BRIEF RESEARCH REPORT article
Getting a Grip on the Grapevine: Extension and Factor Structure of the Motives to Gossip Questionnaire
- 1Department of Organization Sciences, Vrije Universiteit Amsterdam, Amsterdam, Netherlands
- 2Department of Experimental and Applied Psychology, Vrije Universiteit Amsterdam, Amsterdam, Netherlands
- 3Department of Social Psychology, Universiteit van Amsterdam, Amsterdam, Netherlands
- 4Department of Psychology, Università Milano-Bicocca, Milan, Italy
Gossip is condemned but also ubiquitous and thought to be essential for groups. This triggers the question of which motives explain gossip behavior. Hitherto, negative influence, social enjoyment, group protection, and information gathering and validation are established as motives to gossip. However, venting emotions—discussed as a potentially important motive—has been overlooked empirically. Furthermore, a lack of consensus about a definition of gossip may have affected previous conclusions about gossip motives. This study (N = 460) expands the Motives to Gossip Questionnaire (MGQ; Beersma and Van Kleef, 2012) by including a subscale measuring emotion venting, the desire to share emotionally evocative experiences. To validate the five motives to gossip across definitions, we asked participants to report the most recent gossip event they experienced, randomly assigning them to one of three instructions containing different gossip definitions commonly used in the literature: (1) broad instructions (sharing information about third parties who have no knowledge of the exchanged information), (2) narrower instructions (adding that the shared information must be evaluative), and (3) instructions using the word gossip. After participants recalled and described a gossip event, they completed the 25-item measure of five motives to gossip: social enjoyment, information gathering and validation, negative influence, group protection, and emotion venting. Confirmatory factor analysis confirmed the five-factor structure. Multi-group confirmatory factor analysis supported full invariance across the three definition conditions. This indicates the Motives to Gossip Questionnaire successfully measures the five dimensions argued to motivate gossip and can be applied in research conceptualizing gossip both narrowly and broadly.
Gossip is omnipresent across societies, despite being condemned in public opinion (Wilson et al., 2000; Foster, 2004). However, recent literature suggests gossip is essential for social groups because it fosters cooperation and social bonding (e.g., Dunbar, 2004; Nowak and Sigmund, 2005; Wu et al., 2016). The paradoxical nature of gossip as condemned, yet supposedly essential for groups (Dunbar, 2004), raises the question of what motivates people to gossip. The Motives to Gossip questionnaire (MGQ) was developed by Beersma and Van Kleef (2012) to measure four theoretically derived motives: (1) social enjoyment, (2) information gathering and validation, (3) group protection, and (4) social influence (cf. Foster, 2004). The MGQ resolves issues affecting other scales related to gossip; some existing scales assess attitudes toward gossip and gossip use as dispositions, which neglects that gossip motives can differ across situations (Foster, 2004; Litman and Pezzo, 2005). Others solely measure gossip frequency, which does not allow for assessing why people gossip (Nevo et al., 1993; Wittek and Wielers, 1998; Brady et al., 2017).
To our knowledge, the MGQ is the only questionnaire assessing motives to gossip in a specific situation. However, several aspects required improvements. First, despite several authors arguing an important motive to gossip is venting emotions—to share emotionally evocative experiences (e.g., Grosser et al., 2012)—the MGQ lacked a subscale assessing this motive. People have a lay theory of catharsis and this lay theory can regulate their behavior in line with this theory (for example, catharsis beliefs were related to aggressing; Bushman et al., 2001). Whereas the effectiveness of venting emotions via intrapersonal processes is questionable, ample research indicates sharing emotions can elicit responses from others which can contribute to more positive affect (Nils and Rimé, 2012). For instance, people may vent emotions to elicit empathy and support (Pauw et al., 2018b), which are common responses to emotional sharing (Pauw et al., 2018a). Therefore, the desire to vent may motivate gossip about emotional events (Duprez et al., 2015). Indeed, Feinberg et al. (2012) showed reducing negative affect motivated gossip that could successfully alleviate negative affect. Likewise, Brady et al. (2017) found venting emotions was a prominent reason for workplace gossip and Waddington and Fletcher (2005) found nurses gossiped to express emotions. In sum, several studies show venting emotions can be a motive to gossip, suggesting measuring emotion venting is needed.
Second, whereas growing evidence suggests the desire to protect group members against norm violators is a prominent driver of gossip (Beersma and Van Kleef, 2011; Feinberg et al., 2012), this motive was measured with only three items in the MGQ, the minimum number of items to form a scale, making the scale vulnerable to response errors. Therefore, we extended the group protection subscale with two items to increase the robustness of the measure.
Third, for reasons of parsimony, we removed four redundant items from the information gathering and validation scale1, such that in the resulting questionnaire, each motive to gossip is measured with five items. Fourth, we carefully checked the wording of existing items to eradicate ambiguities and improve face validity.
Finally, we tested the validity of the MGQ by assessing the invariance of the underlying factor structure of the MGQ against variations in gossip conceptualizations. Gossip has been defined differently across studies. Sometimes it has been broadly defined as information shared about a third party who has no knowledge of the communicated information (cf. Peters and Kashima, 2015). Other studies have used a narrower definition, restricting gossip to information that portrays the target as positive or negative (cf. Foster, 2004), and yet others simply used the word “gossip” (e.g., Jazaieri et al., 2018). Currently, we do not know whether the MGQ enables distinguishing between different motives to gossip across these different conceptualizations. Therefore, we first examined its factorial validity in a representative sample and investigated whether a five-factor structure, distinguishing five motives for gossip as intended, fitted the data best. Second, we examined the measurement invariance of the MGQ across the three gossip definitions.
The total sample comprised a diverse community sample of 493 participants recruited through Dutch panel agency Flycatcher. Thirty-three participants indicated not recalling gossip and were removed. Demographic information was available for 453 participants (50.4% females); age ranged from 18 to 91 (M = 46.10, SD = 16.86).
Participants read one of three definitions of gossip in a broad (N = 152), narrow (N = 155), or simple (N = 153) instruction condition. The broad instructions instructed participants to report situations in which they communicated information about another person, who was absent or unable to hear the conversation. Participants were informed such information can relate to personal characteristics, attributes, events, behaviors, or needs of this person and could be positive, negative, or neutral. The information could be shared through any medium with any number of people present. Lastly, the information had to concern a specific person and did not have to remain secret to the third person (target of gossip). The narrow instructions were identical except for specifying the information must be positive or negative. The simple instructions instructed participants to report situations in which they communicated gossip instead of a definition of gossip, which could be shared through any medium and must concern a specific person (Complete instructions reported in Supplementary Materials).
Motives to Gossip
We extended the original MGQ by including a five-item subscale measuring emotion venting, adding two items to the group protection subscale, and removing four items from the information gathering and validation subscale2, resulting in a 25-item self-report questionnaire measuring five motives to gossip on seven-point Likert scales [1 (completely disagree), 7 (completely agree)]. Finally, we carefully checked the wording of all items for ambiguities and made minor changes in wording (overview in Supplementary Table 8). Items and descriptive statistics are reported in Supplementary Tables 1–4.
After providing informed consent, participants were randomly assigned to read either the broad, narrow, or simple instructions. Participants described the most recent situation in which they shared gossip (first for the current day, if not, they recalled the next most recent situation). After this, participants completed the MGQ3.
To examine whether the added fifth emotion venting dimension statistically contributes beyond the original four dimensions of the MGQ, we tested the five-factor solution against four-factor solutions using confirmatory factor analysis where emotion venting indicators load on the other dimensions.
We used the R-package Lavaan (Rosseel, 2012; R. Core Team, 2018), applying MLM Robust to correct for substantial multivariate kurtosis (Mardia's coefficients > 159.04, normalized estimates > 21.67) and allowing factors to correlate.
Next, we tested whether factor structures were similar across different gossip definitions by examination of the measurement invariance of the MGQ across the three instructions conditions using the SemTools-package (Hirschfeld and Von Brachel, 2014; Jorgensen et al., 2018). First, we tested for configural invariance (i.e., whether the number of latent variables and the pattern of indicators loading on factors is similar across definitions). Second, we tested for weak (metric) invariance (i.e., whether factor loadings are equal across definitions, indicating indicators reflect the same latent variables with the same intensity across definitions). Third, we tested for strong (scalar) invariance (i.e., whether item intercepts are equal across definitions, indicating the absence of systematic response biases and allowing comparisons of latent variable means across definitions). Fourth, we tested for strict invariance (i.e., equal residual variances across definitions, indicating the indicators show equal explained variance across definitions). Finally, we tested for the equality of factor means across definition conditions.
To assess the five-factor model fit, we used the following criteria for at least a satisfactory fit: CFI > 0.90; TLI > 0.90, RMSEA < 0.08, SRMR < 0.08, χ2/df < 3.00 (Hu and Bentler, 1999; Hooper et al., 2008). For measurement invariance tests, we used criteria for large samples: p-value of < 0.01 for Δχ2 (likelihood-ratio test) and ΔCFI> 0.002 indicating lack of measurement invariance (Meade et al., 2008).
Results showed that, for the complete sample, the five-factor model including a separate emotion venting dimension was satisfactory and significantly better than any four-factor model where the emotion venting items loaded on another latent factor (Table 1).
Table 1. Model fit statistics for the five-factor model tested against four-factor models including emotion venting in an existing dimension.
All standardized factor loadings were acceptable and were statistically significant (p < 0.001), ranging from 0.94 to 1.77 (Supplementary Table 1). Correlations were weak to moderately positive and statistically significant (Supplementary Table 5), indicating multiple motives are involved in one gossip situation. This is consistent with a priori assumptions, as there is no reason to expect a specific gossip instance involves a single motive. Furthermore, all subscales showed good internal consistency which did not differ between conditions, indicating the subscales measure their respective dimension across definitions (Supplementary Table 6).
Indicating configural variance, the model fit was acceptable in all three definition conditions (Supplementary Table 7). All standardized factor loadings were statistically significant (p < 0.001), ranging from 0.79 to 1.88 (Supplementary Tables 1–4).
Indicating weak invariance, strong invariance, and strict invariance, imposing the constraints of equal factor loadings, item intercepts, and residual variances did not lead to statistically significant differences in the fit of the five-factor model (Table 2).
Imposing the constraint of equal factor means led to a significantly worse model fit (Table 2), indicating at least two of the conditions differed on at least one factor. Comparing all factor means across conditions showed the participants in the simple condition scored higher on negative influence (ΔMbroad = 0.30, p = 0.033) compared to the broad condition and higher than broad and narrow conditions on information gathering and validation (ΔMbroad = 0.32, p = 0.039; ΔMnarrow = 0.34, p = 0.029). It is possible that the gossip condition fits better with lay views of gossip as negative and informative about others. However, ΔCFI did not exceed the 0.002 criterion. Given the ΔCFI and the large sample, this difference among means does not seem important or very reliable (supported by a MANOVA testing mean differences for all motives per condition [Pillai's Trace = 0.35, F(10, 908) = 1.61, p = 0.099].
In summary, across definitions, the MGQ reliably measures gossip motives that can be clearly distinguished, despite positive inter-correlations.
The five-factor Motives to Gossip model showed satisfactory fit and good reliability across subscales, indicating the revised MGQ successfully distinguishes social enjoyment, information gathering and validation, negative influence, group protection, and emotion venting motives to gossip. Results support Beersma and Van Kleef's (2012) theoretical model of gossip motives and extend it by including emotion venting. Enabling the measurement of this motive enhances the MGQ's ability to assess why people gossip.
Furthermore, the MGQ factor structure was completely invariant across broad, narrow, and simple definitions of gossip, indicating the MGQ is a stable instrument unaffected by different gossip definitions, which is corroborated by the dimensions' equal internal consistency across definitions. This implicates the MGQ can be used regardless of how gossip is defined. Furthermore, this enables comparing results across studies such as in meta-analyses.
Limitations and Strengths
Our study also had limitations. Firstly, we do not consider criterion variables such as gossip frequency or intensity, therefore it remains unclear whether emotion venting improves the predictive validity of the MGQ and whether the MGQ motives can predict criteria similarly across gossip definitions. Therefore, future research should investigate the MGQ's predictive validity for relevant criterion variables. Secondly, we did not consider contextual variables (e.g., group characteristics, Grosser et al., 2010, or personality, Lyons and Hughes, 2015) which could influence the MGQ's functioning.
Despite these limitations, our study is unique in its focus on measurement invariance across gossip definitions. We contribute to both understanding gossip motives and integrating gossip literature. Moreover, we used a large, diverse sample. This affords adequate power and allows generalizability beyond student or employee samples commonly used in gossip research. Lastly, asking respondents to report on real gossip events shows the motives captured in the MGQ are important in the natural setting where gossip occurs.
We demonstrated the MGQ is a stable and reliable instrument capturing several motives underlying gossip and aids our understanding of why people engage in the sometimes-puzzling behavior of gossip.
This study was carried out in accordance with the recommendations of Code of Ethics for Research in the Social and Behavioral Sciences, Scientific and Ethical Review Board (VCWE) of the Faculty of Behavioral and Movement Sciences at the VU University Amsterdam with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the Scientific and Ethical Review Board (VCWE).
BB and GvK developed the MGQ. TD and DB collected the data. TD and ES analyzed the data. TD wrote the first draft of the manuscript. DB, ES, BB, GvK, MG, and TD commented and worked on sections resulting in the current 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.
The authors would like to thank Flycatcher for their kind service.
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpsyg.2019.01190/full#supplementary-material
1. ^Results did not differ using all items.
2. ^An earlier adaptation of the MGQ (Dores Cruz et al., under review) measured emotion venting with seven items. For parsimony and because factor analysis demonstrated 2 items were redundant, we excluded two items from this earlier adaptation to create the current five-item subscale. Results did not differ using all items.
Beersma, B., and Van Kleef, G. A. (2012). Why people gossip: an empirical analysis of socialmotives, antecedents, and consequences. J. Appl. Soc. Psychol. 42, 2640–2670. doi: 10.1111/j.1559-1816.2012.00956.x
Brady, D. L., Brown, D. J., and Liang, L. H. (2017). Moving beyond assumptions of deviance: there conceptualization and measurement of workplace gossip. J. Appl. Psychol. 102, 1–25. doi: 10.1037/apl0000164
Bushman, B. J., Baumeister, R. F., and Phillips, C. M. (2001). Do people aggress to improve their mood? Catharsis beliefs, affect regulation opportunity, and aggressive responding. J. Pers. Soc. Psychol. 81, 17–32. doi: 10.1037/0022-35188.8.131.52
Feinberg, M., Willer, R., Stellar, J., and Keltner, D. (2012). The virtues of gossip: reputational information sharing as prosocial behavior. J. Pers. Soc. Psychol. 102, 1015–1030. doi: 10.1037/a0026650
Grosser, T. J., Lopez-Kidwell, V., and Labianca, G. (2010). A social network analysis of positive and negative gossip in organizational life. Group Org. Manage. 35, 177–212. doi: 10.1177/1059601109360391
Grosser, T. J., Lopez-Kidwell, V., Labianca, G., and Ellwardt, L. (2012). Hearing it through the grapevine: positive and negative workplace gossip. Org. Dyn. 41, 52–61. doi: 10.1016/j.orgdyn.2011.12.007
Hirschfeld, G., and Von Brachel, R. (2014). Multiple-Group confirmatory factor analysis in R: A tutorial in measurement invariance with continuous and ordinal indicators. Pract. Assess. Res. Evalu. 19, 1–12. Available online at: http://pareonline.net/getvn.asp?v=19&n=7
Hooper, D., Coughlan, J., and Mullen, M. (2008). Structural equation modelling: guidelines for determining model fit. Electron. J. Bus. Res. Methods 6, 53–60. Available online at: www.ejbrm.com
Hu, L. T., and Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct. Equat. Model. 6, 1–55. doi: 10.1080/10705519909540118
Jazaieri, H., Logli Allison, M., Campos, B., Young, R. C., and Keltner, D. (2018). Content, structure, and dynamics of personal reputation: the role of trust and status potential within social networks. Group Process. Intergroup Relat. doi: 10.1177/1368430218806056
Jorgensen, T. D., Pornprasertmanit, S., Schoemann, A. M., and Rosseel, Y. (2018). semTools: Useful Tools for Structural Equation Modeling. R Package Version 0.5-1. Retrieved from: https://CRAN.R-project.org/package=semTools (accessed January 18, 2019).
Meade, A. W., Johnson, E. C., and Braddy, P. W. (2008). Power and sensitivity of alternative fit indices in tests of measurement invariance. J. Appl. Psychol. 93, 568–592. doi: 10.1037/0021-9010.93.3.568
Nevo, O., Nevo, B., and Derech-Zehavi, A. (1993). The development of the tendency to gossip questionnaire: construct and concurrent validation for a sample of Israeli college students. Educ. Psychol. Meas. 53, 973–981. doi: 10.1177/0013164493053004010
Pauw, L. S., Sauter, D. A., Van Kleef, G. A., and Fischer, A. H. (2018a). I hear you (not): sharers' expressions and listeners' inferences of the need for support in response to negative emotions. Cogn. Emot. doi: 10.1080/02699931.2018.1536036
Pauw, L. S., Sauter, D. A., Van Kleef, G. A., and Fischer, A. H. (2018b). Sense or sensibility? Social sharers' evaluations of socio-affective vs. cognitive support in response to negative emotions. Cogn. Emot. 32, 1247–1264. doi: 10.1080/02699931.2017.1400949
R. Core Team (2018). R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing. Available online at: http://www.R-project.org/
Satorra, A. (2000). “Scaled and adjusted restricted tests in multi-sample analysis of moment structures,” in Innovations in Multivariate Statistical Analysis, eds R.D.H. Heijmans, D.S.G. Pollock and A. Satorra (Boston, MA: Springer), 233–247. doi: 10.1007/978-1-4615-4603-0_17
Wilson, D. S., Wilczynski, C., Wells, A., and Weiser, L. (2000). “Gossip and other aspects of language as group-level adaptations,” in The Evolution of Cognition, eds C. Heyes and L. Huber (Cambridge, MA: MIT Press), 347–365.
Keywords: motives to gossip, factorial validity, confirmatory factor analysis, measurement invariance, motives to gossip questionnaire, emotion venting, gossip definition
Citation: Dores Cruz TD, Balliet D, Sleebos E, Beersma B, Van Kleef GA and Gallucci M (2019) Getting a Grip on the Grapevine: Extension and Factor Structure of the Motives to Gossip Questionnaire. Front. Psychol. 10:1190. doi: 10.3389/fpsyg.2019.01190
Received: 19 January 2019; Accepted: 06 May 2019;
Published: 24 May 2019.
Edited by:Eva G. Krumhuber, University College London, United Kingdom
Reviewed by:Gary John McKeown, Queen's University Belfast, United Kingdom
Gordon Patrick Ingram, University of Los Andes, Colombia
Copyright © 2019 Dores Cruz, Balliet, Sleebos, Beersma, Van Kleef and Gallucci. 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.
*Correspondence: Terence D. Dores Cruz, firstname.lastname@example.org