that leader (Howell and Shamir, 2005). Their desire to gain acceptance and approval from their leader also explains their motivation to be compliant and their susceptibility to exploitation (Barbuto, 2000; Howell and Shamir, 2005; Thoroughgood et al., 2012). Followers with low self-esteem are less likely to object to exploitation out of fear of rejection and disapproval. Weierter (1997) even goes as far as to argue that people with low selfesteem are more likely to identify with leaders who want to control and manipulate others, because those followers feel they deserve such treatment, thereby perpetuating a negative cycle of exploitation. It seems thus that persons with low self-esteem are susceptible to the influence of leaders with self-serving motivations.

Empirical support for this idea comes mostly from studies outside the leadership field. For instance, prior research on

abusive behavior and workplace victimization points to low selfesteem, dependence, social isolation, and social incompetence as characteristics of potential victims (Harvey and Keashly, 2003; Zapf and Einarsen, 2003; Bowling and Beehr, 2006; Matthiesen and Einarsen, 2007). Moreover, those with low self-esteem are more likely to report problems related to being self-sacrificing and overly accommodating (Paz et al., 2017). Finally, research has shown that those with low self-esteem react to ego-threats with behavior that is more friendly, cautious and restrained, arguably because they are more focussed on belongingness needs and thus are focused on establishing a relationship with others, whether they respect them or not (Heatherton and Vohs, 2000; Vohs and Heatherton, 2001). Thus, those with low self-esteem may have less of a defense system against exploitation by those who are looking to further their own interests.

The compliant tendencies and the openness to exploitation of people with low self-esteem, however, may not be the only reason why those with low self-esteem may fall victim to the selfserving tendencies of leaders with psychopathy traits. In addition, subtle behavioral patterns and gestures of the low self-esteem followers might indicate that they offer little resistance in case of abuse (i.e., gestural hinting, see Grayson and Stein, 1981). It has been suggested that those with psychopathic traits are particularly capable of recognizing others' vulnerability and have a willingness to exploit that. The infamous psychopathic serial killer Ted Bundy, for instance, boasted about his observational competencies by stating that he "could tell a victim by the way she walked down the street, the tilt of her head, the manner in which she carried herself, etc. . . " (as cited in Holmes and Holmes, 2009, p. 221). Several scholars indeed confirm that victims share certain characteristics that seem to predispose them for abuse and exploitation (Grayson and Stein, 1981; Richards et al., 1991; Gunns et al., 2002; Sakaguchi and Hasegawa, 2007). Other studies indicate that those with psychopathic traits are particularly likely to pick up on those characteristics (Book et al., 2013). For instance, Wheeler et al. (2009) found that psychopathic traits in a non-referred (and presumably not clinically psychopathic) sample increased the accuracy of perceptions of victim vulnerability. More recently, Demetrioff et al. (2017) found that individuals' psychopathy scores were even positively associated with a heightened ability to identify sadness micro-expressions (note that low self-esteem often goes hand in hand with negative emotions) which further indicates their prowess in vulnerability assessment.

The idea of psychopaths being "social predators" (e.g., Hare, 2001; Book et al., 2007) hence seems to be justified. Boddy (2011) argues that such predatory behavior can be found in organizational contexts as well and conjectured that corporate psychopaths would mainly exploit those followers who are unlikely to defend themselves. The likelihood that leaders will engage in more self-serving behavior vis-a-vis followers that have low self-esteem might thus stem from followers' own compliant tendencies, as well as from the psychopathic leader's competencies in recognizing vulnerability and their willingness to take advantage of that. Notably, this resonates with Trait Activation theory (e.g., Christiansen and Tett, 2008), where it is argued that traits can be seen as latent propensities to behave in a certain way as a response to trait relevant cues (such as social cues). We posit that psychopathic traits in leaders carry the propensity to behave self-servingly as a response to cues that signal low self-esteem in followers. Given that expressing one's traits is intrinsically satisfying, we expect that:

Hypothesis 2: Leader psychopathy will be positively related to leader self-serving behavior to the extent that leaders are dealing with followers suffering from low self-esteem.

# Overview of the Present Research

We opted for a multiple-study, multiple-method approach so that comparable results between studies increase the confidence in our findings. In Study 1, a laboratory study with business leaders, we measured psychopathy and assigned all participants to a leader role. We then manipulated follower's self-esteem, and asked the leader to perform a task in which they had the possibility to display self-serving behavior at the expense of the follower (using an ultimatum game). Study 2, was a multi-source field study (N = 124 unique leader–follower dyads) using questionnaires to assess leader psychopathy, follower self-esteem, and leader self-serving behavior as rated by the follower.

# STUDY 1

### Materials and Methods Participants and Design

The study was conducted as an online survey of people in leadership positions from the United States, holding a job for at least 3 days a week. The 156 participants were randomly assigned to one of two conditions (Follower self-esteem: low vs. high), and participants' psychopathy scores were added to the design as a continuous variable. Most of our participants worked in technology (17.9%), business and finance (17.3%), manufacturing (9.6%), education (8.3%), or human services (8.3%). Participants were predominantly male (57.7%), and their mean age was 36.33 (SD = 11.00). Most of them had a Bachelor degree or higher (76.3%), had been working on average for 16.07 years (SD = 10.46), and supervised on average 13.00 employees (SD = 34.86).

### Procedure

Leaders were recruited using Amazon's Mechanical Turk (Mturk). There are several scholars that advocate the use of Mturk data (Paolacci et al., 2010; Buhrmester et al., 2011; Mason and Suri, 2012), also specifically for organizational research (e.g., Rietzschel et al., 2017), as well as studies that have actually used Mturk for collecting leader data. Van Houwelingen et al. (2017), for example, used Mturk for sampling leaders and found results comparable to other samples they used in the same paper<sup>1</sup> . Participants were informed that the research would take approximately 15 min to complete, that the data collected with this study would be treated confidentially, and that they would

<sup>1</sup> In line with recommendation for studies using Mturk samples, we deleted participants who indicated, we should not use their data or had absolute studentized residuals larger than 3 (Cheung et al., 2017).

receive \$1.30 in return for their participation. Participants also learned that they had the opportunity to earn a bonus payment (based on task performance).

The study consisted of two parts. In the first part, participants provided their informed consent via the program software and filled out several questionnaires - including our psychopathy measure. In the second part, participants played an ultimatum game with a fictitious other Mturk worker, with whom they were allegedly randomly paired up. Participants were told that, based on a comparison of their own and the other person's answers on the questions in the first part, one of them would be assigned the role of leader and the other one the role of subordinate. In reality, all participants were assigned the leader role (there were no subordinates). Next, participants were presented with the instructions for the task – an ultimatum game – in which they were to divide a bonus payment between themselves and their subordinate (the fictitious other Mturk Worker). Participants then performed the task, answered some (demographical and manipulation check) questions, were debriefed, thanked, and paid (a base pay of \$1.30 and a bonus [up to \$0.60]). The experimental procedure was approved by the Ethics Committee of Psychology of the University of Groningen.

### The Experimental Task: An Ultimatum Game

In a (symmetrical) ultimatum game people are asked to divide money or other rewards between themselves and another person (Handgraaf et al., 2003). The game usually has two players: an allocator and a recipient. The allocator is asked to divide the money, and the recipient has the chance to either accept or reject the offer made by the allocator. If the recipient accepts the offer, the money will be divided based on the proposal made by the allocator. If the recipient rejects the offer, both get nothing. In a symmetrical ultimatum game, both the allocator and the recipient know how much money can be divided (symmetric information; e.g., van Dijk and Vermunt, 2000). As keeping more money for oneself automatically results in less money for the other, the game has been used in previous research to assess self-serving behavior that comes at the expense of another person (van Dijk and Vermunt, 2000; Sanders et al., 2016). In our experiment, all participants had the allocator role and had to decide on the distribution of a bonus payment of 60 dollar cents between themselves and their (fictitious) subordinate. The allocators did not know that there were no actual recipients. The ultimatum game in this case is strictly speaking not symmetrical, since there are no real recipients involved.

### Self-Esteem Manipulation

Before playing the ultimatum game, participants received some information about their subordinate. Specifically, participants were presented with a table displaying their subordinate's alleged scores on the first five items of the Rosenberg (1965) self-esteem scale, which we used to manipulate the subordinate's level of selfesteem. In the low self-esteem condition, participants could see that the recipient scored a 1 or a 2 on a 7-point scale on items such as: "On the whole I am satisfied with myself." In the high selfesteem condition, participants could see that the recipient scored a 6 or a 7 on these items.

# Measures

### **Psychopathy**

Leaders' psychopathic traits were assessed with Levenson's Self-Report Primary Psychopathy Scale (LSRPA; Levenson et al., 1995). This 16-item scale for the assessment of primary psychopathy includes items such as: "I enjoy manipulating other people's feelings" and "For me, what is right is whatever I can get away with." Leaders indicated their agreement with the statements using a 4-point Likert scale (1 = disagree strongly, 4 = agree strongly). In a recent study, Tsang et al. (2018, p. 316) found that the items of the Levenson scale "assessing primary psychopathy are better at differentiating between individuals with varying levels of psychopathic traits than items measuring secondary psychopathy features." This study also found confirmatory support for the primary, but not the secondary psychopathy scale. The reliability of the primary psychopathy scale in the present study was very good with Cronbach's alpha = 0.93 (M = 1.86; SD = 0.59).

### **Manipulation check**

To assess the effectiveness of the self-esteem manipulation, we used the final five items of Rosenberg's (1965) self-esteem scale. Participants were for instance asked to what extent the other person ". . . has a positive attitude toward him/herself," ". . . is inclined to feel that he/she is failure" (R). Participants' responses were assessed using a 5-point Likert scale (1 = not at all, 5 = extremely). Cronbach's alpha of the scale = 0.96 (M = 2.99; SD = 1.46).

### **Leader self-serving behavior**

The number of cents leaders allotted to themselves (at the expense of their subordinate) in the ultimatum game comprised our behavioral measure of leaders' self-serving behavior (M = 33.43; SD = 6.71).

### **Controls**

We controlled for supervisor age (Barlett and Barlett, 2015), and gender (Webster and Jonason, 2013; coded 1 = male; 2 = female), because previous research found these variables to be related to psychopathy (cf. Jonason and Webster, 2010; Wisse and Sleebos, 2016).

# Results

### Manipulation Check

An independent samples t-test showed that participants in the high self-esteem condition (M = 4.31, SD = 0.61) perceived their subordinate to have higher self-esteem than those in the low self-esteem condition (M = 1.70, SD = 0.66), t(154) = −25.57, p < 0.001 (mean difference = −2.61, 95% CI = [−2.81, −2.41]). These results demonstrate that the manipulation worked as intended.

### Leader Self-Serving Behavior

We predicted that leader psychopathy and subordinate selfesteem would interact in such a way that particularly when subordinate self-esteem is low, leader psychopathy would be related to self-serving behavior. To test this hypothesized moderation, we relied on a procedure suggested by Hayes (2013;

model 1; see **Table 1**). We controlled for supervisor age and gender. We found a main effect of supervisor psychopathy, showing that supervisors were more self-serving when they scored higher on psychopathy. In addition, and in line with our hypothesis, we found that the interaction term of supervisor psychopathy and employee self-esteem significantly predicted the amount of money that the supervisor took for him/herself. We tested the conditional direct effects of supervisor psychopathy on the dependent variable (self-serving behavior) at different levels of employee self-esteem. Bootstrapping (5,000 samples) confirmed that the direct effect of supervisor psychopathy on self-serving behavior was significant for employees with low selfesteem (b = 4.49, 95% CI = [2.08, 6.90]; 1 SD below the mean), but not for employees with high self-esteem (b = 0.73, 95% CI = [−2.09, 3.55]; 1 SD above the mean) (**Figure 1**).

# STUDY 2

### Materials and Methods Respondents

We approached 300 dyads of Dutch subordinates and their direct supervisors. After initial screening (removing respondents who did not fill out all psychopathy, self-serving behavior or selfesteem questions), we had a dataset of 124 dyads (response rate 41.33%). A total of 42.7% of the subordinates and 65.3% of the supervisors indicated to be male. Subordinates' mean age was 31.20 (SD = 11.52) and supervisors' mean age was 40.48 (SD = 11.07). A total of 65.0% of the supervisors and 30.6% of the subordinates indicated having worked more than 5 years in their current position. Most supervisors and subordinates worked more than 25 h a week (92.7% and 53.2%, respectively). The majority of our respondents worked in commercially oriented (service) organizations (e.g., shops, financial institutions, health care organizations, etc.; 73.3%).

TABLE 1 | Regression results for the (conditional) effects of Study 1.


<sup>a</sup>Unstandardized regression coefficients; <sup>∗</sup>p < 0.05, ∗∗p < 0.01.

### Procedure

Data were collected as part of a study on the role of personality in the workplace. Research assistants used their own work environment, their personal network and that of acquaintances to get in contact with employees and supervisors. In addition, they actively visited business and shopping centers. Potential participants were approached via email, through phone calls, or face-to-face contact. We stressed the fact that participation was voluntary and that data would be treated confidentially. If subordinates and their supervisors were interested in participating, they were asked to fill out the paper-and-pencil questionnaires without consulting their colleagues, subordinates or supervisor, and to return the questionnaires in the enclosed envelope. This envelope was then picked up by the research assistant or returned by mail. Because people often filled out the questionnaires during work hours, we kept the survey short and to the point. Respondents also had the option to fill in the questionnaire during their free time (e.g., during lunch breaks or at home). Participants gave their informed consent, and the study was approved by the Ethics committee of the University of Groningen.

### Measures

The following measures were used in this study:

### **Psychopathy**

Leaders' primary psychopathy was again assessed by asking supervisors to fill out LSRPA (Levenson et al., 1995; 1 = disagree strongly, 4 = agree strongly).

### **Self-esteem**

To measure the self-esteem of the follower, we asked subordinates to fill out the 10 items of the Rosenberg Self-Esteem Scale (Rosenberg, 1965), using a 5-point Likert scale (1 = completely disagree, 5 = completely agree).

### **Leader self-serving behavior**

Perceptions of the degree to which leaders demonstrated selfserving behavior were assessed using the scale developed by

Rus et al. (2010). Because, we asked subordinates to assess their leader's behavior, we removed one item from the original 8 item scale as subordinates generally do not have access to that information ("My supervisor negotiated a bonus for him/herself that was substantially higher than the bonus we receive"). The scale includes items such as "My supervisor has used his/her leadership position to obtain benefits for him/herself," and "Instead of giving credit to me or my colleagues for jobs requiring a lot of time and effort, my supervisor took the credit him/herself "). Subordinates rated their leaders' self-serving behavior using a 4-point Likert scale (1 = never, 4 = often).

### **Controls**

We again controlled for supervisor age and gender (1 = male; 2 = female). Additionally, we controlled for length of collaboration and frequency of contact (as indicated by the subordinate) because previous research suggests that others' perceptions of people scoring high on Dark Triad traits may change once they get to know them better (cf. Campbell and Campbell, 2009). Length of collaboration was assessed using five categories which were coded 1 (less than 6 months) to 5 (5 years or longer). Frequency of contact was assessed using a 5-point scale (1 = sporadic; 5 = very often) 2 .

### Results

**Table 2** shows means, standard deviations, zero-order Pearson correlations, and Cronbach's alphas for the study variables<sup>3</sup> . Cronbach's alpha's were all sufficiently high. Note that, we found a significant positive correlation between leader psychopathy and perceptions of leader self-serving behavior (r = 0.24, p < 0.01)

### Leader Self-Serving Behavior

We predicted that leader psychopathy would be more strongly related to (perceived) leader self-serving behavior to the extent that subordinate self-esteem is low. To test this hypothesized moderation, we again relied on Hayes (2013; model 1; see **Table 3**). We controlled for supervisor age, gender and length of collaboration and frequency of contact. We found a main effect of supervisor psychopathy, showing that supervisors were rated as more self-serving when they scored higher on psychopathy. We also found a main effect of employee self-esteem, showing that employees perceived more self-serving behavior when they had lower self-esteem. In addition, and in line with our hypothesis, we found that the interaction term of supervisor psychopathy and employee self-esteem significantly predicted (perceptions of) leader self-serving behavior. We tested the conditional direct effects supervisor psychopathy on the dependent variable (selfserving behavior) at different levels of employee self-esteem. Bootstrapping (5,000 samples) confirmed that the direct effect of supervisor psychopathy on (perceptions of) leader self-serving behavior was significant for employees with low self-esteem (b = 0.67, 95% CI = [0.33, 1.02]; 1 SD below the mean), but not for employees with high self-esteem (b = 0.00, 95% CI = [−0.32, 0.33]; 1 SD above the mean) (also see **Figure 2**).

# DISCUSSION

This study focused on leader psychopathic personality traits, follower self-esteem and leader self-serving behavior. Across two studies, one experimental study and one field study, we found that leader psychopathy positively predicts (perceived) leader selfserving behavior. More importantly, we found that follower selfesteem moderated the relationship between leader psychopathy and leader self-serving behavior. Only when followers had low self-esteem, we found that leader psychopathy and leader selfserving behavior were positively related. The results support and extend previous studies in several ways.

First, the study adds to the growing list of potential consequences of employing people with psychopathic tendencies by showing that leader psychopathy is associated with selfserving behavior. Given the adverse effects that leader selfserving behavior may have on outcomes for subordinates and for organizations (Carmeli and Sheaffer, 2009; Peterson et al., 2012) it is crucial to understand its determinants. Notably, the occurrence of self-interested behavior without heeding to the needs of others may prove particularly detrimental when larger resources are at stake (Wisse and Rus, 2012). This renders leader behavior particularly important, because leaders tend to have more control over resources than rank and file employees.

Second, the study indicates that followers can have an effect on the extent to which leaders' psychopathic traits will be reflected in their behavior. This finding thus confirms the notion that destructive organizational outcomes are not exclusively the result of destructive leaders, but are also products of 'susceptible followers' (Padilla et al., 2007; Thoroughgood et al., 2012). Thoroughgood et al. (2012) distinguished between two classes of susceptible followers: the conformers and the colluders. While conformers are prone to obedience, colluders actively contribute to the leaders' mission. In our study, we have considered the influence of low self-esteem; a characteristic that is likely to make a follower belong to the conformer category. These followers are considered vulnerable to leaders wishing to exploit them, arguably out of a fear of confrontation that creates a weakness to social pressures, or out of a need to gain the approval of someone who is able to provide clarity, direction, and increased self-esteem. Perhaps future research could focus on if the relationship between leader psychopathy and self-serving behavior is also strengthened by the presence of 'authoritarian' followers. Authoritarians also belong to the conformer category, but these follower possess rigid, hierarchical attitudes that prescribe leaders' legitimate right to exert (Altemeyer, 1998), and those with psychopathic traits may be inclined to make use of

<sup>2</sup>The present study was part of a large-scale study that also included scales for the other two Dark Triad traits: Machiavellianism (a shortened version of the MACH-IV; Belschak et al., 2015) and narcissism (the NPI-16; Ames et al., 2006). Because, we focus on psychopathy here, we did not use these scales. These scales were also not available in Study 1.

<sup>3</sup>The structure of the items assessing self-esteem, leader self-serving behavior, and psychopathy was examined by means of EFA. The EFA produced three oblique factors that very closely resembled the three a priori scales. Congruencies (Tucker's phi; Tucker, 1951) between the three EFA factors and the three a priori scales were subsequently computed in Matlab. The congruencies, after orthogonal Procrustes rotation (Kiers and Groenen, 1996), were: self-esteem ϕ = 0.95, leader self-serving behavior ϕ = 0.91, and psychopathy ϕ = 0.90. These results support the expected structure of the data (e.g., Lorenzo-Seva and Ten Berge, 2006).



N = 124; <sup>∗</sup>p < 0.05; ∗∗p < 0.01 (two-tailed); Cronbach's alpha between brackets.

TABLE 3 | Regression results for the (conditional) effects of Study 2.


<sup>a</sup>Unstandardized regression coefficients; <sup>∗</sup>p < 0.05, ∗∗p < 0.01.

that to their own advantage. Moreover, future research may want to investigate the role of colluders in more detail. Anecdotal evidence suggests that psychopaths may sometimes work through or with their 'henchmen' (Babiak and Hare, 2006; Langbert, 2010) to accomplish their self-serving goals, but more research on the matter is needed.

In addition, our study suggests that those with higher levels of psychopathy differentiate between the one person and the other in terms of victim selection. That is, leaders with psychopathic traits victimized followers with low selfesteem more than those with high self-esteem. Interestingly, a recent study (Black et al., 2014) argued that psychopathic personalities may not be so 'picky' when choosing a victim. This study examined the relation between the Dark Triad (psychopathy, Machiavellianism, and narcissism) and strategies used in the assessment of personality and emotional states related to vulnerability in others. Their results indicated that dark personalities engaged in a relatively superficial interpersonal analysis and generally perceived all targets as weak and vulnerable. The authors proposed that instead of being keen

"readers" of others, dark personalities, including psychopathic ones, may rely on their ability to draw in vulnerable victims (for instance based on their charisma or good looks) or adopt a "quantity over quality" strategy to find targets and then use manipulation tactics to exploit them. Our studies' results are more in line with other findings that suggest that psychopaths do differentiate between targets (Wheeler et al., 2009; Book et al., 2013; Demetrioff et al., 2017). Perhaps future research could look at potential moderators in order to explain when psychopaths make a distinction between potential targets and when they do not.

We would like to draw attention to a couple more issues that could fruitfully be addressed in future research. For instance, we mentioned that both a lack of self-esteem as well as the surplus in psychopathy may set in motion certain process that may explain why those with low self-esteem may fall prey to the exploitative tendencies of those scoring high on psychopathy. Indeed, we argued that low self-esteem may engender compliant behavior and vulnerability on the part of the follower, and that psychopathy may come with a knack for recognizing vulnerability and the willingness to misuse that on the part of the leader. Although we indeed find that leader psychopathy and follower self-esteem interact to explain

leader self-serving behavior, our results are mute when it comes to the underlying process. Of course, in the experiment the role of the follower was manipulated and there was no actual follower present, but in the field study the follower did exist and his/her behavior may have set exploitative processes in motion. Therefore, it remains unclear to what extent various processes are set in motion by leader psychopathy and follower self-esteem that may explain their combined effect. Future studies could explore the potential mediating roles of for instance compliance, a focus on vulnerability cues, etc. Moreover, it may be worthwhile to examine the implications of psychopathy for self-serving behavior over time using longitudinal designs (Smith and Lilienfeld, 2013). Psychopathy may be adaptive for first impressions, but over time co-workers and employers may begin to grow weary of psychopathic individuals and associated selfserving tendencies. It would be valuable to get more insight into to the long term development of leader exploitive behavior of followers.

# Strengths and Limitations

As with every study, the present study has its strengths and limitations. One strength is that by conducting both an experiment and a field study, we adopted a multiple-study, multiple-method approach in which the strengths of one method may compensate for any weaknesses in the others (Eid and Diener, 2006). For instance, the multiple method approach allowed us to assess leader self-serving behavior in different ways. In the field study, we asked followers to indicate the extent to which their leader displayed self-serving behavior. With this approach, we followed contemporary practices in leadership research and recommendations for research on consequences of Dark Triad personalities in the workplace (Smith and Lilienfeld, 2013). The use of subordinate perception data has the advantage of not having to rely on leaders own perceptions of their selfserving behavior (and thus of avoiding self-serving bias on the part of the leader). However, one potential drawback is that perceptions of observers and reality may differ as well (Hansbrough et al., 2015). Indeed, one may argue that those with low self-esteem are plainly more aware of or sensitive to negative behaviors of others, or that that their perception of negative behaviors of others is more negative than the perception of those with high self-esteem. Our experimental study shows, however, that these arguments cannot satisfactorily explain our findings. After all, in the experiment, we assessed self-serving behavior using a behavioral measure instead of a perception measure. In this study, we also found that leader psychopathy is only related to self-serving tendencies in the actual division of monetary rewards when the follower has low self-esteem.

Other compensatory advantages of the present paper's multimethod approach are, for instance, that even though we took special care to achieve a high degree of experimental realism, the experiment could still be criticized for its artificial character. That is, findings generated in an experimental environment provide no evidence that the same relationships actually exist outside the laboratory (Goodwin et al., 2000). Moreover, although an ultimatum game may be perceived as a simple form of leader– follower exchange (Price and Van Vugt, 2014), and ultimatum games have been used in previous research to assess self-serving behavior that comes at the expense of another person (e.g., van Dijk and Vermunt, 2000; Sanders et al., 2016), one might question the ecological validity of such a game. Our second study may alleviate these concerns as it shows that these relationships may indeed be observed in the field.

Another limitation of the present study is that we only examined primary psychopathy, not secondary psychopathy. Because primary psychopathy, contrary to secondary psychopathy, does not seem to hinder individuals from functioning reasonably well in society, and also appears to capture the core of the psychopathy concept as defined by Cleckley (1964; e.g., Murphy and Vess, 2003), we focused on primary psychopathy only. Future studies might, however, also examine secondary psychopathy, in order to examine the effect of both elements in the Dual Process model of psychopathy on leader self-serving behavior, or, additionally, all three Dark Triad traits.

# Practical Implications

Several scholars have cautioned against studying the role of victim personality and suggested that one has to be careful with respect to these issues, in order to avoid being accused of "blaming the victim" (Zapf and Einarsen, 2003). Of course, in studying victim personality, it is not our intention to hold victims responsible for their exploitation, neither do we suggest that others should do so. However, we hope our study makes clear that there are legitimate reasons to examine follower personality and the role it may play in self-serving leadership. Developing effective intervention techniques in order to prevent self-serving behavior by leaders depends upon a comprehensive understanding of the phenomenon. If follower personality and associated behaviors and needs (non-verbal behavior, compliance, need to belong) trigger self-serving behavior in leaders, followers are better off if they are aware of it so they can address it. Moreover, even if follower personality plays role in self-serving behavior of leaders, that does not mean that organizations and employers are cleared of a responsibility in the prevention and termination of such behaviors in the workplace. On the one hand organizations may assist targets in addressing or dealing with the issue, and on the other hand they may want to hold self-serving leaders accountable for their actions. Indeed, it has been suggested that accountability, a lack of ambiguity, and a clear set of values and norms may mitigate the negative impact of psychopaths in the workplace (Cohen, 2016). Moreover, the results of this study give further credence to the idea that organizations may want to be cautious with promoting those with dark personalities into positions of leadership (Wisse and Sleebos, 2016). Notably, to prevent those with higher levels of psychopathic traits from being promoted to or hired into leadership positions more research on how to screen for psychopathic traits is needed (Stevens et al., 2012; Spain et al., 2014).

All in all, given the negative consequences of leader selfserving behavior for subordinates as well as for the organization at large, more insight into the conditions that prompt supervisors to engage in such behavior is essential. We found that follower self-esteem and leader psychopathy jointly determined leader

self-serving behavior, as such that leader psychopathy predict leader self-serving behavior to the extent to followers had low self-esteem. We hope that our study stimulates research that employs an interactionist perspective (integrating both leader and follower characteristics) on the influence of corporate psychopaths at work. This may further insight into how organizations can protect themselves against the destructive influences of supervisors with psychopathic traits.

### REFERENCES


### AUTHOR CONTRIBUTIONS

All authors provided substantial contributions to the conception or design of the work; were responsible for drafting the work or revising it critically for important intellectual content; approved the final version of this manuscript; and agreed to be accountable for all aspects of the work.



negotiation. Pers. Individ. Diff. 79, 116–122. doi: 10.1016/j.paid.2015. 02.001


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer IK and handling Editor declared their shared affiliation at the time of review.

Copyright © 2018 Barelds, Wisse, Sanders and Laurijssen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Angels and Demons: The Effect of Ethical Leadership on Machiavellian Employees' Work Behaviors

### Frank D. Belschak\*, Deanne N. Den Hartog and Annebel H. B. De Hoogh

Section of Leadership and Management, Amsterdam Business School, University of Amsterdam, Amsterdam, Netherlands

Machiavellians can be characterized as goal-driven people who are willing to use all possible means to achieve their ends, and employees scoring high on Machiavellianism are thus predisposed to engage in unethical and organizationally undesirable behaviors. We propose that leadership can help to manage such employees in a way that reduces undesirable and increases desirable behaviors. Studies on the effects of leadership styles on Machiavellian employees are scarce. Here we investigate the relationship of ethical leadership with prosocial (helping colleagues or affiliative OCB) and antisocial work behavior (knowledge hiding and emotional manipulation) for employees who are higher or lower in Machiavellianism. The effect of an ethical leadership style on employees predisposed to engage in unethical behaviors has not been investigated so far. In a cross-sectional multi-source survey study among a sample of 159 unique leader–follower dyads, we find interaction effects between leadership and employee Machiavellianism for prosocial and antisocial work behavior. As expected, employee Machiavellianism comes with reduced helping behavior and increased knowledge hiding and emotional manipulation, but only when ethical leadership is low. Under highly ethical leaders, such increases in organizationally undesirable behaviors of Machiavellian employees do not occur. While the cross-sectional design precludes conclusions about the direction of causality, findings of our study suggest to further explore (and from a practical perspective to invest in) ethical leadership as a potential remedy for undesirable behavior of Machiavellian employees.

Keywords: Machiavellianism, ethical leadership, organizational citizenship behavior, knowledge hiding, emotional manipulation

# INTRODUCTION

The psychological literature describes Machiavellians as master manipulators who are willing to use all possible means to achieve their ends (e.g., Wilson et al., 1996; Jones and Paulhus, 2009). Employees scoring high on Machiavellianism (high-Machs) have been consistently found to engage in a plethora of unethical and counterproductive behaviors including lying, theft, sabotage, and bullying in numerous studies (see Dahling et al., 2012). High-Machs might eventually even contribute to the creation of an unethical organizational culture by acting as role models and signaling to others that "anything goes" (e.g., Felps et al., 2006; Pinto et al., 2008). As a consequence, the recommendation of most studies has been to identify and avoid high-Mach employees (e.g., Dahling et al., 2009; Kiazad et al., 2010). However, high-Mach individuals are proficient in

### Edited by:

Barbara Wisse, University of Groningen, Netherlands

### Reviewed by:

Athena Xenikou, Hellenic Air Force Academy, Greece Rita Berger, University of Barcelona, Spain

> \*Correspondence: Frank D. Belschak F.D.Belschak@uva.nl

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 28 September 2017 Accepted: 07 June 2018 Published: 28 June 2018

### Citation:

Belschak FD, Den Hartog DN and De Hoogh AHB (2018) Angels and Demons: The Effect of Ethical Leadership on Machiavellian Employees' Work Behaviors. Front. Psychol. 9:1082. doi: 10.3389/fpsyg.2018.01082

**103**

deceiving and manipulating their social environment (Davies and Stone, 2003; McIlwain, 2003), thus it might not always be easy to identify Machiavellians in organizations. Also, some authors have noted that high-Machs do not always engage in unethical and counterproductive behaviors; they also show pro-organizational behavior as long as they feel that this is instrumental for achieving their goals (Wilson et al., 1996; Belschak et al., 2015). As Belschak et al. (2015, p. 1935) argue, organizations cannot always avoid having some Machiavellian employees on board, and they suggest to move the focus toward having a better understanding of how to manage high-Mach employees in a way that reduces organizationally undesirable and increases desirable behaviors. Here, we propose that ethical leadership can offer effective ways to do so.

Research on leading Machiavellian employees is hardly available, and the effects of different leadership styles and behaviors on Machiavellian employees have not received much attention to date. The few existing studies focus on the effects of transformational leadership (Belschak et al., 2015), managerial control (Bagozzi et al., 2013), and leader Machiavellianism (Wisse et al., 2015; Belschak et al., 2016). None of these studies explore how to decrease high-Machs' highly undesirable tendency to engage in unethical behaviors. Somewhat related, Belschak et al. (2015) address how to increase high-Machs' pro-organizational behavior and show that transformational leaders, who emphasize the importance of new missions and organizational change, are able to stimulate challenging organizational citizenship behavior (OCB) such as making suggestions for change initiatives in high-Mach followers (Belschak et al., 2015). High-Mach employees have a strong goal orientation and instrumental focus (see Christie and Geis, 1970; Jones and Paulhus, 2009) and by emphasizing the importance of change and changeoriented behavior, showing their appreciation of such change initiatives, and empowering employees to make such changes, transformational leaders seem to stimulate high-Machs in particular to engage in such behavior. Yet, this strict goal orientation of high-Machs also implies that such increases in challenging OCB under transformational leaders might not generalize to a wider range of behaviors (e.g., helping colleagues if this is not clearly to their own benefit) and might not reduce unethical work behaviors (e.g., manipulating, cutting corners, or hiding knowledge from others). To stimulate these types of behaviors, we propose that leaders may need to emphasize specifically the importance of employees showing ethical behavior and hence explicitly engage in ethical leadership.

Ethical leaders (i.e., leaders who demonstrate "normatively appropriate conduct through personal actions and interpersonal relationships," Brown et al., 2005, p. 120) act as role models of ethical behavior, communicate ethical standards, reward ethical behavior, and punish unethical behaviors (see Den Hartog, 2015). Their behavior sends strong signals to their employees that ethical behavior is important and will be rewarded while unethical behavior is undesirable and will be punished. As noted, high-Machs' strong goal orientation ("doing what it takes to achieve one's ends") should make them particularly sensitive to the signals leaders send about what is appreciated, and high-Machs should hence react to ethical leadership with reduced unethical, antisocial work behavior (manipulation and knowledge hiding) and increased ethical, prosocial behavior (helping colleagues or affiliative OCB). Specifically, we hypothesize that compared to low-Mach employees, high-Mach employees show increased affiliative OCB and decreased knowledge hiding and emotional manipulation under highly ethical leaders and, vice versa, they show less affiliative OCB and more knowledge hiding and manipulation when ethical leadership is low. Greenbaum et al. (2017) argued that abusive leaders stimulate manipulative and unethical behavior in Machiavellians. Here, we similarly reason that low ethical leadership may stimulate unethical behavior such as deception and manipulation, whereas high ethical leadership may inhibit such behavior and rather stimulate ethical behavior including helping others in need rather than manipulating and hiding knowledge from them.

Our study adds to both the literature on leadership and on Machiavellianism. In particular, we contribute to the stream of literature investigating the impact of "dark-side" traits like the dark triad (Machiavellianism, psychopathy, and narcissism; Paulhus and Williams, 2002). While the main effects of Mach are well researched, the interactive effects of Mach in leader–follower interactions and the outcomes of these interactions only received attention more recently (e.g., Nevicka et al., 2011; Den Hartog and Belschak, 2012; Belschak et al., 2015; Wisse et al., 2015). We also add to a stream of research in leadership focusing on how to lead specific groups of employees. Based on their traits and values, employees seem to react differently to their leaders (e.g., Ehrhart and Klein, 2001; De Hoogh and Den Hartog, 2009), and here we investigate the role of Machiavellianism on employees' reactions to ethical leadership. This study contributes specifically to research on ethical leadership and Machiavellianism by showing that ethical leader behavior is suitable for countering antisocial behavioral tendencies in a group of employees (high-Machs) that bears a high risk of engaging in unethical behaviors (e.g., Dahling et al., 2012). Finally, by studying the effect of employee Machiavellianism on their behavioral reactions to ethical leader behaviors, we provide empirical support for scholars who argue that even high-Mach employees do not always engage in unethical behaviors and are also able to show cooperative behavior (e.g., Wilson et al., 1996; Kessler et al., 2010).

# Machiavellianism in Organizations

In the psychological literature (e.g., Christie and Geis, 1970; Jones and Paulhus, 2009), Machiavellianism is defined as a personality trait that refers to "a strategy of social conduct that involves manipulating others for personal gain, often against the other's self-interest" (Wilson et al., 1996, p. 285). It is regarded as a quantitative trait which implies that all individuals may show manipulative behavior at times, but some may be prone to showing such behavior more often than others. High-Mach individuals are characterized by a specific constellation of characteristics which can be summarized by (a) a strong goal focus and (b) the willingness to use all possible means to achieve their goals.

High-Machs show a strong goal focus and stress achievement and winning (Jones and Paulhus, 2009). This goal focus motivates them to use all possible means to achieve their ends ("winning

above all") and ultimately allows high-Machs to show high performance especially if given the opportunity to manipulate and bend rules (Shultz, 1993; Bagozzi et al., 2013), even under conditions of constrained access to resources (Kuyumcu and Dahling, 2014). Supervisors, however, usually evaluate high-Mach employees less positively than low-Machs (Ricks and Fraedrich, 1999). Thus, while the unmitigated use of all means to achieve their ends helps high-Machs to achieve high performance or other goals, it often negatively affects their evaluations by others at the same time, at least in the long run (see Jones and Paulhus, 2009).

High-Machs' willingness to deploy antisocial and unethical strategies can be explained by several mechanisms. First, Machiavellian individuals have a cynical, negative worldview, always expecting the worst from other people (Christie and Geis, 1970). This provides them with a justification for showing unethical behavior, "others would have acted similarly." Consistently, high-Machs trusted others less in economic situations than low-Machs (Sakalaki et al., 2007). At the same time, high-Machs are emotionally detached from their own actions, allowing them to engage in unethical behaviors without experiencing negative feelings like guilt or remorse (e.g., McHoskey et al., 1998; Wastell and Booth, 2003). The regulatory social function of (self-conscious) negative emotions is thus not equally strongly available to Machiavellians as it is to those low on Mach (Bagozzi et al., 2013, provide a neurological explanation for this deficit). Finally, Machiavellianism comes with a strong self-focus and egoism (Fehr et al., 1992) resulting in a lack of attachment and commitment toward others or the organization (Zettler et al., 2007). Consistently, McLeod and Genereux (2008) note that high-Machs only lie if they profit, not if others profit (i.e., no "white lies").

The mentioned characteristics of Machiavellianism all provide explanations for high-Machs' low threshold to engage in unethical behaviors, even when being antisocial and (potentially) harming others, and their lack of willingness to engage in behavior that benefits others if it not also clearly benefits them. In line with the arguments above, high-Mach individuals tend to show a number of unethical and counterproductive work behaviors (see Dahling et al., 2009, 2012). For instance, high-Machs are found to lie and deceive others (e.g., Williams et al., 2010), steal (e.g., Harrell and Hartnagel, 1976), defect during bargaining (Gunnthorsdottir et al., 2002), engage in sabotage (e.g., McLeod and Genereux, 2008), and use emotional manipulation (e.g., Austin et al., 2007). Some studies report that they engage in less helping behaviors (Wolfson, 1981; Becker and O'Hair, 2007), while other studies (e.g., Dahling et al., 2009; Bagozzi et al., 2013) report a non-significant relationship of Mach with OCB, which suggests that moderating variables might play a role here.

Studies on Machiavellianism and leadership are scarce. The limited research available on Machiavellian leaders suggests that high-Mach leaders stimulate less positive responses in their followers than low-Mach leaders (e.g., Den Hartog and Belschak, 2012; Belschak et al., 2016). They are also more often perceived as abusive leaders by their followers (Kiazad et al., 2010). Yet, research has also found that high-Mach leaders can be seen as determined and charismatic by followers (Deluga, 2001), and are able to increase employee engagement when showing ethical leader behavior (even though their effect was less strong than when low-Machs engaged in ethical leader behaviors; Den Hartog and Belschak, 2012). This demonstrates high-Mach leaders' ability to adapt their behavior to the situation despite of being detached from their followers' interpersonal concerns (Deluga, 2001; Dahling et al., 2009).

Even fewer studies than on Machiavellian leaders have been conducted on leading Machiavellian employees, and thus the effects of different leadership styles on Machiavellian employees have not received much attention to date. Noteworthy exceptions are the studies by Belschak et al. (2015) who have investigated the reactions of Machiavellian employees to transformational leaders, by Belschak et al. (2016) who explored the effects of high-Mach leaders on high- versus low-Mach followers, and by Wisse et al. (2015) who address the role of all three dark triad traits in leaders and followers. Here, we add to this stream of research by testing the effects of ethical leadership on high-Mach versus low-Mach employees' ethical (affiliative OCB) and unethical work behavior (knowledge hiding and (emotional) manipulation). To our knowledge, research has not yet explored which leadership style might be effective in reducing high-Machs' highly undesirable tendency to engage in unethical work behaviors.

# Ethical Leadership and Machiavellian Employees

Ethical leadership can be defined as "the demonstration of normatively appropriate conduct through personal actions and interpersonal relationships, and the promotion of such conduct to followers through two-way communication, reinforcement, and decision-making" (Brown et al., 2005, p. 120). Past research on ethical leadership has shown that such leaders foster their followers' ethical behavior and decrease their unethical behavior (e.g., Brown et al., 2005; Den Hartog and De Hoogh, 2009; Piccolo et al., 2010; Kalshoven et al., 2011b) and has been linked specifically to OCB (e.g., Mayer et al., 2009; Den Hartog and Belschak, 2012).

Ethical leaders are value driven and act in line with their principles (Brown and Treviño, 2006). They stress the importance of fair, moral, and ethical behavior and the avoidance of unethical behavior, and they live up to the values they espouse (Den Hartog, 2015). Ethical leaders act as role models of ethical behavior and stimulate ethical behavior and conduct by rewarding (ethical employee behavior) and punishing (unethical employee behavior) of their followers. They send strong and clear signals to their employees that ethical behavior is desirable and will be noticed and rewarded while unethical behaviors are undesirable and will be punished when detected. In contrast, leaders low on ethical leadership do not signal and model the importance of integrity and ethical conduct, and do not monitor for or use rewards or sanctioning to stimulate such conduct.

As noted, high-Mach employees are self-centered and goaldriven, and they are thus likely to be more sensitive than low-Machs to messages about what type of behavior is likely to result in the highest rewards for them and will adapt their own behavior

accordingly. For example, Wilson et al. (1996, p. 287) describe "Machiavellianism as a kind of master strategy that includes both cooperative and defecting substrategies, plus a system of rules for when to use them." Similarly, Kessler et al. (2010) note that high-Machs can use manipulation and deceit but can also be genuinely accommodating and respectful, depending on what seems most advantageous for achieving their goals in a given situation. While low-Machs may generally show more ethical behavior than high-Machs, high-Machs may be more sensitive to cues from the environment about which behaviors are rewarded.

High-Machs have a strong preference for money and power (Stewart and Stewart, 2006; Sakalaki et al., 2007) suggesting that they strongly value the extrinsic motivational aspects of their work (e.g., promotions, status, power, and money). We therefore expect that followers will show increased ethical forms of behavior under ethical leaders given that this behavior is clearly expected, monitored for, and rewarded by the leader, and that this positive relationship will even be stronger for high-Machs than for low-Machs due to high-Machs' strong goal orientation and their sensitivity to rewards (Jones and Paulhus, 2009; Kessler et al., 2010). Also, ethical leaders' own ethical behavior sends a signal to employees that such behavior will facilitate achieving a leadership position in the organization, encouraging high-Machs who strongly value positions of status and power to engage in vicarious learning and copy such ethical behavior. In contrast, low ethical leaders do not expect or monitor for ethical behavior and may send the signal that "anything goes." Under such leaders, we expect that high-Machs do not show increased ethical behaviors and rather engage more in unethical means to reach their goals, including particularly deception and manipulation.

Scholars have argued that a communal and people orientation (showing respect, supporting and helping others) is an essential part of ethical leadership (e.g., Treviño et al., 2003; Kalshoven et al., 2011b; see also Den Hartog, 2015). This implies ethical leaders will emphasize the importance of and reward showing affiliative behavior. High-Machs are more sensitive to such rewards than low-Machs and are therefore likely to show increased affiliative behavior only when ethical leadership is high, not under low ethical leadership as such leaders do not emphasize the importance nor reward employees for supporting and helping colleagues, and helping others is not something that high-Machs would typically do if they did not explicitly expect to be rewarded for it (Dahling et al., 2009). We therefore hypothesize the following.

Hypothesis 1. Machiavellianism and ethical leadership will have interactive effects on affiliative OCB, such that the relationship between Machiavellianism and affiliative OCB will be more positive under highly ethical leaders than under low ethical leadership.

While generally high-Machs will show more unethical behavior than low-Machs, unethical behaviors by high-Mach employees should strongly decrease under ethical leaders. These leaders monitor follower behaviors on an ethical dimension, communicate clearly that unethical behaviors are not acceptable, and punish such behaviors when detected. This active monitoring decreases high-Machs' room to maneuver and signals that unethical behavior is likely not to lead to reward but to punishment. As noted, high-Machs adapt their behavior to the situation and do not always engage in unethical behaviors; in particular, they do not show manipulation and deception if it is not advantageous or might even be detrimental for achieving their goals (Wilson et al., 1996; Kessler et al., 2010). Here, we investigate two specific types of manipulation and deception behavior, namely, knowledge hiding (e.g., Webster et al., 2008; Connelly et al., 2012) and (emotional) manipulation (Austin et al., 2007). Both of these behaviors can be labeled unethical behaviors. In this respect, Gini (1998) argues that, in order to act ethically, individuals need to consider and respect the interests and rights of all affected parties in their behaviors. Yet, when hiding knowledge from others, the knowledge hider accepts that the interests of others might be harmed due to a lack of information. Emotional manipulation refers to the instrumental use of reading and managing others' emotions to suit one's interests, even against the interests of others (Austin et al., 2007). Such behaviors are in conflict with being a "moral person" who carefully considers the consequences of one's actions (cf. Treviño et al., 2000). Both knowledge hiding and emotional manipulation thus refer to behaviors that ignore and neglect others' needs or interests and may even go against those needs in order to maximize satisfaction of one's own (or one's own group's) interests and can thus be considered as unethical (see Den Hartog, 2015).

Knowledge hiding refers to employees' efforts to withhold or conceal knowledge from colleagues rather than share it, even if that knowledge is useful for or needed by them (e.g., Connelly et al., 2012; Connelly and Zweig, 2015). Knowledge hiding is hence the opposite of sharing knowledge with and helping colleagues and forms the antisocial, unethical counterpart of prosocial affiliative OCB as it refers to an active and intentional attempt of employees to hide their knowledge from colleagues.

As noted, ethical leaders advocate communal and peopleoriented behaviors (see Den Hartog, 2015), and knowledge and information sharing has been identified as part of ethical leadership (Kalshoven et al., 2011b). In line with this, followers of ethical leaders were found to show increased knowledge sharing (Ma et al., 2013). Thus, we argue that, under highly ethical leaders, high-Machs' adaptivity and goal focus will not only lead to increased affiliative OCB but also to decreased knowledge hiding activities compared to low-Machs who are more likely to already be willing to share knowledge regardless of their leader's behavior, as high-Machs likely perceive that under ethical leaders, who monitor them, knowledge hiding will be easily discovered and is detrimental for their career (Connelly et al., 2012). In contrast, leaders low on ethical leadership do not emphasize and monitor employee behavior on a moral dimension and thus are more likely to give the signal to high-Machs that they do not need to pay attention to ethical behaviors but can cut corners and deceive others without being punished. Given high-Machs predisposition to fall back on unethical behaviors (Jones and Paulhus, 2009), we thus expect their knowledge hiding to increase when ethical leadership is low compared to when it is high. We therefore hypothesize the following.

Hypothesis 2. Machiavellianism and ethical leadership will have interactive effects on knowledge hiding, such that the relationship between Machiavellianism and knowledge hiding will be less positive under highly ethical leaders than under low ethical leadership.

While knowledge hiding refers to an unethical behavior targeting specifically colleagues (Connelly et al., 2012), we argue that high-Machs' tendencies to engage in unethical behaviors will generalize and also show in other social contexts (Christie and Geis, 1970; Jones and Paulhus, 2009). We therefore investigate a second unethical, antisocial behavior aimed at a different target, namely the use of manipulative behavior toward leaders. The use of manipulation is one of the defining characteristics of Machiavellianism (Christie and Geis, 1970) and an important part of measures of Machiavellianism (cf. Christie and Geis, 1970; Dahling et al., 2009; Kessler et al., 2010). Emotional manipulation is defined as manipulating others' emotions within a self-serving framework (e.g., Grieve and Mahar, 2010) and has been positively linked to both Machiavellianism (Austin et al., 2007) and psychopathy (Grieve and Mahar, 2010), potentially because it is an effective but more covert type of manipulation compared to other manipulative behaviors (e.g., lying, providing false information). Due to this reduced risk of discovery and the power differential between leaders and followers, emotional manipulation seems a type of manipulation particularly suitable to be used by followers with their leaders.

Ethical leaders emphasize fairness, are trustworthy and honest, advocate integrity, and communicate the importance of such behaviors to employees (Brown et al., 2005; Kalshoven et al., 2011b). Ethical leaders thus do not use manipulation themselves and clearly signal to employees that manipulative behavior is not acceptable and will lead to negative consequences (see Den Hartog, 2015). As noted above, we argue that high-Mach employees are particularly sensitive to their leaders' signals and expectations about desirable behaviors (e.g., Kessler et al., 2010) and will therefore avoid (or at least reduce) the use of manipulative behaviors under highly ethical leaders. Leaders low on ethical leadership, in contrast, do not discuss or model ethical behavior nor do they monitor or punish (un)ethical employee behaviors, and high-Machs should therefore more freely engage in emotional manipulation under such leaders. We therefore expect the following.

Hypothesis 3. Machiavellianism and ethical leadership will have interactive effects on emotional manipulation, such that the relationship between Machiavellianism and emotional manipulation will be less positive under highly ethical leaders than under low ethical leadership.

### MATERIALS AND METHODS

### Procedure and Sample

We tested the three moderation hypotheses presented above in a multi-source survey study among 159 unique employeesupervisor dyads in Netherlands. We used business school contacts to get access to organizations and asked these organizations whether they would be willing to participate in a study on leadership in organizations and its impact on employees. The organization had to provide contact information of one of their employees and his/her supervisor. We then sent employee and supervisor a paper-and-pencil version of the survey by email accompanied by a letter explaining the purpose and purely academic nature of the study and the voluntary and confidential nature of participation. Respondents did not receive anything in return for participating in the study. After having read this information, respondents filled in the survey. The study was carried out in accordance with the recommendations of the Economics and Business Ethics Committee, University of Amsterdam, who approved the protocol for the study (request nr 20171124121141). In total, we sent out 240 employee–supervisor surveys, and we received 159 employee–supervisor dyads back, resulting in a response rate of 66%. Surveys were administered in Dutch. All survey scales came from validated measures and were carefully translated and back-translated by native speakers, respecting the norms of the International Test Commission.

Respondents worked in a wide range of industries including health services, IT, architecture, account management, consultancy, education, and financial services. Of the participating employees, 37% were male and 63% female. The mean age of the employees was 34.98 years (SD = 13.36), and the average tenure at their current organization was 6.80 years (SD = 8.85). In total, 40% of the employees had attained a university (master's) degree. Of the participating supervisors, 57% were male and 43% female. The mean age of the supervisors was 42.23 years (SD = 12.15); their mean organizational tenure was 10.27 years (SD = 9.18). Supervisors had worked with the participating employee together for 3.21 years on average (SD = 3.73); 45% of the supervisors held a university master's degree.

### Measures

Employees rated their own degree of Machiavellianism, their supervisors' ethical leader behaviors, and their own knowledge hiding behaviors toward their colleagues and emotional manipulation toward their supervisor. Supervisors rated their employees' affiliative OCB. All responses were measured on a seven-point Likert scale ranging from 1 ("completely disagree") to 7 ("completely agree").

Employee Machiavellianism was measured with eight items from the Mach-IV scale by Christie and Geis (1970) which is still the most widely used measure in studies on Machiavellianism. This Dutch eight-item short measure of Machiavellianism was used successfully in several recent studies in the Netherlands (e.g., Den Hartog and Belschak, 2012; Belschak et al., 2015, 2016). Sample items are "It is wise to flatter important people" and "Never tell anyone the real reason you did something unless it is useful to do so." Cronbach's alpha of the scale was 0.80.

Employees' perception of their leaders' ethical leadership was measured with the oft-used 10-item scale by Brown et al. (2005). This measure is well validated and was used in the Dutch context successfully before (e.g., Kalshoven et al., 2011a,b; Den Hartog and Belschak, 2012). Sample items are "My leader discusses business ethics or values with employees," "sets an example

of how to do things the right way in terms of ethics," or "disciplines employees who violate ethical standards." Cronbach's alpha was 0.84.

Due to the conceptual overlap (e.g., caring about others, acting as role models for followers; see Brown and Treviño, 2006) and the substantial empirical correlations usually found between ethical and transformational leadership (e.g., Brown et al., 2005; Toor and Ofori, 2009; see Ng and Feldman, 2015), leadership scholars have emphasized the need to control for transformational leadership in studies regarding ethical leadership (see Den Hartog, 2015). We therefore also included transformational leadership in our survey and used the 11-item measure of the Dutch "Charismatic Leadership in Organizations (CLIO)" questionnaire to measure employees' perception of their leaders' transformational leadership (e.g., "My leader has a clear vision and an image of the future" and "stimulates subordinates to think independently"). This Dutch measure covers content similar to other measures of transformational leadership like the MLQ (e.g., Bass and Avolio, 1990; House, 1998). It is well validated and has been used in several leadership studies in the Netherlands before (e.g., De Hoogh et al., 2004, 2005; De Hoogh and Den Hartog, 2009). Cronbach's alpha was 0.90.

Affiliative OCB was measured with seven items by MacKenzie et al. (1991). The items cover the helping and the courtesy dimension of this widely used multi-dimensional measure of OCB. Sample items are "This employee is always willing to help the people around him/her" and "considers the impact of his/her actions on others." Cronbach's alpha was 0.84.

The knowledge hiding scale is a relatively new measure which was first introduced by Connelly et al. (2012). We used seven items of this measure capturing all different strategies of knowledge hiding (playing dumb, evasive hiding, and rationalized hiding). Sample items read "When a colleague recently asked for information I agreed to help the colleague but provided different information than the requested one" and "I pretended that I did not know the information." Cronbach's alpha was 0.86.

The emotional manipulation measure was also relatively recently developed and introduced to the literature (see Kessler et al., 2010) and was taken from Austin et al. (2007). It consists of five items. Sample items are "I used my emotional skills to make my supervisor feel guilty" and "I made my supervisor feel uneasy." Cronbach's alpha was 0.92.

### RESULTS

We conducted a confirmatory factor analysis to test the factor structure and the convergent and discriminatory validity of our scales. Statisticians have noted that a prerequisite for reliable results of a CFA is a satisfactory indicator to sample ratio (see, e.g., Bentler and Chou, 1987; Bentler, 1995). Due to the relatively high number of items compared to the sample size, we therefore used a parceling approach, as recommended (e.g., Bagozzi and Heatherton, 1994). For building the parcels, we followed a factorial algorithm by combining items into parcels according to the factor loadings of the items (e.g., Little et al., 2002; Rogers and Schmitt, 2004). To minimize loss of information, we only built parcels for the two longer and wellestablished leadership scales, and parcels consisted only of two items (and one parcel of three items in case of transformational leadership due to the uneven number of items). The CFA showed a satisfactory fit of the hypothesized six-factor structure (i.e., employee Mach, ethical leadership, transformational leadership, employee affiliative OCB, employee knowledge hiding, and employee emotional manipulation): χ 2 (614) = 942.95 (p = 0.00); CFI = 0.90; IFI = 0.90; RMSEA = 0.06. Factor loadings were satisfactory ranging from 0.45 to 0.75 for employee Mach, from 0.53 to 0.82 for ethical leadership, from 0.73 to 0.89 for transformational leadership, from 0.56 to 0.72 for affiliative OCB, from 0.42 to 0.89 for knowledge hiding, and from 0.73 to 0.94 for emotional manipulation. Factor inter-correlations ranged from −0.33 (ethical leadership and emotional manipulation) to 0.78 (ethical leadership and transformational leadership).

While one of our dependent variables was rated by leaders (affiliative OCB), the other dependent variables (knowledge hiding and emotional manipulation) were measured as employee ratings and might hence be subject to common source bias. Such bias may inflate or deflate observed relationships between constructs (e.g., Podsakoff et al., 2003). To test for common method variance, we therefore included the same-source firstorder common method factor to the CFA. This factor was defined as having as indicators all employee-rated items, and this controls for the portion of variance attributable to obtaining all measures from a single source (see Podsakoff et al., 2003). If common source variance played a role, factor loadings and/ or intercorrelations should differ substantially for CFAs including versus not including the common method factor. A comparison of the CFAs showed that factor loadings and factor inter-correlations were almost identical in both computations, thus suggesting that common source bias might not play a substantial role in our data.

The means, standard deviations, Cronbach's alphas, and (Pearson) inter-correlations of the variables are presented in **Table 1**. Employee Mach was positively correlated with knowledge hiding (r = 0.42; p = 0.00) and emotional manipulation (r = 0.28; p = 0.00) and negatively correlated with ethical leader behavior (r = −0.20; p = 0.01). Consistent with earlier studies (e.g., Brown et al., 2005; Toor and Ofori, 2009), transformational and ethical leaderships were substantially correlated with each other (r = 0.65; p = 0.00), thus illustrating the need to simultaneously include both variables in subsequent analyses to be able to draw better conclusions about the unique effects of ethical leadership. Finally, employee affiliative OCB was significantly correlated with transformational leadership (r = 0.20; p = 0.01), but not correlated with ethical leadership, and knowledge hiding and emotional manipulation were both negatively correlated with ethical leadership (r = −0.23; p = 0.00; and r = −0.32; p = 0.00).

To test our hypotheses, we used the PROCESS macro (version 2.13.2; developed by Hayes, 2013) to conduct our analyses. More specifically, we regressed employee affiliative OCB, knowledge hiding, and emotional manipulation on employee Mach, ethical and transformational leadership, and the interaction term of employee Mach and ethical leadership. In the analyses, we used



N = 159. <sup>∗</sup>p < 0.05. ∗∗p < 0.01.

TABLE 2 | Results of the moderation analysis using the PROCESS macro.


N = 159. <sup>∗</sup>p < 0.05. ∗∗p < 0.01.

the PROCESS option to center the predictors around their respective means and based the interaction term (Mach x ethical leadership) on these mean-centered scores to ease interpretation. As research on OCB and on dark personality traits often includes demographics as control variables, we also added employee age and gender (1 = male, 2 = female; both measured as employee ratings) as well as leader age and gender (1 = male, 2 = female) and the length of the relationship between leader and employee (all three measured as leader ratings) as covariates. The results of the moderation analyses are presented in **Table 2**. Indeed most of the demographics were significantly related to our outcome variables.

Ethical leadership only had a significant main effect on emotional manipulation (B = −0.49, p = 0.00); the other main effects of ethical and transformational leadership were non-significant. More importantly though, and (mostly) in line with Hypotheses1–3, the main effects were qualified by significant interaction effects of employee Mach and ethical leadership for affiliative OCB (B = 0.16, p = 0.04), knowledge hiding (B = −0.19, p = 0.03), and emotional manipulation (B = −0.34, p = 0.00). To facilitate interpretation of these interaction effects, we plotted the relationship between employee Mach and the three outcome variables (affiliative OCB, knowledge hiding, and emotional manipulation) for high and low values of ethical leadership (**Figures 1**–**3**), while controlling for the effects of transformational leadership.

First, Mach is significantly and negatively related with affiliative OCB for low ethical leadership (B = −0.21, p = 0.05) but non-significantly for high ethical leadership (B = 0.13, p = 0.22, **Figure 1**). Next, the relationship between Mach and knowledge hiding is significant and positive for low ethical leadership (B = 0.53, p = 0.00) and non-significant for highly ethical leaders (B = 0.15, p = 0.21, **Figure 2**). Finally, the relationship between Mach and emotional manipulation is also significant and positive for low ethical leadership (B = 0.52, p = 0.00) and non-significant for highly ethical leaders (B = −0.16, p = 0.24, **Figure 3**).

# DISCUSSION

High-Mach employees are a group of employees that is usually depicted as negative in the literature and sometimes even as dangerous for organizations (e.g., Dahling et al., 2009, 2012). Research has shown that high-Machs often make unethical choices and have the tendency to use manipulation and deception in social situations (e.g., Williams et al., 2010; Dahling et al., 2012). In line with this literature, we indeed found employee Mach to be significantly positively related to both hiding knowledge from colleagues and emotionally manipulating supervisors. Similarly, we replicated earlier findings that Mach is not significantly related to affiliative OCB (e.g., Dahling et al., 2009; Bagozzi et al., 2013). Yet, other authors found a negative link between Mach and OCB (e.g., Becker and O'Hair, 2007),

suggesting that moderators might play a role and explain these inconsistent results in the literature. High-Machs might only help others if they expect to receive a reward in return for their help, for instance, using OCB as an impression management tactic to receive a more positive supervisor evaluation (Becker and O'Hair, 2007). We therefore investigated the interactive effects between employee Mach and supervisors' leadership style on employee unethical behavior and OCB.

Belschak et al. (2015) have argued that leadership might offer the possibility to influence high-Mach employees' behaviors in positive ways, and specifically they show that transformational leadership can increase high-Machs' challenging OCB. However, they also caution this may not generalize to other outcomes. Building on this idea, we argued that high ethical leadership would reduce high-Machs' unethical work behaviors and increase their motivation to show affiliative OCB, whereas low ethical leadership would have the opposite effect. Indeed,

the relationship between Machiavellianism and affiliative OCB, knowledge hiding, and emotional manipulation was moderated by leaders' ethical leadership. The findings of our study show that in particular under low ethical leadership high-Machs show undesirable reactions, while Machiavellianism was not significantly linked to affiliative OCB, knowledge hiding, and emotional manipulation when ethical leadership was high. Thus, low ethical leadership seems to trigger high-Machs to engage in more unethical behavior, whereas high ethical leadership suppresses the expression of such behavior by high-Mach followers, rather than high ethical leadership explicitly stimulating ethical behavior in high-Machs. By ignoring the ethical dimension in employee behaviors and not caring about or monitoring employees' (un)ethical behavior, low ethical leaders seem to signal to their followers that it is acceptable to use unethical means and hence trigger undesirable behaviors particularly in high-Machs who have a predisposition to fall back on unethical behavior to achieve their ends.

Similarly, Greenbaum et al. (2017) found that high-Machs engage in unethical behavior under abusive supervisors and argue that abusive supervisors may provide cues that activate employees' Mach trait, stimulating the expression of traitconsistent behavior. Our findings provide further support for the concept of Mach trait activation and for the notion that high-Mach employees can at least to some extent be managed as their behavior is linked to specific leadership styles (see Wilson et al., 1996; Belschak et al., 2015).

Somewhat surprisingly, we found that employees generally showed the highest affiliative OCB under low ethical leadership. A potential explanation of this unexpected finding is that colleagues may compensate for a lack of people-oriented leader behavior in a team. If followers are facing a lack of guidance, support, and help from their leader (i.e., low ethical leadership), they might look for and receive help from their colleagues who fall in and compensate for their leader's deficiency. A similar compensatory model has been reported for perceived

organizational support and perceived supervisor support (Maertz et al., 2007). Future research should further investigate this compensation mechanism in which followers step in for their leader and help each other where the leader fails to support them.

Literature on knowledge hiding has argued that such behavior harms the organization and thus, in turn, the knowledge hider him/herself (e.g., less money available for financial bonuses due to reduced work unit performance; e.g., Evans et al., 2015). If showing strategic and calculated behavior, high-Machs should thus avoid such behavior as they ultimately would also suffer themselves from its negative consequences. Yet, our findings show that knowledge hiding is strongly positively linked to Mach, despite of its potential for longer term detrimental effects. In this respect, the literature on Mach suggests that high-Machs might not adapt their behavior to potential longer term indirect effects (see Wilson et al., 1996). In game theoretical experiments, high-Machs aim for short-term profit maximization (e.g., Sakalaki et al., 2007) and easily change groups if needed (see Wilson et al., 1996); they thus seem more likely to strive for instant gratification than delayed rewards (see Christie and Geis, 1970). While knowledge hiding might harm the company in the long run, in the short run, it provides high-Machs with a source of power (cf. French and Raven, 1959) and status, hence giving them the opportunity to outperform others and achieve other external rewards (e.g., a bonus or promotion; Webster et al., 2008). Overall, the findings thus indicate that high-Mach individuals prioritize short-term profit maximization over long-term profit maximization, which would be of interest to test in future research.

Despite of the increased risk of targeting supervisors with unethical behavior, our results show that high-Machs not only engage in knowledge hiding toward colleagues but also in emotional manipulation toward their supervisors, in particular for supervisors low on ethical leadership. High-Machs' tendency to use unethical behaviors when they have sufficient room to maneuver and the ethicality of their actions is not closely monitored thus seems to generalize to a broad range of manipulation and deception behaviors and to different targets. This result resonates with the results of an earlier study (Austin et al., 2007) which also found a positive link between emotional manipulation and Mach and extends it by introducing a contingency variable, (low) ethical leadership. While Mach was uncorrelated (Kessler et al., 2010) or even negatively linked with emotional intelligence in earlier studies (Austin et al., 2007), Bagozzi et al. (2013) found in fMRI studies evidence that high-Machs seem to use (non-conscious) emotional resonance processes which might allow them to "intuitively" feel and manipulate others' emotions. Future research should further investigate the link between Mach and emotional manipulation and its underlying mechanisms.

A strength of our study is that we controlled for transformational leadership. Ethical leadership shows similarities with transformational leadership (e.g., the strong value orientation; see Brown and Treviño, 2006), and correlations between the two constructs are usually high (see Ng and Feldman, 2015). It is therefore important to control for transformational leadership in empirical studies on ethical leadership to be able to determine the variance explained uniquely by each construct (see Den Hartog, 2015).

# Practical Implications

Our findings offer several practical implications. First, high-Mach employees should be managed carefully. Our results show that high-Machs are sensitive to the behavior of their leaders and adapt their behaviors to leaders who emphasize and reward certain practices. Yet, our study also suggests that leadership effects seem to be limited to very specific employee behaviors. Leaders thus need to be clear and explicit to high-Mach employees about employee practices that are acceptable and those that are not. For instance, transformational leaders' emphasis on change stimulates change-related behaviors like challenging OCB in high-Machs (Belschak et al., 2015), whereas ethical leaders' focus on ethical behavior motivates them to avoid unethical work behaviors. In this respect, organizations are also well advised to introduce (ethical) organizational values and policies to communicate acceptable and desirable employee behaviors. Developing reward systems that clearly reward ethical behavior and punish unethical behavior could further help in establishing such norms and values.

Also, high-Machs seem to perceive a lack of specification of desirable behaviors as a signal that all means are acceptable to reach their goals and hence easily engage in unethical and other organizationally undesirable behaviors. High-Machs therefore form a group of employees that are particularly in need of guidance by leaders. While passive leadership generally comes with negative employee reactions in terms of increased incivility (Harold and Holtz, 2015), a lack of leadership seems to lead to even more pronounced effects for high-Machs who strongly fall back on unethical work behaviors that are particularly damaging to the organization.

Ethical leadership seems especially suitable to counter high-Machs' tendency to engage in unethical work behaviors, and a lack of such leadership can be easily interpreted by high-Machs as a signal that "anything goes." Fortunately, ethical leadership can be combined with other leadership styles like transformational or transactional leadership (see Den Hartog, 2015). It therefore seems good advice for leaders to always show ethical leader behaviors when high-Mach followers are part of their work unit. To suppress unethical behavior from these employees, organizations should therefore offer leadership training for leaders that particularly focuses on ethical leadership (emphasis on ethical behavior) and transactional leadership aspects (systematic use of monitoring, rewards, and punishments).

Finally, organizations might consider to include measures of Machiavellianism or (ethical) values in their personnel selection procedures. While measures of Mach are generally valid, it might be difficult though to measure high-Machs' true personality during selection as this group of individuals is likely to manipulate their answers in socially desirable ways in situations in which they perceive the outcome may depend on a specific type of answer. In this sense, organizations might rather want to rely on long-term experiences of colleagues and supervisors to identify high-Machs and carefully consider this information in

promotion decisions to avoid that high-Machs rise into higher management positions in the organizational hierarchy.

### Limitations

As most studies, this study also suffers from a number of limitations. First, we used a cross-sectional design and collected our data at one moment of time, and therefore we cannot make claims about the direction of causality. Experimental research is warranted to establish the direction of causality. For instance, work units that are characterized by highly unethical practices and an "all means are acceptable to meet one's targets" employee attitude might be appealing to and attract in particular low ethical leaders. In this sense, it would also be interesting to investigate our topic from a longitudinal perspective and explore how processes unfold over time.

Second, we investigated only a limited range of employee behaviors. We have focused our study specifically on one ethical and two unethical behaviors, and our results suggest that high-Machs react clearly differently to leaders on these behaviors than on challenging OCB, as found in earlier studies. Thus, future research should further investigate and specify the different types of work behaviors that high-Machs adapt as a reaction to a specific leadership style. This would also be helpful for offering further advice to practitioners on how to manage high-Mach employees.

Also, our sample is not representative for the population of Dutch organizations, and there might be differences in ethical values across different industries that may have affected our findings. However, respondents in the study came from a broad range of different industries and organizations with no single industry being substantially more strongly represented than the others, which makes it unlikely that ethical values are systematically biased in any specific direction in our sample. Future research should consider and control for potential industry-related differences in ethical values and norms.

Finally, while we measured affiliative OCB as a supervisor rating and hence from a different source than the other variables in our study, employee Mach, leadership styles, knowledge hiding, and emotional manipulation were measured as employee ratings, which comes with the risk of common source variance. While our test for the effects of common source variance did not provide any evidence that common source bias may have affected our results, we cannot exclude this possibility. However, the main contribution of our study lies in the investigation of the interactive effects of employee Machiavellianism and leadership behaviors, and scholars have noted that analyses including interaction terms do not suffer from inflated interaction effects

### REFERENCES


due to common method bias; rather, measurement error reduces the probability to find significant interactions (e.g., Busemeyer and Jones, 1983; Siemsen et al., 2010). Also, it is difficult to measure variables such as personality traits and covert, deceptive behaviors such as knowledge hiding and emotional manipulation through other ratings. In particular, there are currently no wellvalidated non-self-reported measures of Mach available. Future studies might collect leadership data from other sources though (e.g., colleague ratings) or develop other-rated measures of Mach thus including even more different data sources.

# CONCLUSION

This study investigated the link between leadership, employee Machiavellianism, and ethical (affiliative OCB) as well as unethical employee work behavior (knowledge hiding and emotional manipulation). We found that the relationship between Machiavellianism and these behaviors was strongly influenced by leaders' ethical leadership style. Employee Machiavellianism came with reduced affiliative OCB and increased knowledge hiding and emotional manipulation, but only when ethical leadership was low. More research is warranted in the area of Machiavellians' reactions to different leadership styles in order to help managing this group of organizational members.

# ETHICS STATEMENT

All subjects gave written informed consent in accordance with the Declaration of Helsinki. More specifically, at the beginning of the survey, respondents were informed about the content and purpose of the study; the academic nature of the study, i.e., the study was conducted by a university for research purposes; the voluntary nature of participation, i.e., participants did not receive anything in return for participation and participants were free to not respond to any question; and the fully anonymous nature of the study, i.e., it is impossible to identify either individual respondents or participating organizations.

# AUTHOR CONTRIBUTIONS

FB, DDH, and ADH conceived and developed the project. FB and ADH coordinated and conducted data collection. All authors contributed to all parts of the manuscript, agreed to all aspects of the work, and approved the final version of the manuscript.




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Belschak, Den Hartog and De Hoogh. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# 'Killing Me Softly With His/Her Song': How Leaders Dismantle Followers' Sense of Work Meaningfulness

Petra Kipfelsberger<sup>1</sup> \* and Ronit Kark<sup>2</sup>

1 Institute for Leadership and Human Resource Management, University of St. Gallen, St. Gallen, Switzerland, <sup>2</sup> Department of Psychology, Bar-Ilan University, Ramat Gan, Israel

Leaders influence followers' meaning and play a key role in shaping their employees' experience of work meaningfulness. While the dominant perspective in theory and in empirical work focuses on the positive influence of leaders on followers' work meaningfulness, our conceptual model explores conditions in which leaders may harm followers' sense of meaning. We introduce six types of conditions: leaders' personality traits, leaders' behaviors, the relationship between leader and follower, followers' attributions, followers' characteristics, and job design under which leaders' meaning making efforts might harm or 'kill' followers' sense of work meaningfulness. Accordingly, we explore how these conditions may interact with leaders' meaning making efforts to lower levels of followers' sense of meaning, and in turn, lead to negative personal outcomes (cynicism, lower well-being, and disengagement), as well as negative organizational outcomes (corrosive organizational energy, higher turnover rates, and lower organizational productivity). By doing so, our research extends the current literature, enabling a more comprehensive understanding of leaders' influence on followers' work meaningfulness, while considering the dark side of meaning making.

### Edited by:

Konstantinos G. Kafetsios, University of Crete, Greece

### Reviewed by:

Gro Ellen Mathisen, University of Stavanger, Norway Dritjon Gruda, Maynooth University, Ireland

\*Correspondence: Petra Kipfelsberger petra.kipfelsberger@unisg.ch

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 07 December 2017 Accepted: 16 April 2018 Published: 09 May 2018

### Citation:

Kipfelsberger P and Kark R (2018) 'Killing Me Softly With His/Her Song': How Leaders Dismantle Followers' Sense of Work Meaningfulness. Front. Psychol. 9:654. doi: 10.3389/fpsyg.2018.00654 Keywords: work meaningfulness, meaning making, leadership, followership, dark triad

# INTRODUCTION

The dominant perspective on leaders' meaning making role, in both the leadership and the meaningfulness literature, focuses on leaders' positive influence on followers' work meaningfulness. This line of research has shown that transformational leadership, empowering leadership, and high-quality leader-member relationships are positively related to followers' perceptions of work meaningfulness (e.g., Piccolo and Colquitt, 2006; Arnold et al., 2007; Grant, 2012; Tummers and Knies, 2013). However, research on the dark side of leaders' meaning making is scarce (for exceptions, see Amabile and Kramer, 2012; Neal et al., 2013; Bailey and Madden, 2016). While prior research has already shown that not all leaders' efforts to infuse the work of followers with positive meaning are effective (e.g., Kirkpatrick and Locke, 1996), we know very little regarding whether and under which conditions leaders' attempts to instill work meaningfulness among followers might have detrimental effects on followers' work meaningfulness. Investigating the potentially harmful effects of leaders' meaning making is a pressing endeavor since the quest for meaningful work among many employees is increasing (Cascio, 2003). People across generations and particularly today's emerging adults [Millennials, born after 1980, also described as "generation me" by Twenge (2006)] are motivated to realize their selves at work and focus on having work opportunities that will

enhance their personal sense of meaningfulness. However, at the same time, many people lack a deeper sense of meaning at work and are searching for something that is larger than themselves and that goes beyond their ego (Lancaster and Stillman, 2010; Lepisto and Pratt, 2017). Organizations might react to the increasing need for meaningful work by intensifying leaders' meaning making efforts (Cascio, 1995; Feintzeig, 2015). Therefore, it is important to better understand how and under which conditions intense efforts of leaders' meaning making may backfire and decrease work meaningfulness among followers.

Our conceptual model introduces six types of conditions under which leaders' meaning making efforts might harm followers' meaningfulness at work: leaders' personality traits, leader behaviors, the relationship between leaders and followers, followers' attributions toward their leaders, followers' characteristics, and job design. This model makes important contributions to the literature on meaning making, work meaningfulness, and leadership in the following ways. First, our research provides a conceptualization, grounded in theory and empirical findings, of several conditions under which leaders' meaning making may have detrimental effects on followers' work meaningfulness. In such, we provide a novel perspective on the influence of leadership on meaningfulness (e.g., Shamir et al., 1993; Podolny et al., 2005). The conceptual model we develop contributes to the growing literature on the 'dark side of leadership' (e.g., Mathieu et al., 2014) and destructive leadership (e.g., Einarsen et al., 2007). Second, by incorporating the recent advancements of research on work meaningfulness (Martela and Steger, 2016; Lepisto and Pratt, 2017), we develop a more detailed understanding regarding the particular dimensions of work meaningfulness (coherence, purpose, or significance), as well as of the pathways (realization or justification perspective) that can be harmed by the interaction of leaders' meaning making and the identified conditions. Third, through revealing the factors under which leaders' meaning making might harm followers' work meaningfulness, we extend prior writings on the critical perspective on the management of meaning (Lips-Wiersma and Morris, 2009; Bailey et al., 2017). Finally, we present an agenda for future research and discuss practical implications of our conceptualization.

# Harming Followers' Sense of Meaningfulness

Studies on the cultivation of work meaningfulness have mostly focused either on work conditions (e.g., May et al., 2004; Humphrey et al., 2007), on individuals' personal behaviors (e.g., Vuori et al., 2012) that increase their sense of meaningfulness, or on a combination of both (e.g., Chalofsky and Krishna, 2009) as antecedents to work meaningfulness. Studies that have addressed leadership as an antecedent, have generally focused on how leaders help construct work meaningfulness among followers and contribute to their sense of meaning (e.g., Shamir et al., 1993; Zhang and Bartol, 2010; Tummers and Knies, 2013). There are different views of leaders' role in meaning making and how meaning making leads to meaningful work. While the top-down views (e.g., job design, leadership style and behaviors; Piccolo and Colquitt, 2006; Grant, 2012) consider leaders as the agents in meaningful work, the bottom-up view (e.g., job crafting; Wrzesniewski and Dutton, 2001) treats employees as the agents that construct their own sense of meaning at work. We integrate both perspectives (top–down and bottom–up; leaders as well as followers as agents) in our model. We transfer recent research advancements on meaning in life to the context of meaningful work and suggest that meaningful work consists of three components: coherence, which refers to employees' understanding and ability to make sense of what is happening at work; purpose, which refers to directionality of employees' work and the ability to connect their work to a higher-order goal; and significance, which refers to employees' evaluation of their work worth (Martela and Steger, 2016).

Harming followers' work meaningfulness refers to situations in which any of the three components- coherence, purpose or significance- of followers' work is diminished. First, regarding coherence, this implies that followers struggle to understand the meaning of their work; they find their work chaotic, unstructured, do not know or lose track of what their work is all about and are unable to grasp the point of their work (lack of coherence). Second, a reduction of followers' purpose means that the experience, belief, or hope of followers that their work makes a positive difference in the world is reduced, and that they do not see a clear direction and contribution of their work. They might feel that their work is going nowhere or that their work does not serve a higher-order goal (lack of purpose). Third, regarding significance, the reduction of feelings of one's work significance refers to the worth, value, and importance of one's work. Followers might experience that their work is useless and not worthwhile. They might struggle to explain the worth of their work, be unable to justify the worth of their work, or consider the worth of their work and tasks as ambiguous. Furthermore, they may lack solid accounts for the worth of their work, or their established accounts, based on personal values and what matters for them, might have been impaired (lack of significance; Martela and Steger, 2016; Lepisto and Pratt, 2017).

While these three components of work meaningfulness are the elements that might be dismantled by leaders' meaning making, we draw on Lepisto and Pratt's (2017) recent work to explain what drives and underlies the process of meaning erosion. These scholars differentiate between two ideal-type conceptualizations of meaningful work in the vast literature: the realization and justification perspective (Lepisto and Pratt, 2017). We propose that this dual perspective on meaningful work helps shed light upon the deeper, underlying mechanisms that can be harmed by leaders' meaning making. The realization perspective of work meaningfulness refers to the idea that individuals strive to express and realize themselves through their work. According to this perspective, meaningfulness is achieved through fulfillment of motivations, desires, and needs associated with self-actualization (Lepisto and Pratt, 2017). For example, leaders that have a strong ideological base and are obsessed with their own frame(s), voice or "song" of meaning, can offer a frame of meaningfulness that will override employees' inner and authentic sense of meaningfulness. This can hinder employees' ability to realize and take ownership over their own sense of work meaning.

The justification perspective, on the other hand, refers to an individual's ability to account for one's work worth and to consider one's work as worthy (Lepisto and Pratt, 2017). For example, leaders that use incoherent accounts, suggest unethical purposes, or rely on unethical means to mobilize followers, can distort followers' foundation for justifying the worth of their work toward others or for considering their work worthwhile for themselves.

In the following sections, we elaborate in depth upon each of the circumstances that we propose might harm followers' sense of meaning at work, discuss the particular component of meaningful work that might be negatively affected and refer to the underlying mechanism of realization and/or justification. We provide propositions for each condition. **Figure 1** depicts the conceptual model.

# COMPONENTS OF THE CONCEPTUAL MODEL

# Leaders' Personality Traits: The Dark Triad

We begin with leaders' personality traits and focus specifically on the dark triad because we expect a particularly large negative effect of these traits on followers' meaningfulness. All three personality dimensions– namely, psychopathy, narcissism, and Machiavellianism – entail a socially malevolent and rather insincere character and the behavioral tendencies to promote oneself, while interacting with others in an aggressive and emotionally cold way (Paulhus and Williams, 2002). More specifically, psychopaths lack a conscience, and are therefore not bothered by and do not try to change their bad and hurtful behavior (Tamayo and Raymond, 1977; Babiak and Hare, 2006). In fact, they have been shown to get an exciting thrill from hurting others (Clarke, 2009). They also lack emotions (Nadis, 1995; Stout, 2005), do not consider others' pain when making moral judgments (Blair et al., 1995), and are not sensitive to criticism. Narcissists are primarily driven by self-enhancement and need constant external self-affirmation. Although they do not intend to harm others, they ignore others' welfare (Braun, 2017). Machiavellians, though they do not lack a conscience, they do lack concern for conventional morality, ignore interpersonal affect, have low ideological commitment, and are quite adapt at manipulating others and are willing to do so through all means (McHoskey et al., 1998).

Though psychopathy, narcissism, and Machiavellianism are distinct constructs, all are characterized by a high degree of selfishness and a willingness to put one's own needs ahead of others (O'Boyle et al., 2012). We argue that the combination of leaders' meaning making and any one of the dark triad traits might reduce followers' work meaningfulness due to the overly self-focused and overly socially dominant characteristics of such leaders (Rosenthal and Pittinsky, 2006), which might overshadow followers' ability to develop or follow their own inner sense of meaning. In fact, such self-focused personalities might expect unquestioning obedience (O'Connor et al., 1996) as well as full identification and compliance with their sense of meaning from their followers (Jones et al., 2004; Vaknin, 2009). Their strong attachment to their self-promoting motives is very likely to hamper or dismantle employees' inner sense of meaning. Besides their desire for compliance, these personalities are also associated with a lack of empathy, which is the inability to understand and share the feelings of others. Their lack of empathy makes it difficult or impossible to promote the meaning making of their followers because they do not know and value the needs, desires, and feelings of their employees.

Beyond the theoretical links between dark triadic leaders' characteristics and their effects on followers' work meaning, such leaders' behaviors might affect and interfere with the various components of work meaningfulness (coherence, purpose, and significance). In terms of coherence, there is evidence that corporate psychopaths engage in extreme forms of mismanagement, characterized by poor personnel management, directionless leadership, and mismanagement of resources (Babiak and Hare, 2006; Boddy et al., 2015). In addition, employees working under corporate psychopaths receive less instruction, less training, less help and they experience more unfairness from their supervisors (Boddy, 2010). This sort of chaotic and precarious managing style is antithetical to clarity, and to the ability to make sense of one's work and work environment and can therefore directly obstruct employees' sense of coherence.

In terms of purpose, empirical studies of corporate psychopaths have shown that they create a toxic work environment, characterized by conflict, bullying, increased workload, low levels of job satisfaction, and unnecessary organizational constraints (Babiak and Hare, 2006; Boddy, 2010). Moreover, dark triad leaders are motivated by personal gain and self-promotion (Jakobwitz and Egan, 2006; MacNeil and Holden, 2006) rather than that of their organization, employees, or other stakeholders (see Boddy, 2006 on corporate psychopaths; Holian, 2006 on narcissists) or greater society. For example, a prior study suggests that there is a negative relationship between corporate psychopaths in organizations and employees' perceptions of corporate social responsibility (Boddy et al., 2010). Such lack of a higher-order prosocial goal might reduce followers' feelings of purpose. Moreover, employees of managers possessing dark triad traits may come to dread their work due to adverse experiences at their workplace, leading them to overlook any purpose their work might serve. Indeed, all three traits have been associated with bullying behavior (Baughman et al., 2012), with the use of manipulation tactics (Jonason et al., 2012), and have been linked to counterproductive work behavior (O'Boyle et al., 2012). Therefore, it is reasonable to assume that managers possessing any one of these traits create unpleasant work environments for their employees, void of safety, integrity, and pleasantness, which may overshadow employees' ability to view their work positively and as serving a higher purpose.

In terms of significance, there is evidence that leaders with dark triad traits do not provide followers with grounds for feeling that their work is appreciated, valued and important but rather cause them to feel quite the opposite. Research has shown that employees were significantly less likely to feel that they receive

recognition for doing a good job, that their work is appreciated, and that they were properly rewarded for their efforts, when corporate psychopaths hold leadership positions (Boddy et al., 2010). The dominance and obsession by the self-serving motives of such leaders might degrade followers as human beings, their ideas, and their contributions to the collective. Research has shown that leaders' narcissism lowers employees' self-esteem and, in turn, their level of creativity (Eissa et al., 2017). Hence, we argue that followers who are treated in an instrumental manner and used to serve the leader's interest, might not only feel deliberately misused and devalued but also may feel disenabled to realize their self or their interests at work and therefore their sense of significance at work will be lowered.

Proposition 1: Leaders' dark triad personality traits (narcissism, psychopathy, and Machiavellianism) will moderate the effect leaders' meaning making has on followers' work meaningfulness, such that this relationship is reduced if the leader scores high on any of the dark triad traits.

# Leaders' Behavior: Rigidity and Inflexibility

There are multiple leadership behaviors that can reduce followers' sense of meaning. Here, as a central example, we focus on behaviors that highlight leaders' rigid and inflexible behaviors. Several facets of inflexibility and rigidity that pertain to leadership are introduced in the organizational literature (Good and Sharma, 2010). In our conceptual model, we discuss two specific facets that we estimate to moderate the effects of leaders' meaning making attempts on followers' sense of meaning. We argue that leaders with low cognitive flexibility and low communication flexibility – accumulating in rigid behaviors of leaders – can hurt followers' sense of work meaningfulness, through leaders' inability and unwillingness to adjust to and incorporate new situations (those pertaining to followers, in particular).

Cognitive flexibility is defined as the ability to shift attention in order to respond to the environment in a new way (Good and Sharma, 2010). Leaders low on cognitive flexibility struggle to overcome their fixed mental schemas and are unable to come up with situation appropriate responses. Cognitive inflexibility is therefore closely related to rigidity, which has been defined as "the inability to produce novel or changed responses" (Vacchiano et al., 1969, p. 268). Consequently, such leaders are "closedminded," inflexible, rigid, and do not adjust their behavior according to situational demands but rather adhere to their way of seeing the world and acting upon it.

Another aspect of flexibility that can affect followers' sense of meaning is communication flexibility, which has been defined as the ability to generate and select communicative options according to the needs of the situation (Martin and Rubin, 1994). This form of flexibility has been proposed to be important for communicating continuously shifting goals and expectations in the dynamic environments in which leaders operate. In other words, leaders are often required to change their verbal and nonverbal communicative behaviors according to the situation and audience (Stevens and Campion, 1994). Indeed, studies show that leaders' choice of words, symbols, and expressions influence the degree to which the audience becomes inspired, aroused, and committed (Awamleh and Gardner, 1999).

Taken together, if leaders are low on cognitive flexibility thus, unable to shift the work goals and tasks according to the

situational demands— and low on communication flexibility thus, unable to communicate clear expectations, goals, and tasks— they may weaken followers' sense of coherence, purpose, and significance in the following ways. Regarding followers' coherence, if employees do not understand what the leader is trying to convey due to leaders' very limited range of words and phrases, overly abstract or overly detailed speech (low communication flexibility), followers' ability to understand and make sense of their work will be compromised. In addition, if the leader has low cognitive flexibility, the leader may not be able to change the meaning they provide for work according to their audience. In these situations, the leader is more likely to distort followers' ways of realizing their selves at work and their ways of justifying the worth of their work rather than facilitate these. Followers are likely to be left with an unstructured, chaotic picture of their work and might struggle to make sense of their work based on leaders' thoughts and words.

Even worse, regarding followers' work significance, cognitively inflexible leaders might be so rigid in their opinions and evaluations that they squelch followers' sense of meaningfulness of and at their work as they provide too little space for followers to self-realize and to express themselves. Metaphorically speaking, 'his/her personal song is so loud,' pervasive, and insistent that followers' 'song' is not heard or that followers stop singing. In addition, lack of cognitive flexibility implies an inability to engage in perspective taking. Lacking the ability to see things from the other's point of view and to integrate another's perspective, makes it difficult for a leader to tap into the values of their followers. In addition, as such leaders overlook what is important to their followers and only emphasize their own view and values, followers' feelings about their self-worth and the worth of their work are likely to be reduced.

Taking these characteristics one step further and considering the dimension of purpose, cognitive and behavioral rigidity is a characteristic of the ideological or dogmatic leader who holds strong personal convictions and values that mostly refer to the past: "The ideological leader, moreover, will justify actions based on a limited number of relatively inflexible core beliefs and values. Appeal to others will be based, not on the leader per se, but rather the truth embedded in these beliefs and values" (Mumford and Strange, 2013, p. 133). In addition, ideological leaders have been described as extremely focused on the past, which they oftentimes idealize (Mumford, 2006), so that their mental models are not constructed around future goals but rather upon goals that have served them well in the past (Mumford et al., 2008). While followers might be looking for something valuable to contribute to in the future and might wish to pursue different goals than their leaders, who glorify the past and their personal (at times limited) thinking, followers' sense of purpose is likely to be diminished by leaders' rigid ideological past-oriented convictions.

Overall, the inflexible and rigid thinking and behaviors of leaders is likely to hurt followers' potential for self-realization at work. Followers may be provided with too little space to discover and express their true selves at work, be intimidated and miniaturized when leaders are so small-minded, fixated, and focused on their personal truth and convictions. Besides hurting their self-realization, followers' justification base might be diminished if only leaders' thoughts and words are allowed to be used to justify the worth of their own work. The narrowmindedness of leaders might also limit followers' ability to think more openly and flexibly in the long-term hurting the likelihood of them finding salient accounts for the worth of their work on their own in the long-term.

Proposition 2: Leaders' behavior will moderate the effect of leaders' meaning making on followers' sense of meaning such that followers' sense of meaning will be lowest, if leaders behave in rigid ways (low cognitive and communication flexibility and strong adherence to their own ideological convictions).

# Leader–Follower Relationship: Low Level of Trust

The relationship between leaders and followers is likely to affect the ways in which followers make sense of how leaders interpret situations and craft meaningfulness. One of the central elements of a relationship is the level of trust among leaders and their followers (Barnard, 1938). We suggest that leaders' meaning making might reduce followers' sense of work meaningfulness if followers have a low level of trust in their leaders. Followers' trust in their leaders depends on their perceptions of leaders' ability, integrity, and benevolence (Mayer et al., 1995). Leaders' ability refers to their capabilities, competence, experience, skills, and qualifications and to their professional behavior related to their role as leaders (Lapidot et al., 2007). Perception of leaders' integrity refers to followers' perceptions that their leader sticks to a set of values and principles, that are acceptable to them (Mayer et al., 1995) and includes leaders' behavior that displays honesty, loyalty, and taking responsibility (Lapidot et al., 2007). Perceptions of leaders' benevolence refers to followers' belief that their leader has good intentions to contribute to followers' wellbeing (Mayer et al., 1995) including support, caring, and encouragement (Lapidot et al., 2007).

If leaders are seen by followers as low on their ability to perform their role and lead the organization to reach its expectations, followers' sense of meaningfulness might decrease for several reasons. First, if leaders are unable to effectively coordinate and orchestrate followers' activities, followers' work coherence is likely to be compromised. If a leader's ability is lacking to the extent that s/he cannot effectively direct employees toward completing tasks and reaching goals, then followers' sense that their work has the potential to make a meaningful contribution and hence, its worth, are likely to be reduced. Second, when leaders are unable to effectively communicate, instruct and manage complexity, followers' work coherence, and, in turn, significance, are likely to be harmed. If followers do not comprehend the meaning of their work and find coherence, they are quite unlikely to consider their work as important and significant (Martela and Steger, 2016). Third, leaders' ability to make decisions and their comfort in and after doing so might also have crucial influence on followers' sense of meaning, based on cross-cultural research on the differentiating factors of leaders' effectiveness (Brodbeck et al., 2000). If leaders are indecisive, feel uncomfortable and unconfident in making decisions (Brodbeck et al., 2000), followers might not only struggle to trust leaders'

decisions but also might not see a clear course of action, worry about wasted efforts and resources, so that their sense of coherence and purpose is threatened and reduced. Research has indicated that leaders' indecisiveness might evoke employees' anxieties of being 'adrift in the ocean' (Mulki et al., 2012), which might be directly linked with followers' lower work coherence and, hence, indirectly also hurt their feeling of purpose. If leaders do not make decisions, or if they tend to regret their decisions immediately after making them, or take them back frequently (Amabile and Kramer, 2012), followers might have a hard time understanding how their work activities fit together (coherence) and how they might contribute to the broader picture (purpose), as their leader takes no clear course of action.

Concerning integrity, the second major aspect of trust, followers may not believe the leaders' construction and framing of the meaning of their mutual work, if they perceive their leader as being low on integrity. If leaders do not serve as role models, do not 'practice what they preach,' do not keep their promises and do not tell the truth (Cha and Edmondson, 2006), followers might question leaders' meaning making attempts, and find it difficult to embrace their messages regarding the purpose and significance of their work. Leaders' inconsistencies, such as changing their messages too often or provision of contradicting messages, not only thwart work coherence, but also reduce followers' perceptions of leaders' credibility, including that of the messages they provide regarding work significance and purpose, and thus impair followers' belief in their work's meaningfulness.

Regarding benevolence, the third component of trust, when followers question leaders' benevolence, meaning they do not believe their leader has their best interest at heart, they might be inclined to reject the leader's attempts to tap into their self-realization goals and justifications, as they may view such attempts as insincere acts brought on by ulterior motives. Worse, if they think their work serves a malevolent agenda of their leader, they may view their work as such that yields negative consequences, which is likely to negatively affect their perspective regarding their work as positive and significant.

In addition to a low level of trust in one's own leader, a general distrust toward all leaders, based on previous personal experience with other leaders (i.e., role-based trust, Kramer, 1999) or bias stemming from a perceived prevalence of managerial misconduct (often reported by the media; Greve et al., 2010), might also negatively influence the degree to which followers trust their leaders, including what these leaders say, preach, or do. Thus, we propose that:

Proposition 3: Followers' trust in their leaders will moderate the effect of leaders' meaning making on followers' sense of meaning such that followers' sense of meaning will be reduced the lower their trust in their leaders is (e.g., lowest if they lack trust in their leaders).

# Followers' Attributions Toward Leaders' Intentions

Above, we focused on the characteristics of leaders, their behaviors, and their relationships with followers. While these actual characteristics interact with leaders' ability to foster or harm a sense of meaning among followers, another key aspect, which needs to be taken into consideration, is what followers ascribe and attribute to their leaders. A major stream in the leadership literature highlights the crucial role of followers' attributions toward their leaders in shaping how leaders' behaviors and intentions are understood and evaluated by followers, as well as how they affect the leader-follower interaction (e.g., Ferris et al., 1995; Martinko et al., 2007). For example, the romance of leadership phenomenon, suggested by Meindl et al. (1985), captures a phenomenon showing that extreme performance outcomes (very poor or very high outcomes) is often attributed to the effects of leadership, although in fact the resulting performance might be due to the situation, such as market development. Furthermore, extant research showed that the way followers perceive the intentionality of leaders' behaviors influenced their interpretations of leaders' actions as well as the followers' reactions (Ferris et al., 1995). Although leaders might have good intentions when they try to infuse meaningfulness into the workplace, the effects of their attempts interact with the attributions of followers, namely how they understand and interpret this behavior. In the following, we focus on the situations, in which followers perceive leaders' intentions as self-serving and manipulative or as unethical, because both might negatively affect followers' subsequent work meaningfulness.

If followers attribute leaders' meaning making to leaders' self-serving motives, their sense of meaningfulness might be distorted. Leaders, who are perceived as aiming to influence followers for their own self-interests, such as solely increasing their own power, prestige, or advancing their own career, might reduce followers' ability and willingness to relate their own work activities to a higher-order purpose. Feelings of 'I am being used' (Dienesch and Liden, 1986) might arise among followers and lower their self-esteem, self-confidence and, importantly, their ability to realize their selves at work. Even if followers enjoyed their work initially, the feeling that their talents, efforts, and potentials are being used for leaders' self-serving purposes might harm their willingness to fully engage in their work and undermine their answer to the question "does my work reflect and fulfill who I am" (Lepisto and Pratt, 2017, p. 111). Furthermore, followers are likely to experience anger and disappointment if their sole answer to the question of "why their work is worthy" is because it helps their leader get ahead or gain more benefits (salary, prestige, etc.). This is because people across different cultures and professions are motivated by the need to make a difference to others (Schwartz and Bardi, 2001), which likely does not mean serving leaders' egocentric or profane interests. While leaders might vary in the degree to which they are able to hide their self-serving motives, followers' work meaningfulness might be even more negatively impacted if leaders try to manipulate followers by their meaning making attempts. Leaders' manipulation has been shown to hurt the leader-follower relationship and followers' outcomes (Lin et al., 2016). Although the leader might have collective intentions in mind and refer to collective motives (e.g., a higher-order purpose), their meaning making behavior might backfire if followers attribute hypocrisy to it. A hypocritical leader

violates and only pretends to care about the company's values but in fact has self-serving intentions since hypocrisy is a deliberate violation of (espoused) values (House and Howell, 1992; Cha and Edmondson, 2006).

Another possible attribution of followers toward leader behaviors refers to unethical motives, that is, to intentions that are morally unacceptable to the larger community (Jones, 1991) and that violate moral norms (cf. Kaptein, 2008). If leaders make meaning and are perceived by followers as such that rely on unethical motives, followers might experience a moral dilemma because they are supposed to support their leader and to contribute to the leader's success through their work, while at the same time they experience that their leader represents and/or engages in unethical issues. The unethicality of leaders might particularly hurt followers' base of justifying the worth of their work and their experience of their own work's worth. If followers feel that part or all their work activities violate morale standards or norms, they might have a hard time viewing their work as significant and positive. The experience of meaningfulness is socially constructed and determined (Wrzesniewski et al., 2003; Lepisto and Pratt, 2017). Hence, if leaders' meaning making is altogether or partly morally inacceptable to the larger community, the accounts that the leader provides, might also be morally unacceptable for their followers and prevent them from developing and maintaining a positive work identity. Moreover, followers' attributions of leaders' motives as unethical might also diminish their ability and willingness to realize their (full) selves at work. The violation of moral standards might motivate followers to (emotionally) distance themselves from their leader and/or their work instead of unleashing their potential to fully engage in their work. Such distancing – necessary for followers to not blindly comply with leaders' unethicality – is likely to reduce their level of purpose and significance.

Proposition 4: Followers' attributions toward leaders' meaning making attempts will affect the way they interpret and internalize a sense of work meaningfulness, such that when followers perceive leaders a) as manipulative and self-serving, or b) unethical, leaders' conveyed sense of meaningfulness will lead to lower levels of followers' work meaningfulness.

### Followers' Characteristics

We also consider the characteristics of followers as conditions that might influence the effects of leaders' meaning making attempts on followers' sense of meaning. In particular, we focus on followers' personal values and their extent of misfit with leaders' values as well as on followers' emotional state. We argue that the combination of leaders' meaning making and the misfit between leaders' and followers' values will reduce followers' level of meaningfulness because the values prescribed by leaders as important might not matter or might be less meaningful to followers and hence, will reduce the level of their work significance. Researchers studying incidents in which followers lost their sense of meaningfulness discovered that employees related to events of leaders' behavior that were disconnecting them from their own set of values (Bailey and Madden, 2016). If followers value something different than what their leader values, their ability to realize their self through work and what matters to them are distorted. Feelings of self-alienation among followers whose values are in conflict with leaders' meaning making might increase. Notably, the extent of value misfit among leaders and followers might play an important role. While a small incongruence could even be beneficial in order to expand followers' perspective and to transcend their own values and interests, which is a fundamental aspect of transformational leadership (Sosik, 2005; Brown and Treviño, 2009), a fundamental conflict between their values and the values the leader promotes – be it the leader's own personal values or organizational values – might undermine the experience of followers' work meaningfulness.

Besides the threat to one's self-realization, the justification of meaning might also suffer in case of a value misfit because followers need anchors, such as values, to competently justify the worth of their work (Sennett, 2006; Lepisto and Pratt, 2017). If the leader offers accounts for the worth of one's work, but the follower does not buy into due to the value incongruence or value conflict, the individual's sense of coherence and purpose might be crumbled. The reason for the reduction of coherence is that followers, who have a basic or solid understanding of how their work activities yield a holistic entity, might be irritated by a different perspective that is constantly suggested by the leader that is incongruent with their personal values. Furthermore, followers' ability to experience a higher-order purpose might also be harmed if the leader preaches a higher-order purpose (based on personal and/or organizational values) that fundamentally conflicts with the followers' values. Followers, who are unable to internalize the higher-order purpose proposed by the leader, due to the value conflict, might struggle to understand the worth of their work. Together, both perspectives of realization and justification suggest that followers' misfit with leaders' values is a factor that is likely to yield negative effects for leaders' meaning making on followers.

While the misfit with leaders' values is deeply rooted in followers' more stable personal characteristics, there may also be emotional components of followers, which are likely to interact with leaders' meaning making processes. For example, if followers are emotionally exhausted by their work due to overwhelming intense situations (e.g., conflicts with customers; Schaufeli and Bakker, 2004), they might enter survival operational mode; meaning, they may try to solve immediate problems and to get things done (Kahn et al., 1964; Edwards, 1992). Accordingly, we argue that this type of followers' mode might clash with leaders' meaning making attempts so that it reduces followers' sense of meaning.

We assume that leaders' meaning making entails an idealized, promotion-oriented focus. The leader might talk about the bigger picture and articulate ultimate aspirations (Kark and Van Dijk, 2007). If such idealized messages, aimed at inspiring followers, meet with followers' situation of survival mode and emotional exhaustion, said followers might feel uneasy with the leaders' messages due to the discrepancy between the leader's and followers' emotional level (Damen et al., 2008). While leaders' emotional level might be intense and positive, followers' emotional level in such a prevention survival mode would

be of low intensity and high negativity (Feldman Barrett and Russell, 1999). We expect that such a clash might harm followers' experience of purpose and significance. First, while the purpose of followers in such a state is to solve the immediate, concrete, short-term problems, leaders' purpose would be abstract and long-term (Kark and Van Dijk, 2007). If followers cannot see any connection between leaders' higher-order abstract meaning provision and the ways they themselves construe their immediate situation, their sense of purpose of their current work might be distorted (Carton, 2017). Second, whereas the follower in such a state might consider solving the immediate problems as worthwhile and significant, the leader may undermine followers' feelings of significance by referring to an abstract, idealistic, and positive future. Leaders, who overlook the precarious and stressful situation of their followers and focus on ultimate, higherorder, abstract aspirations, are likely to harm followers' emotional and evaluative baseline. Overall, leaders' ignorance of followers' emotional situation and survival mode, is likely to interact with their meaning making process and reduce followers' sense of meaning at work.

Proposition 5: Followers' characteristics (misfit between followers' and leaders' values as well as followers' emotional exhaustion) will moderate the effect of leaders' meaning making attempts on followers' sense of work meaningfulness, such that (a) followers' experience of a misfit between their own and leaders' values, or (b) followers' emotional exhaustion, will reduce this relationship.

### Job Design: Irrelevant and Disconnected

Another way through which leaders influence followers' meaning is through structuring the work of their followers and hence, through job design (Hackman and Oldham, 1980). This mode of influence has also been referred to as indirect leadership (Kerr and Jermier, 1978). Amabile and Kramer (2011) describe three ways in which managers unwittingly drain work of its meaning through job design. First, managers may destroy employees' sense of ownership of their work, for example, by frequent and abrupt reassignments. Second, managers may convey the message that the work employees are doing will never see the light of the day, for example, by constantly changing their priorities or their minds. Third, leaders may forget or neglect to inform employees about unexpected changes in a customer's priorities so that followers' work efforts lose their relevance. All three ways refer to some form of disconnection between followers' work, their self, and their ability to make a meaningful contribution (Amabile and Kramer, 2011; Bailey and Madden, 2016). The frequent and abrupt reassignments might hurt followers' sense of meaning most if leaders do not provide a logical reason for changing followers' work activities. Consequently, followers' sense of coherence might be distorted because followers might struggle to understand the logic behind leaders' decision and to see how their piece of work fits into the bigger picture (Heintzelman and King, 2014). Beyond that, the lack of comprehensibility and coherence is likely to harm followers' sense that their work has significance and serves a purpose (Martela and Steger, 2016). Furthermore, leaders who structure or delegate tasks so that they never can be finished, make it difficult for followers to derive meaning from work. Even if the followers' assignments bear inherent significance and value to them, if followers feel their work will "never see the light of day" then effectively, their work is futile and meaningless. Without a final product or any sense that the process has reached an endpoint, there are far less opportunities for the follower to receive positive feedback on their work, which could reinforce its worth (Grant, 2007). If leaders, customers, or any other beneficiaries, that followers' work is supposed to serve, change their priorities, followers' work loses its relevance. In such cases, followers are likely to become frustrated and upset by the loss of their work significance (Chadi et al., 2017).

The aforementioned negative effects might even be exacerbated when leaders try to inspire followers with some higher-order goal, while simultaneously giving them ever changing, non-achievable tasks. Consider a leader who paints a picture of an idealized future and of how the organization will make a difference in the world, but at the same time randomly re-assigns tasks or gives irrelevant tasks (McGregor and Little, 1998). Followers might become very frustrated, as there is great dissonance between what they do and can achieve in reality and what they are being inspired to achieve and contribute to (Carton, 2017). Their aspiration might become very distant, unreachable, and even ridiculous through the huge experienced contrast between reality (which is full of irrelevant, mismanaged tasks) and their aspirations, such as changing the world (Schwarz and Bless, 1992). In addition, the lack of consistency between the goals set by leaders and their actions might cause followers to question the goals themselves as they grow increasingly skeptical of their leaders' judgment due to their poor business running and work structuring capabilities. The inconsistency experienced by followers might cause them to question why they are doing their work or why they are there (Pratt and Ashforth, 2003), signaling that they are uncertain about the purpose and significance of their work and their contribution.

Disconnecting followers from products and results is not the only form of disconnection a leader can bring about to impair followers' sense of meaningfulness. Sometimes leaders create interpersonal isolation or distance among their followers or between themselves and their followers. Both create disconnection from supportive relationships, which might harm followers' work meaningfulness (Rosso et al., 2010; Bailey and Madden, 2016). Research on relational job design has shown that interactions with others at work, such as leaders, coworkers, or customers helps to experience the impact and purpose of one's work (Grant, 2007). However, if leaders create an isolating work environment, in which there is little to no interpersonal interaction with others or in which the leader does not engage in supporting behaviors, followers might experience feelings of isolation and loneliness. Such a lack of interpersonal connectedness might thereby reduce followers' self and collective efficacy beliefs, due to decreased levels of social identification (Kark et al., 2003). If leaders talk about a high-level aspiration while neglecting to provide some type of support or additional resources to followers, followers' sense of meaning might be reduced by the felt impossibility to take a meaningful step toward the higher

goal. Leaders' espoused high-level aspiration might oppress followers, when they feel isolated and receive limited or no social support.

Proposition 6: Job design aspects will moderate the effect of leaders' meaning making on followers' sense of meaning, such that followers' sense of meaning will be reduced, when the job is designed as irrelevant and disconnected from others.

# Follower-Related Consequences of Reduced Work Meaningfulness

We argue that the decrease of followers' work meaningfulness will have negative consequences for followers' cognitions, well-being, and behaviors. In particular, we expect that followers, whose sense of meaningfulness has been reduced, will show a heightened level of cynicism. Cynicism has been defined as both a general and specific attitude, characterized by frustration, hopelessness, and disillusionment, as well as contempt toward and distrust of a person, group, ideology, social convention, or institution (Andersson, 1996). In the present case, we expect that followers will experience frustration, hopelessness, and disillusionment because their high expectations of realizing their self through work might have been disenabled or prevented by their leaders.

We also expect that a lower level of work meaningfulness will lead to a lower level of well-being, including heightened levels of negative affect, distress, exhaustion and depression among followers and reduced levels of vitality and thriving (Sonnentag, 2015). While prior research has shown that helping others at work or through one's work increases one's positive affect (e.g., Sonnentag and Grant, 2012), we expect that a low level of meaningfulness at work increases one's negative affect and lowers followers' overall well-being based on the frustration of both meaning pathways (lack of realization and justification). On the one hand, followers' limited possibilities to realize their self and their potential at work is likely to lead to negative affect, distress, and exhaustion, which might potentially culminate into depression or sickness. Followers are likely to be disappointed and frustrated, when they gain awareness of the gap between their actual self at work, which performs activities with low impact on others or with low personal significance, and their ideal self at work, which would express their true self or grow personally (Higgins, 1987; Carver and Scheier, 1990). On the other hand, followers' lack of meaningful accounts to explain and justify the value of their work might reduce their well-being and their level of experienced vitality and thriving. Followers, who question why their work is meaningful for themselves and/or for others, and thus, cannot competently justify the worth of their work (Lepisto and Pratt, 2017), might feel useless, distressed, irritated, or depressed. Hence, it is quite likely that these followers score lower on well-being.

Moreover, we expect that followers' reduced work meaningfulness influences followers' behaviors; in particular, we expect a higher level of disengagement. Disengagement is defined as the uncoupling of selves from their work roles, which means that followers are physically uninvolved in tasks, and that role demands guide their task behavior (Kahn, 1990). When followers' sense of meaning at work is reduced, meaning they consider their work as less personally significant and worthwhile, it is likely that such followers reduce their efforts to fulfill their job and work activities.

Proposition 7: Followers' reduced sense of meaningfulness will (a) enhance followers' cynicism, (b) reduce followers' well-being, and (c) enhance followers' disengagement.

# Organizational-Level Consequences of Reduced Work Meaningfulness

Furthermore, we suggest that followers' decreased levels of work meaningfulness will have organizational-level consequences. First, we argue that the level of corrosive organizational energy increases when followers' work meaningfulness decreases. Corrosive organizational energy describes the level of shared destructive energy within the organization characterized by aggression and destructive behavior (Bruch and Ghoshal, 2003). Importantly, corrosive energy entails self-reinforcing negativity (Bruch and Ghoshal, 2003), which means that if some followers are frustrated by their work and/or their leader, these followers are likely to infect other organizational members with their negative feelings, thoughts, and behaviors. It is likely that the experience of frustration and cynicism of followers not only results in disengagement but also evokes destructive behaviors against the organization creating a toxic, destructive, and corrosive work environment.

Additionally, based on previous insights, we expect that reduced levels of work meaningfulness leads to higher turnover rates in organizations. For example, Lancaster and Stillman (2010) cited the absence of meaningfulness as a key reason for turnover, and the meta-analysis of Humphrey et al. (2007) on work design features provided evidence that there is a negative linkage between experienced work meaningfulness and turnover intentions. We argue that if organizations and leaders cannot meet followers' desire for meaningful work but rather reduce followers' feelings of meaning at work, followers are very likely to leave the organization.

Moreover, we expect that a decrease in followers' work meaningfulness leads to lower organizational productivity. If followers do not fully engage in their tasks, they are less productive, which is likely to yield an overall lower level of organizational productivity (Harter et al., 2002). In addition, due to followers' lowered meaning coherence, which implies that followers have only a fuzzy shared mental understanding of the overall purpose of their work (Carton et al., 2014), their forms of cooperation and team collaborative work might be less effective, so that organizational productivity is reduced.

Proposition 8: Followers' reduced sense of meaningfulness will (a) increase corrosive organizational energy, (b) increase turnover rates, and (c) decrease organizational productivity.

# DISCUSSION

According to Shamir et al. (1993) identity motivational theory of leadership, followers' sense of meaning and their need to find meaning is a significant aspect of their organizational life. Our

conceptual model brings forth a set of different conditions under which leaders' meaning making can have detrimental effects on followers' work meaningfulness. We suggest different aspects that can interact with leaders' provision of a sense of meaning and lead to reduced meaning and eventually bring negative consequences for followers and organizations.

The theory of positive organizational psychology highlights the role of positive upward emotional spirals in organizations (Fredrickson and Joiner, 2002) as a power that feeds growth and advancement. Scholars in this field also highlight the importance of relationships and interactions that are "life giving" (Dutton and Heaphy, 2003; Kark, 2012). However, leaders' and leader-follower dynamics can actually lead to negative downward spirals and to interactions that are "life depleting," when leaders harm followers' sense of meaningfulness. This spiral can become contagious and lead to further negative effects in the organizational life.

Harming one's sense of meaningfulness may be easier and more common than enhancing followers' sense of meaningfulness. This is due to a negativity bias that is evident in numerous psychological phenomena (Baumeister et al., 2001). These works show that events with a negative valence (e.g., losing capital, breaking up with a friend, and being criticized) will have a stronger and longer lasting influence on people than similar events that have a positive valence (e.g., gaining capital, making friends, and receiving positive feedback). This was named the negativity bias (Rozin and Royzman, 2001) and the asymmetry effect (Peeters, 2002). With regards to emotions and possibly work meaningfulness, the effects of negative affect at work and within organizational life are stronger and more memorable and detailed than positive affect (e.g., Kaplan et al., 2009; van Kleef, 2009; George, 2011). This has also been demonstrated for the effect of leadership behavior (Dasborough and Ashkanasy, 2002; Sy et al., 2005; Kark et al., 2017). For example, Kark et al. (2017) found that it is easier for leaders to hinder creative behaviors of followers than to encourage such behaviors, since people are more attentive to negative versus positive leadership behaviors and signals to the prevention versus the promotion self-regulatory focus. When leaders are monitoring and looking for mistakes and exceptions, and when they elicit a self-regulatory focus of prevention, this may have a stronger effect on hindering creativity than the effect of charismatic and transformational leadership to promote novel ideas, thinking on the ideals and, creativity. This phenomenon was acknowledged by Amabile (1998, p. 77). In her words: "When I consider all the organizations I have studied and worked with over the past 22 years, there can be no doubt: creativity gets killed much more often than it gets supported."

With regards to work meaningfulness, this implies that the model we suggest, which aims to understand how contextual characteristics in the leader–follower process may harm followers' meaningfulness, may have significant effects on the ability of managers to maintain followers' meaningfulness without harming it. Furthermore, we contribute to the leadership and followership literature that considers the 'dark side of leadership' by showing under which conditions leaders might have a negative influence on followers' motivating forces, namely meaningfulness.

By incorporating the recent advancements of research on work meaningfulness (Martela and Steger, 2016; Lepisto and Pratt, 2017), we developed a more detailed understanding regarding the particular dimensions of work meaningfulness, i.e., coherence, purpose, and significance, and the underlying pathways, i.e., the realization or justification perspective, that can be harmed by the interaction of leaders' meaning making and the identified conditions. At times when many employees search for a deeper meaning at work (Cascio, 2003; Twenge, 2006), leaders might be asked and encouraged to offer new solid accounts for the worth of employees' work. However, our model indicates that the outcome of this endeavor hinges upon many different facets including the ways followers perceive and interpret leaders' attempts and to which degree leaders manage to reach an alignment between their, the organization's, and followers' values. While recent research on respectful inquiry, a leadership technique of asking followers open questions and attentively listening to them (Van Quaquebeke and Felps, 2018), suggests that leaders might be able to incorporate followers' perspective by engaging in listening, leaders still need to provide their point of view, show their value base and take a stance, when providing accounts for the meaningfulness of work.

Moreover, our research extends prior writings on the critical perspective on the management of meaning (Lips-Wiersma and Morris, 2009; Bailey et al., 2017) by revealing the factors under which leaders' meaning making might harm followers' work meaningfulness. Although the answers to the question of whose responsibility meaningful work is (Michaelson, 2011) might vary, our research is in alignment with the scholars of the critical perspective of meaning management inasmuch as both would respond to the question that it is the responsibility of leaders not to dismantle followers' work meaningfulness.

### Research Agenda

Our conceptual model offers a wide terrain for future research. It suggests that there are multiple contextual conditions that interact with the process of meaning making to effect its outcomes. We offer various dimensions for underrating the process of work meaningfulness hindrance as a multi-focal process. In order to study this complex phenomenon, qualitative studies would be helpful at a first stage. In-depth interviews and ethnographies that explore the complex processes of meaningfulness hindrance over time will enable a more nuanced and deeper understanding of the mechanisms of this process. At a later stage, quantitative studies are warranted. Following and extending the work of Martela and Steger (2016) and Lepisto and Pratt (2017), our model proposes that leaders may harm followers' work meaningfulness by reducing their sense of coherence, purpose, and significance. This three-dimensional model of work meaningfulness needs the development of a new scale that can be used to explore how leaders contribute to followers' sense of meaningfulness. Thus, in future studies researchers can explore the ways in which the different aspects of reducing meaning affects these different components of the meaning making process. The meaningfulness scale can be used to explore the negative side looking into what extent leaders can harm meaningfulness. To be even more concrete, future studies

might rely on experience sampling designs. Followers might report their level of perceived work meaningfulness multiple times throughout a working day, while they might also report the level of perceived meaning making of leaders in combination with the attributed intentionality on leaders' behaviors (Proposition 4). Last, laboratory experiments may be used to evaluate in a better controlled environment the causality of leaders' behavior and contextual aspects and how they influence the meaning reduction process. For example, future laboratory studies might build upon existing experimental studies on meaningless tasks (Ariely et al., 2008; Chadi et al., 2017) and create jobs and scenarios in which leaders' meaning making attempts are varied in combination, for example, with the job design (Proposition 6) or with the followers' current emotional state (e.g., survival mode; Proposition 5).

Future research might also develop further conditions, which might harm followers' work meaningfulness. While we focused on leaders, followers, their relationship and job design, considering these the major building blocks of the factors harming followers' work meaningfulness, future studies might add factors to each of our categories or even suggest new categories. For example, the phenomenon of multiple team memberships (O'Leary et al., 2011) and the accompanying constant change of team constellations might imply that followers lose track of a coherent understanding of what their work is contributing to. Regarding the consequences of reduced levels of followers' work meaningfulness, future research might also investigate which implications the lowered meaningfulness has for followers' level of intrinsic motivation and behavioral coping strategies. We suspect that certain individuals might react with a general reduction of their (intrinsic) motivation, while others might engage in certain behaviors to cope with or change the situation, for example, engage in job crafting. Revealing the personal (e.g., proactive personality, socio-economic status) and contextual factors (e.g., organizational climate), which might evoke certain reactions, represents an exciting avenue for future studies. Overall, we hope that our broad framework will inspire more focused and deeper theoretical developments on different aspects of the presented model, as well as empirical studies that focus on some of our more specific propositions.

### Limitations

Although the current model offers a wide variety of contextual conditions that are likely to harm followers' sense of work meaningfulness, we could not come up with an exhaustive list of conditions that harm followers' work meaningfulness. Future studies may want to consider additional aspects of the leader-follower relationship, such as gender bias or followers' attachment orientation, leaders' abusive supervision, the financial climate, time pressure of the leaders and other aspects that may harm the meaningfulness making process. Importantly, while our model mainly considered particularly negative factors, such as leaders' dark triad traits, also seemingly positive characteristics and circumstances might undermine followers' sense of work meaningfulness. Furthermore, in our conceptual model each contextual condition is offered as a stand-alone variable. However, in the organizational life, these different aspects interact. For example, leaders who are perceived as Machiavelli may also be seen as unethical and may build a relationship with employees that is characterized by a low level of trust. Thus, future work needs to further develop the interaction among different contextual aspects, or conditions in which more than one of these contextual characteristics is evident, in order to obtain a more complex theoretical perspective. Moreover, our model is focused on the individual level. However, employees in organizations work in teams and in workgroups and their sense of meaningfulness is likely to be shaped and effected by other team members, the team environment and the heterogeneous specific relationships that different team members construct with the leader. Last, our model explores one direction of influence in which leaders shape and harm followers' sense of meaningfulness. However, followers are also active agents and they may also harm leaders' sense of work meaningfulness. Thus, future studies should also explore the opposite direction of influence or multiple and reciprocal directions of influence on reducing meaningfulness.

It is worthy to note that in most circumstances managers may be attuned to fostering work meaningfulness and may be conscious of their behaviors aimed to elicit and affect the sense of meaningfulness. However, the behaviors and conditions that reduce the sense of meaning may be more hidden and managers may act upon them without full awareness. This dynamic should be explored in future studies.

# Practical Implications

Our research makes a major step forward in better understanding the erosion of meaningful work through leaders. While an answer to the quest for meaningful work could be to increase the meaning making of leaders, our research shows that such behavior could decrease followers' work meaningfulness. Thus, our conceptual work suggests that leaders should be trained and taught in the ways they should refrain from harming employees' sense of meaningfulness. Under certain conditions of personality characteristics such as the dark triad, organizations should be attuned to better select managers that are low on such behaviors and personality structure, in order to reduce the possible negative effects of such leaders on employees' sense of meaningfulness. Furthermore, if there is a negativity bias, and an asymmetrical effect of fostering versus hindering a sense of meaningfulness, leaders should be aware of their ability to harm more than to construct a sense of meaning. This implies that managers, practitioners and HR personnel should be mindful of followers' sensitivity to this negative dynamic, and attempt to refrain from giving rise to such a dynamic that can result in the loss of followers inner meaningful 'songs.'

# AUTHOR CONTRIBUTIONS

This project was a joint effort. Both authors developed the theory and ideas and contributed to writing the manuscript.

The manuscript was written by both PK and RK, however, PK took the leading role in the writing process.

### FUNDING

This research has partly been funded by the University of St. Gallen awarded to PK (grant number: 2180285).

### REFERENCES


### ACKNOWLEDGMENTS

We would like to acknowledge the helpful feedback of Anneloes Raes on a prior version of this manuscript and the valuable support of Shulamit Zimmerman-Kalker all along the process.




**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Kipfelsberger and Kark. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# I Am as Incompetent as the Prototypical Group Member: An Investigation of Naturally Occurring Golem Effects in Work Groups

### Alex Leung\* and Thomas Sy

Leadership and Group Dynamics Lab, Department of Psychology, University of California, Riverside, Riverside, CA, United States

Over four decades, research has demonstrated Pygmalion and Galatea effects (positive expectations leading to high performance) across various settings. In contrast, research on the parallel notion of Golem effects (negative expectations leading to low performance) has been largely overlooked. This study is the first to examine the relationship between group-level Implicit Followership Theories (GIFTs) and naturally occurring Golem effects. Integrating the literature on Implicit Followership Theories, Self-fulfilling Prophecies, and Social Identity, we propose that negative GIFTs can serve as proxies of expectations for followers that trigger Golem effects in workgroups. Data from 202 followers and 101 leaders provide support for our hypothesized multilevel model, revealing a top-down relationship between negative GIFTs and follower performance through self-efficacy and effort. Findings highlight the importance of GIFTs in the Golem process, showing that followers' cognitions and behaviors are shaped by the group's prototypical attributes. Suggestions for future research are offered, including interpersonal Golem effects, negative GIFTs and negative outcomes, and influence of organizational culture.

### Edited by:

Susanne Braun, Durham University, United Kingdom

### Reviewed by:

M. Teresa Anguera, University of Barcelona, Spain Allison Duke, Lipscomb University, United States

> \*Correspondence: Alex Leung Alex.leung@email.ucr.edu

### Specialty section:

This article was submitted to Organizational Psychology, a section of the journal Frontiers in Psychology

Received: 01 November 2017 Accepted: 08 August 2018 Published: 11 September 2018

### Citation:

Leung A and Sy T (2018) I Am as Incompetent as the Prototypical Group Member: An Investigation of Naturally Occurring Golem Effects in Work Groups. Front. Psychol. 9:1581. doi: 10.3389/fpsyg.2018.01581 Keywords: golem effects, implicit followership theories, self-fulfilling prophecy, social identity, performance

# INTRODUCTION

For over five decades, research has shown the impact of individuals' expectations on organizational outcomes, such as work performance (e.g., Eden and Zuk, 1995; Whiteley et al., 2012). Since Rosenthal and Jacobson's (1968) first demonstration of expectancy effects in which individuals' positive expectations result in high performance, hundreds of studies have shown the effect of this powerful mechanism across different settings, such as education, military, and industry (McNatt, 2000). In particular, researchers have shed light on the same concept in the workplace and found that leaders' positive expectations of their followers result in greater follower performance (e.g., Eden and Ravid, 1982; Whiteley et al., 2012). Similarly, research has found that employees with more positive self-expectations tend to perform at a higher level (e.g., Eden and Zuk, 1995; McNatt and Judge, 2004). These findings suggest that expectations play an enormous role in individuals' work performance. However, expectations may not always result in positive consequences. While positive expectations may promote individuals' performance, negative expectations, on the flip

side, may hamper performance (Babad et al., 1982). In the past, researchers have primarily focused on positive expectancy effects (i.e., Pygmalion and Galatea effects) (e.g., Rosenthal and Jacobson, 1968; Eden and Zuk, 1995) and studies have yet to examine the consequences of the dark variant — Golem effects—particularly in work settings. Golem effects are a special case of self-fulfilling prophecies in which individuals' negative expectation diminishes their overall performance (Babad et al., 1982). To date, research examining Golem effects is scant, perhaps due to ethical concerns of inducing negative expectancies that may have detrimental outcomes for participants beyond the confines of the study (Oz and Eden, 1994).

The few studies that have investigated Golem effects have done so via indirect means. For example, Oz and Eden (1994) studied Golem effects by manipulating squad leaders' interpretation of low physical examination scores to counter the natural formation of negative expectations for low performing paratroopers. Specifically, squad leaders were informed that the Bar-Or test (i.e., physical examination) along with past experience in other units do not predict future performance, and those with low Bar-Or scores often perform just as well as individuals with high scores. In the control group, the researcher only described the study and no information was provided regarding predictions of future performance. Results showed that Golem effects may be restrained by changing squad leaders' interpretations of the low Bar-Or test scores. In particular, participants in the experimental group made substantial improvements on their Bar-Or test (Golem effects restrained), whereas Golem effects were retained for individuals in the control group as indicated by their persistently lower performance. In short, Oz and Eden (1994) indirectly studied Golem effects by showing that low performing individuals (control group) performed worse than an equivalent peer (experimental) group whose leaders' low expectations for them were mitigated (a process they labeled de-golemization).

More recently, research on Golem effects was investigated directly in an educational setting by inducing supporting instructors with negative expectations in a laboratory study with undergraduate participants (Reynolds, 2007). Specifically, supporting instructors were told that students were put into different conditions based on the result of a management-acumen test, though neither the students nor supporting instructors were aware of the random assignment. Upon dividing the students randomly into three groups (i.e., positive, negative, and control condition), the supporting instructors received different information regarding the students in each group. First, one support instructor was told that she had been assigned the high-performing group and that this group of students would likely perform well on subsequent tests. The second support instructor was informed that she had been given a group with low performance and that these students may perform equally poor in the subsequent assessments. Lastly, the third support instructor was given no information about the students. Results confirmed the linkage between expectations (i.e., positive and negative) and task performance. Specifically,

the change in pretest and post-test scores showed that positive expectation led to higher levels of performance, whereas negative expectation led to lower levels of performance. The latter finding reflects the ethical concerns of researchers. The potential damage that may be inflicted by artificially inducing lower expectations has deterred researchers from studying Golem effects for nearly half a century. As such, little is known about Golem effects, particularly in work settings. To date, we are not aware of any field study investigating Golem effects at work.

One solution that circumvents these ethical concerns and affords investigations into Golem effects is to study it in its natural form (Oz and Eden, 1994). Most research on expectancy effects involves the artificial manipulation of leaders' expectations for their followers. However, leaders' expectations for their followers in most work settings occur naturally, without experimental manipulation (Eden, 1990). Accordingly, we refer to naturally occurring Golem effects as negative expectancy effects that occur without any form of artificial manipulation (Whiteley et al., 2012). Recent developments on implicit followership theories (IFTs) or conceptions of followers (Sy, 2010) offer a new avenue to investigate naturally occurring Golem effects because conceptions of followers can serve as proxies of expectations for followers that trigger Golem effects. In the current study, we investigate naturally occurring Golem effects in organizational settings via IFTs. IFTs exist at both the individual and group levels (Epitropaki et al., 2013). Our focus is on groups' Implicit Followership Theories (GIFTs), specifically, the Incompetency schema—due to its direct relevance to performance. Consistent with recent research (Whiteley et al., 2012), we propose that negative schemas of GIFTs (i.e., incompetence) may serve as proxies for performance expectations that trigger naturally occurring Golem effects. Specifically, we propose that GIFTs are associated with individuals' performance via self-efficacy and effort (see **Figure 1**).

We contribute to the organizational literature in several ways. This study is the first to investigate the relationship between group Implicit Followership Theories (GIFTs) and naturally occurring Golem effects. Our study shows how negative GIFTs may serve as negative in-group expectations for followers which trigger the process of naturally occurring Golem effects, hampering their performance at work. Furthermore, GIFTs provide a new avenue for investigating self-fulfilling prophecies "in the wild" as organizational researchers studying Golem effects are often restricted by the feasibilities of naturalistic organizational settings (e.g., ethical concerns with artificially inducing negative states). We also advance the field of implicit theories by showing the relevance of IFTs in shaping employee outcomes. This insight is particularly important given the decades-long criticism that implicit theories have failed to demonstrate its practical relevance for workplace outcomes (Epitropaki et al., 2013). Finally, we integrate three distinct fields of Implicit Theories, Self-fulfilling Prophecies, and Social Identity to explain how Golem effects may form naturally, shaping followers' cognitions and behaviors negatively in everyday work settings.

# Group Level Implicit Followership Theories

Over decades, researchers have investigated individuals' conceptions of leaders, namely implicit leadership theories (e.g., Lord et al., 1984; Offermann et al., 1994). In comparison, researchers have only begun examining the parallel notion of IFTs recently (IFTs; Sy, 2010; van Gils et al., 2010; Whiteley et al., 2012; Tee et al., 2013; Steffens et al., 2016; Wang and Peng, 2016; Alipour et al., 2017).

Whereas IFTs represent conceptions of followers at the individual level (Sy, 2010), GIFTs represent parallel conceptions of followers at the group level. These conceptions are formed at an early age through socialization (Hunt et al., 1990; Antonakis and Dalgas, 2009) and continue to be developed based on interactions with others (Lord and Maher, 1991; Sy, 2010), such as others within their workgroup. GIFTs are represented in the form of prototypes (Rosch and Lloyd, 1978; Sy, 2010), which may represent ideal (i.e., how followers should be), or typical (i.e., how followers typically are) forms. GIFTs are represented by six dimensions (Sy, 2010): Industry, Good Citizen, Enthusiasm, Incompetence, Insubordination, and Conformity. These six dimensions also represent an overall positive follower prototype (Industry, Good Citizen, Enthusiasm) and an overall negative follower prototype (Incompetence, Insubordination, Conformity) (Sy, 2010). Individuals may use GIFTs as a "sensemaking" function (Weick, 1995) to interpret, understand, and respond to behaviors of their group members (Poole et al., 1989). Moreover, individuals may use GIFTs to make inferences about other followers within the same group (e.g., how similar or dissimilar when compared to the typical follower in the workgroup), which may influence how they think and behave (e.g., think and behave in accordance with the typical follower prototype).

Although individual members could vary in their conceptions of follower prototypes (i.e., IFTs), they also are likely to have shared conceptions of their group's typical follower attributes (i.e., GIFTs) due to recurrent interaction and shared experiences as members of the same group (Klein and Kozlowski, 2000). The process by which group members internalize the key attributes of their groups (i.e., GIFTs) may be explained by the Social Identity Model (e.g., Hogg, 2001; Hogg and van Knippenberg, 2003; van Knippenberg and Hogg, 2003; Giessner et al., 2009). The Social Identity Model of leadership suggests that groups are represented by prototypical attributes and research show that the individual who best represents the group's prototypicality is perceived as the leader of the group.

We propose a parallel notion whereby these prototypical attributes of the group also apply to followers, and followers who best embody these prototypical attributes are perceived as the typical or ideal followers. A fundamental assumption of the Social Identity Model is that individuals perceive peer members by benchmarking the degree to which they match the attributes of the group's leader and follower prototypes (Turner et al., 1987; Hogg, 2001). Given that members of the same workgroup are often exposed to the same information and experience, their leader and follower prototypes tend to be shared (Hogg and Terry, 2000). These shared leader and follower prototypes may influence how group members define themselves through the process of self-categorization (Turner et al., 1987). Such process is important as it determines whether a specific group member gets recognized as an in-group member or an outgroup member, with individuals who best embody the group's prototypical attributes categorized as in-group members while those who diverge from the prototype are classified as out-group members (Hogg and Terry, 2000).

Out-group members may be ostracized as deviants because they threaten the group's collective identity. To avoid ostracism, group members may internalize prototypic attributes of GIFTs (e.g., incompetence) through the process of depersonalization (self-stereotyping) because they want to be accepted by the group (Fenigstein, 1979) and avoid the detriments of being categorized as an out-group member (Hogg and Terry, 2000). For example, less competent peer members may shun and derogate a highly competent person as an overachiever and "know-itall," who violates prototypical norms for selfish gains at the expense of the group. To retain group membership and avoid ostracism, followers may act in accordance with and internalize the group's follower prototype (e.g., scaling back effort and productivity). In short, group member's personal (individual level) IFTs are likely to parallel that of GIFTs because recurrent interactions and shared experiences in the same group facilitate collective identification processes (i.e., "we" as opposed to "I")

(Tajfel and Turner, 1979; Turner, 1985). Moreover, even when individual-level IFTs conflicts with GIFTs, group members are likely to conform to GIFTs rather than relying on their own personal IFTs to the extent that they self-identify with the group and desire to maintain membership (Hogg and Reid, 2006).

Group-level Implicit Followership Theories are expected to activate corresponding behaviors due to the perception-behavior link (Bargh et al., 1996; Chen and Bargh, 1997). Research has found a link between perceptions and behaviors because just as cognitive concepts are represented mentally, so are social behavioral responses, and one is likely to activate the other via spread activation (Collins and Loftus, 1975; Dijksterhuis and Van Knippenberg, 1998). Numerous types of related mental representations and behavioral patterns are triggered when GIFTs are activated. For instance, when negative GIFTs are activated, they trigger related mental representations (e.g., "bad followers" activates the associated notion of "inexperienced followers") and behavioral patterns (e.g., low effort expenditure) that are consistent with the activated concepts. The activation of mental representations (i.e., GIFTs) increases the tendency for individuals to behave in ways that are consistent with those cognitions. Meaning, followers in groups with more negative GIFTs may think and behave more negatively than those in groups that hold less negative GIFTs (McGregor, 1960; Chen and Bargh, 1997). Hence, negative GIFTs are expected to serve as expectations for followers, triggering processes like Golem effects.

# Group's Implicit Followership Theories and Self-Efficacy

Negative GIFTs are the negative conceptions that group members have of followers (Sy, 2010). Although there may be some differences in follower prototypes across workgroups, negative GIFTs have been shown empirically to be shared across workgroups and individuals (e.g., Sy, 2010). On the basis of the perception-behavior link, negative GIFTs should negatively influence how followers feel about their capabilities, generating outcomes like Golem effects (Chen and Bargh, 1997). Concepts relating to "how group's follower prototypes are" (i.e., GIFTs) are highly related and correspond to "how prototypical I am." Given that individuals may internalize and embody the group's negative follower prototypes via the Social Identity Model explained above (Turner et al., 1987; Hogg and Reid, 2006; Hornsey, 2008), individuals who assimilate more negative GIFTs should have lower self-efficacy. That is, followers who internalize more incompetency conceptions of followership may believe they lack capabilities to perform well. Therefore, followers in groups that are exposed to more negative conceptions of followers would likely have lower self-efficacy. Accordingly, we hypothesize that:

**H1.** Negative group-level implicit followership theories (GIFTs) are negatively associated with followers' selfefficacy.

# Self-Efficacy and Effort

We expect that followers' self-efficacy will be positively related to the amount of effort they put forth. According to perceptionbehavior link (Bargh et al., 1996; Chen and Bargh, 1997), individuals who see themselves as more capable of accomplishing tasks (i.e., high self-efficacy) may put forth more effort when they encounter challenges. On the contrary, individuals who think that they are incapable (i.e., low self-efficacy) are likely to abate their effort when faced with obstacles (Bandura, 2000). Indeed, conceptual and empirical evidence have shown support for the linkage between self-efficacy and effort. Researchers have suggested that self-efficacy is key in determining whether employees' work will be initiated and how much effort will be expended (e.g., Stajkovic and Luthans, 1998), which suggest that effort is an outcome of self-efficacy (Bandura, 1997). Moreover, other researchers have found a positive association between self-efficacy and effort, such that having higher self-efficacy encourages individuals to put forth their best effort (e.g., Crawford et al., 1980; Eden, 1990). As such, we hypothesize:

**H2.** Followers' self-efficacy is positively related to their effort.

# Effort and Performance

We posit that followers' effort is positively associated with their performance. Although there are multiple factors that may influence individuals' performance, the most direct influence may stem from the individual—the amount of effort an individual is willing to put forth. As suggested by researchers, individuals' effort is one of the most common factors in influencing performance (e.g., Brown and Leigh, 1996). Moreover, numerous studies have found a positive relationship between effort and performance (Katerberg and Blau, 1983; Gardner et al., 1989; Schermerhorn et al., 1990; Brockner et al., 1992; Blau, 1993). For instance, Katerberg and Blau (1983) found that the amount of effort real estate agents put forth was related to their sales, the number of listings, and commissions. Furthermore, Blau (1993) found that bankers' effort was associated with their overall performance. Altogether, these studies provide strong support for the positive relationship between individuals' effort and performance. As such, we hypothesized that:

**H3.** Followers' effort is positively associated with their performance.

# Multilevel Mediation Through Self-Efficacy and Effort

Our model for Golem effects (see **Figure 1**) and the above hypotheses suggest a mediation effect. Consistent with our prior propositions with the Social Identity Model (Hogg, 2001; Hogg and van Knippenberg, 2003; van Knippenberg and Hogg, 2003) and perception-behavior link (Bargh et al., 1996; Chen and Bargh, 1997), we expect that group's follower prototypes (i.e., negative GIFTs) would serve as a proxy for performance expectation influencing followers' self-efficacy which in turn impact their effort and overall performance. Our model for Golem effects is labeled as a 2-1-1-1 model in which the influence of a level-2 variable (i.e., negative GIFTs) on a level-1 variable (i.e., followers' performance) is conveyed by a sequence of two level-1 variables (i.e., followers' self-efficacy and effort) (Krull and MacKinnon, 2001). Accordingly, we hypothesize:

**H4.** Negative group-level implicit followership theories (negative GIFTs) are significantly and indirectly related to followers' performance through followers' self-efficacy and effort.

# MATERIALS AND METHODS

fpsyg-09-01581 September 8, 2018 Time: 18:36 # 5

# Sample and Data Collection Procedure

A team of trained undergraduate research assistants recruited adult workgroups from their existing network of contacts. Survey data were collected from a wide range of industries (e.g., sales, entertainment, and healthcare). All workgroups consisted of one supervisor and two of his or her direct subordinates. Research assistants received approximately 1 hour of training before the data collection process. Training included discussions of ethical guidelines for the recruiting procedure (e.g., no coercion) and qualifications (e.g., working adults). The workgroup leaders and followers completed different versions of the survey. The follower variables (self-efficacy and effort) were self-reported by followers, whereas followers' performance was assessed by the workgroup leaders. Both leaders and followers rated the group level variable (i.e., GIFTs). Followers' self-reporting on self-efficacy and effort is appropriate because these variables represent individuals' intrapsychic phenomena (Harris and Rosenthal, 1985; Sy, 2010). Hence, it would be more valid to ask followers about their perceptions and behaviors rather than observers who may lack the precision in judging followers. However, this approach may raise the concern of common method bias (Podsakoff et al., 2003). Such concern will be addressed below using different approaches to show that same source ratings did not significantly bias the results.

The sample consisted of 303 participants: 101 workgroup leaders and 202 workgroup followers. Regarding workgroup followers, 58.4% were female, with mean age of 28.52 years (SD = 11.23). Followers were ethnically diverse, including Asians (34.2%), Hispanic/Latinos (25.7%), Caucasians (18.8%), African Americans (4.5%), Native Hawaiian/Pacific Islanders (0.5%), and some identified as "others" (16.3%). Regarding leaders (n = 101), 58.4% were female, with mean age of 29.55 years (SD = 11.82). Leaders were also ethnically diverse, including Asians (39.6%), Hispanic/Latinos (26.7%), Caucasians (13.9%), African Americans (3.0%), Native Hawaiian/Pacific Islanders (3.0%), and some identified as "others" (13.9%).

# MEASURES

# Group Members' Implicit Followership Theories (GIFTs)

Group members' negative conceptions of followers (GIFTs) were adapted and assessed using three negative attributes from the IFTs scale (Sy, 2010). Although implicit measures are rarely used in organizational research, a projective method was used to assess GIFTs in the current study because implicit measures tend to yield more reliable psychological construct in some instances (Roberts et al., 2006). For example, such approach allows us to capture individuals' GIFTs while avoiding potential issues with self-report methods, such as socially desirable responding (Harms and Luthans, 2012; Epitropaki et al., 2013). To assess GIFTs using the projective method, participants were provided the following instructions:

"In the following, you will see three statements describing a story. Imagine the typical members/followers in your workgroup in these stories. Your task is to invent stories for the typical members/followers in your workgroup. Please write a short story in the space provided. . .think about what led to this event, what is happening now, and the outcomes at the end. . .there are no right or wrong stories. Imagine whatever kind of story you like."

Based on the instructions, participants then were asked to invent stories about typical members/followers of their workgroup in typical scenarios at work (e.g., group member's daily experience at work). For example, one participant wrote:

"[Group member] showed up late to work, was already feeling stressed out from problems at home and is overwhelmed but decides to carry on anyways, attempting to act as if nothing is happening. Suddenly a timed order comes through the work system which he feels unable to accomplish on his own. He manages to complete his task at hand even though it did take him longer than expected. [Group member]'s supervisor reprimands him for not being able to finish the task in a timely manner like a manager should and because of such lack in performance has his schedule altered to reflect his supervisor's distrust in his abilities. Such action leads [group member] to feel unappreciated for his effort in his work environment which causes him to care less and less about his involvement and overall progress at work, leaving him frustrated even after leaving the workplace."

Using the IFTs scale, they were then asked to indicate on a 7-point scale how accurate each item described the typical group members in the stories. The IFTs scale comprised of three negative dimensions, each consisting three items. Such method allows leaders and followers to describe and assess typical member in their workgroups based on specific work-related scenarios (versus less relevant social functions). For investigating Golem effects (i.e., negative self-fulfilling prophecies) in this study, it is appropriate for us to focus on the negative dimensions of GIFTs. Specifically, we focus on the Incompetency dimension (i.e., uneducated, slow, inexperienced) because it is directly related to individuals' performance. The Incompetency dimension was constructed by aggregating the three items (i.e., uneducated, slow, inexperienced). The internal consistency coefficient for negative GIFTs was 0.89.

We conceptualize GIFTs as a compositional emergent construct of IFTs because they are measured by the same set of items and are structurally and functionally equivalent (Klein and Kozlowski, 2000). Composition emergence is germane to phenomena that progress through recurrent within-group interactions in which core group elements (i.e., cognition, perception) become shared among all members within a group (Kozlowski and Chao, 2012). Similarly, as group members interact, their conceptions of the characteristics that reflect the group's prototypical follower (i.e., GIFTs) should

converge and become shared over time. Therefore, it is appropriate for us to conceptualize GIFTs as a compositional emergent construct and calculate a single score for each workgroup.

Before calculating a single score for each workgroup by aggregating group members' ratings, we conducted analyses to justify whether there is sufficient support for both withingroup agreement and between-group variation (Klein et al., 1994). Specifically, we followed Klein and Kozlowski's (2000) recommendations to account for group-level analysis as a shared team-construct. First, we conducted a one-way analysis of variance (ANOVA) to estimate between-group variability. Results suggested significant group effects on individuals' ratings of negative GIFTs: F (100, 202) = 2.01, p < 0.01 (ICC<sup>1</sup> = 0.25 and ICC<sup>2</sup> = 0.50). In addition, we calculated rwg(j) statistics to assess the extent of within-group agreement for negative GIFTs (Biemann et al., 2012). The median within-group agreement value for negative GIFTs was considered strong (Negative GIFTs: rwg(j) = 0.84; James et al., 1984; Bliese, 2000; LeBreton and Senter, 2008). All in all, our analyses revealed that our shared group construct, negative GIFTs, have both between-group variability and within-group homogeneity.

# Self-Efficacy

We used five items adapted from Riggs and Knight's (1994) measure of self-efficacy to assess followers' self-efficacy. Workgroup followers were asked to respond to five items on a 7-point scale regarding their own self-efficacy. Example items included, "I have confidence in my ability to do my job" and "I have all the skills needed to perform my job very well." The internal consistency coefficient for the scale was 0.70.

### Effort

Workgroup followers' effort was measured with five items adapted from Brown and Leigh (1996). Followers were asked to rate on a 7-point scale the extent to which each statement describe themselves. Example items included, "I strive as hard as I can to be successful in my work." and "When I work, I really exert myself to the fullest." The internal consistency coefficient for the scale was 0.94.

### Work Performance

Leaders of each workgroup rated their followers' work performance. Followers' work performance was measured with three items using a 7-point scale that adapted from Wayne et al. (1997). Example items included, "This employee has performed his/her job well," and "In my estimation, this employee gets his/her work done very effectively." The internal consistency coefficient for the performance was 0.90.

# Controls

We controlled for participants' age as it may influence performance (e.g., Blanchard-Fields et al., 2007), and because IFTs may continue to be refined and further developed over time as individuals interact with others (Lord and Maher, 1991; Sy, 2010). We also included gender as a second control variable (1 = male, 2 = female) because previous studies have found selffulfilling prophecies to be more potent with men (McNatt, 2000). Finally, we controlled for participants' average hours worked per week because performance can be a function of effort and time spent practicing one's craft (Yeo and Neal, 2004).

# RESULTS

### Analysis

To accommodate the multi-level nature of the study, we used multilevel structural equation modeling (MSEM) to model topdown (2–1) relationships (Preacher et al., 2010). The MSEM models dismantle the variance of a variable into its latent withinunit variance and a latent between unit variance (Lüdtke et al., 2008). By dismantling variance into components at the between and within levels, MSEM avoids potential problems of conflated within and between level relationships in traditional multi-level approach (e.g., hierarchical linear modeling), allowing us to estimate indirect relationships more precisely (Zhang et al., 2009; Preacher et al., 2010).

Hypotheses 1 through 4 suggests an indirect relationship in which negative GIFTs and followers' performance are mediated by followers' self-efficacy and effort. Using MSEM, we could simultaneously evaluate the top-down relationships between (a) negative GIFTs and followers' self-efficacy, (b) the individuallevel relationship between followers' self-efficacy and effort both within- and between-group, (c) the individual-level relationship between followers' effort and performance both within- and between-group. The indirect relationship between negative GIFTs and followers' performance mediated by followers' selfefficacy and effort (Hypothesis 4) was tested using the productof-coefficients methods. Preacher et al. (2010) suggest that a level-2 variable's top-down relationship with a level-1 outcome is a between-group relationship because the level-2 variable could not predict the within-group variances among individuals in the workgroups. Therefore, we examined the coefficient for level 2 predictor (negative GIFTs) and the latent group mean of level-1 outcome (followers' performance).

The means, standard deviations, reliabilities and correlations among the study variables are presented in **Table 1**. All analyses were conducted using Mplus 7.0 (Muthén and Muthén, 2012– 2017) with robust maximum likelihood (MLR) estimation. Model fit for the 2-1-1-1 multi-level mediation was assessed using the root-mean-square error of approximation (RMSEA), the Tucker-Lewis Index (TLI), and the comparative fit index (CFI) (Hu and Bentler, 1999). To compare multilevel models, the scaled chisquare difference test (Satorra, 2000) was used for comparisons.

# Discriminant Validity of Constructs

Prior to testing the hypotheses, we conducted confirmatory factor analyses to examine the distinctiveness of the study variables. Original items were used as indicators for all measures. For the hypothesized four-factor model, results indicated that all factor loadings were significant (p < 0.001). Standardized factor loadings were on average 0.86 for negative GIFTs, 0.60 for selfefficacy, 0.87 for effort, and 0.88 for performance. In addition, results indicated that the model fit for our hypothesized fourfactor model is considered acceptable (χ 2 (98) = 262.49, N = 202;

TABLE 1 | Means, standard deviations, reliability coefficients and correlations.


Note. Internal consistency reliabilities are on the diagonal in parentheses.

∗∗p < 0.01. <sup>∗</sup>p < 0.05.

CFI = 0.92, TLI = 0.91, RMSEA = 0.09) and is better than a onefactor model (1χ 2 (6) = 880.51, N = 202; CFI = 0.51, TLI = 0.44, RMSEA = 0.22) as well as a model in which variables are loaded into three factors based on raters (1χ 2 (3) = 108.63, N = 202; CFI = 0.87, TLI = 0.85, RMSEA = 0.12) (Hu and Bentler, 1999). Hence, given the high factor loadings and the greater fit of the hypothesized four-factor mode, we conclude that our measures captured distinct constructs.

# Common Method Variance

As mentioned in the previous section, both leaders and followers participated to avoid biases from same source ratings. However, common method variance (CMV; i.e., variance as a product of the measurement method rather than the constructs of the measures) may have biased the some of the variables measured from followers' perceptions (i.e., GIFTs, self-efficacy, effort) (Podsakoff et al., 2003). To ensure CMV did not significantly bias the results of the study, we conducted a confirmatory factor analysis to assess whether a single-factor accounted for most of the covariance between the study variables. If CMV is responsible for the relationship among variables, the single-factor CFA would fit the data well (Korsgaard and Roberson, 1995; Mossholder et al., 1998). The results suggested that a single-factor model with followers' same source ratings as indicators was significantly worse fitting compared to the hypothesized model (χ 2 (N = 202, 65) = 671.51, p < 0.01; CFI = 0.63; TLI = 0.56; RMSEA = 0.22), suggesting that CMV did not bias the results of our study.

### Hypothesis Testing

The multilevel structural model, in which negative GIFTs and followers' performance are associated through followers' selfefficacy and effort, showed a good fit overall (χ 2 (4) = 1.68, p = 0.79, CFI = 1.00, TLI = 1.28, and RMSEA = 0.00). All factor loadings were significant (ps < 0.01). We tested an alternative model with a direct path from negative GIFTs to followers' performance. The added pathway did not improve the overall model fit (χ 2 (3) = 1.71, p = 0.63, CFI = 1.00, TLI = 1.21, and RMSEA = 0.00) as indicated by a non-significant scaled chi-square difference test (1χ 2 scaled (1) = 0.06, p = ns), which serves as evidence for a full mediation. Hence, we retain the hypothesized model.

Next, we examined the results of all direct and indirect relationship in our model; shown in **Table 2**. First, negative GIFTs were negatively related to followers' self-efficacy, as indicated by a significant unstandardized structural coefficient (b = −0.27, p < 0.01), which supports Hypothesis 1. Followers' self-efficacy was significantly related to followers' effort at both the within-level of analysis (b = 0.45, p < 0.01) and between-level of analysis (b = 0.78, p < 0.01), supporting Hypothesis 2. Followers' effort was found to be positively related to followers' performance as indicated by an unstandardized structural coefficient in between-level of analysis (b = 0.99, p < 0.01); however, the relationship at the within-level analysis was insignificant (b = −0.11, p = 0.07). This result suggested that individuals' effort was not related to their performance; however, groups that exhibit more effort tend to perform better at work. This result provided partial support for Hypothesis 3. As for the multi-level mediation model, negative GIFTs had a negative and statistically significant indirect relationship with follower performance, through followers' self-efficacy and effort (unstandardized estimate of the product of coefficients = −0.21, p < 0.05, 95% CI = −0.37, −0.05), supporting the mediation in Hypothesis 4.

# DISCUSSION

In this study, we examine the relationship between negative GIFTs and naturally occurring Golem effects. Specifically, we test whether negative GIFTs are associated with followers' performance through their self-efficacy and effort. GIFTs reflect the prototypical follower attributes for the group, from which followers make inferences about themselves as well as other followers. Negative GIFTs are the in-group standard for followers, which serve as negative expectations for followers to fulfill to be accepted and avoid ostracism as an out-group member (Hogg and Terry, 2000). Followers who internalize more negative prototypical follower attributes of the group viewed themselves as less capable. This, in, turn, trigger Golem effects. We find support for the hypothesized multi-level model as shown in **Figure 1**. In line with our hypotheses, we find a negative top-down relationship between negative GIFTs and followers' self-efficacy. Furthermore, followers' self-efficacy is positively related to their effort, which is related to their overall performance. Additionally, we find a negative indirect relationship between negative GIFTs and follower performance through follower self-efficacy and effort.



Note. Unstandardized estimates are reported for both direct and indirect relationships.

1 = level-1 variable; 2 = level-2 variables; CI = confidence interval.

∗∗p < 0.01. <sup>∗</sup>p < 0.05.

Our results indicate that followers' effort and performance are positively related only at the between-level of analysis. Interestingly, although this relationship at the within-level is insignificant, the marginally negative relationship between effort and performance warrants some discussion. One possibility for these results may be due to social loafing (Latané et al., 1979). While each employee may take on different roles on a team, they are likely to be working collectively on tasks or projects rather than individually. As such, it is possible that followers exert less effort while working in groups. This is supported by previous research. For example, Kerr and Bruun (1983) found that individuals working in teams tend to exert less effort when a certain performance level is reached, which may have been influenced by the norms of negative GIFTs; Individuals tend to match their group members' effort while working collectively (Jackson and Harkins, 1985). Moreover, those who attempt to exert more effort may be evaluated by their supervisors more negatively for violating norms associated with GIFTs (i.e., extra effort viewed as a selfish attempt to overachieve at the expense of less competent peer members). Though speculative, these results suggest that GIFTs serve as expectations that are enforced via Social Identity processes regardless if they are positive or negative in nature.

### Implications

Our findings have several significant implications. First, this study contributes to the self-fulfilling prophecies literature by providing the first empirical evidence for naturally occurring Golem at work. Moreover, we offer a solution for investigating Golem effects via IFTs that circumvents the ethical concerns that have hampered research for decades. As Eden (1990) noted, individuals' expectations occur naturally, so this approach captures the phenomenon in its most ecologically valid context. Second, our study further develops the empirical work on IFTs and extends what has been viewed as individual level constructs (IFTs; Sy, 2010; Epitropaki et al., 2013) to group level constructs as viewed from the lens of the Social Identity Model (self-categorization theory; Turner, 1985). Specifically, we examine whether GIFTs (group level construct) is associated with followers' self-efficacy, effort, and performance (individual level constructs). Our model takes the multi-level nature of workgroups into account and provides a more accurate estimate for both within-and between-level of analysis (Zhang et al., 2009; Preacher et al., 2010). Third, while researchers have examined leadership processes via the Social Identity Model (Hogg, 2001; Hogg and van Knippenberg, 2003; van Knippenberg and Hogg, 2003), we advance a parallel model for followership processes that provide further insights and addresses gaps in the followership literature. Indeed, a rich body of research has accumulated around leaders and leadership process (Yukl, 2012), whereas research on followership process is scant (Carsten et al., 2010; Sy, 2010; Bligh et al., 2011). Advancing our understanding of followership is essential given that leadership can only occur if there is followership; there can be no leaders if there are no followers (Uhl-Bien et al., 2014). Hence, we appropriately examine the process of followership within a group context and how such a process may influence followers' overall performance. Lastly, as we bring to light that Golem effects occur naturally, it may benefit organizations to assess whether GIFTs are a factor affecting group performance. Organizations may intervene to restrain Golem effects by creating more positive expectations about group members. For example, group members may engage in a writing intervention describing their ideal or "best possible member" (Layous et al., 2013). In line with our prior propositions, this intervention may create more positive expectancies by redefining the salient attributes of group identity by which members self-identify and internalize (Turner et al., 1987; Hogg and Reid, 2006).

### Limitations

Despite its contributions and methodological strength such as the use of MSEM, this study, like any other studies, is not without its limitations. First, the cross-sectional nature of the study design prevents us from demonstrating causal directions. However, this design is a necessity for examining naturally occurring (i.e., without any form of artificial manipulation) Golem effects at work. Moreover, our approach offers an appropriate solution to circumvent the ethical concerns mentioned previously. Second, although the direction of relationships in our model is derived theoretically, followers' performance could also influence groups' conceptions of followers. Meaning, negative GIFTs could also be an outcome variable rather than a predictor variable (i.e., a recursive loop from followers' performance to negative GIFTs). Lastly, it is possible that the duration group members spent working together may impact how their GIFTs are formed because individuals' GIFTs may change based on their interactions with other group members (Lord and Maher, 1991). Hence, researchers may consider investigating how GIFTs emerged and transformed by implementing a longitudinal study design (Kozlowski and Chao, 2012).

### Future Research Directions

fpsyg-09-01581 September 8, 2018 Time: 18:36 # 9

Future research should investigate the impact of negative LIFTs (Leader's IFTs) on Golem effects and followers' performance (Babad et al., 1982). While research has demonstrated the positive interpersonal expectancy effect (Naturally occurring Pygmalion effects; Whiteley et al., 2012), studies have yet to investigate this negative interpersonal expectancy effect. Aligned with the claim of our study and the literature on selffulfilling prophecies (Eden, 2003), leaders with more negative conceptions of followers (i.e., negative LIFTs) should have more negative expectations of their followers which may trigger Golem effects, impairing followers' performance. As mentioned previously, research on Golem effects is often bounded by their inability to manipulate negative expectations due to ethical concerns. Using negative LIFTs, however, allow researchers to investigate interpersonal Golem effects while avoiding these ethical concerns.

The current study focused on the Incompetence dimension of negative GIFTs because we aimed to examine the traditional framework of Golem effects (i.e., Incompetence is most relevant to performance expectations). However, researchers may investigate an expanded theory of Golem effects using the Insubordinate and Conformity dimensions of GIFTs. For instance, how might the Insubordination dimension influence relationships among group members? Followers who internalize more insubordinate attributes may engage in more adverse behaviors (e.g., followers may be arrogant and mistreat members of the group) because they assume these behaviors are normative. This, in turn, may damage the relationships among members that cause detrimental outcomes. Good relationships lead to positive outcomes, whereas bad relationships have the opposite effect. For example, leader-follower dyads that have good relationships often lead to positive work outcomes, such as higher job satisfaction, job commitment, and organizational citizenship behaviors (Gerstner and Day, 1997; Ilies et al., 2007; Dulebohn et al., 2012). Thus, we propose an expansion of Golem effects by suggesting that it may operate via other mediators (e.g., relationship quality) beyond its core variables (i.e., efficacy and effort).

In addition, it might be fruitful to examine the outcomes related to the Conformity dimension of GIFTs. Although conformity is often viewed as a negative attribute in Western cultures, it may be a positive feature for followers in other cultures that endorse different types of follower attributes (Epitropaki et al., 2013). Indeed, researchers have found that individuals have different expectations for followers in Eastern and the Western cultures (Sy et al., 2017). Whereas followers in Western

### REFERENCES

Alipour, K. K., Mohammed, S., and Martinez, P. N. (2017). Incorporating temporality into implicit leadership and followership theories: exploring inconsistencies between time-based expectations and actual behaviors. Leadersh. Q. 28, 300–316. doi: 10.1016/j.leaqua.2016.11.006

cultures are expected to take on a more proactive role (e.g., make suggestions and speak up in meetings), followers in Eastern Cultures tend to conform to their leaders as a respectful gesture (e.g., execute tasks without questioning their leaders, and remain silent during meetings). As such, investigating the Conformity dimension using an Eastern cultural sample may lead to opposing predictions compared to what is expected in Western cultures planting Galatea effects in Eastern soil with a Golem seed.

Lastly, future studies on GIFTs may emulate the multi-level structural equation method and investigate how a level three variable (e.g., culture or organizational structure) may transform GIFTs. Organizational culture may be a source of alignment or discrepancy for IFTs at the group and individual levels (Fitzsimons et al., 2008; Sy, 2010). In addition, GIFTs that are endorsed by companies with traditional hierarchical structures may differ from those with flat or horizontal structures.

# CONCLUSION

Self-fulfilling prophecies reflect a double-edged sword (Eden, 1990). While positive expectations may promote positive outcomes, negative expectations can lead to detrimental outcomes. Much knowledge has accumulated on positive expectancy effects. In contrast, we know little about the dark side of self-fulfilling prophecies. There is still much to be learned about Golem effects. This study is a first step toward understanding how GIFTs may play a key role in summoning the detrimental consequences of Golem effects. It is important to note that even the most productive and gifted employees may be constrained when operating in workgroups that have high negative GIFTs. All in all, insights gained from GIFTs research may allow researchers to understand how Golem effects may be restrained so that employees may unleash their talents and transform as positive gifts of group performance.

### ETHICS STATEMENT

This study was conducted in accordance with the ethical guidelines set by the Institutional Review Board of the University of California, Riverside with written informed consent from all subjects. Employees participated in our study voluntarily. The protocol was approved by the Institutional Review Board committee of the University of California, Riverside.

# AUTHOR CONTRIBUTIONS

AL and TS designed the study and collected the data. AL analyzed the data and drafted the paper. TS revised and edited the paper.

Antonakis, J., and Dalgas, O. (2009). Predicting elections: child's play! Science 323, 1183–1183. doi: 10.1126/science.1167748

Babad, E. Y., Inbar, J., and Rosenthal, R. (1982). Pygmalion, galatea, and the golem: investigations of biased and unbiased teachers. J. Edu. Psychol. 74:459. doi: 10.1037/0022-0663.74.4.459

Bandura, A. (1997). Self-Efficacy: The Exercise of Control. London: Macmillan.



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Leung and Sy. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.