Gender stereotypes hinder women’s ascent to leadership. However, theory and empirical evidence regarding their changeability are divided. Social Role Theory (Eagly, 1987; Eagly and Wood, 2012) postulates that observing increasing numbers of women in agentic roles – e.g., leadership – changes female stereotypes toward higher agency (Koenig and Eagly, 2014). Discordantly, Backlash Hypothesis (Rudman and Glick, 2001; Rudman et al., 2012) questions the changeability of gender stereotypes, arguing that stereotype incongruent behavior gets punished. There is empirical support for both arguments (Diekman and Eagly, 2000; Haines et al., 2016).

Distinguishing descriptive and prescriptive stereotypes might reconcile such divided views. Descriptive stereotypes depict what men and women are like and lower the probability of recognizing leadership ability in women. Women stereotypically characterized as communal (e.g., kind, caring) are perceived as lacking the predominantly agentic qualities (e.g., assertive, independent) associated with successful leadership (Heilman, 1983, 2012; Eagly and Karau, 2002). However, they might change through observations of women in male domains as suggested by Social Role Theory.

Prescriptive stereotypes define what men and women should be like and thus exclude women from the agentic qualities demanded of leaders. Women – stereotypically required to be communal – suffered personal derogation and dislike when acting agentic (Heilman et al., 2004; Heilman and Okimoto, 2007). Thus, leadership ability in women – once recognized – is associated with high personal cost. Resonating with Backlash Hypothesis, prescriptive stereotypes might not change over time.

Obituaries are a unique source of information when studying stereotypes. While they are positively biased, they reveal what was appropriate to write about deceased leaders and reflect the zeitgeist of stereotypes. However, past studies that analyzed obituaries to infer changes in gender stereotypes did not distinguish between descriptive and prescriptive stereotypes (Kirchler, 1992; Rodler et al., 2001; Hartl et al., 2013). Addressing this shortcoming, we re-analyzed data of these previous studies from 1974 to 2010 with new data for 2016.

Changes in descriptive stereotypes are directly observable in obituaries as they contain information on leaders’ character and demeanor (i.e., what someone was like). In case of prescriptive stereotypes, however, changes are not directly observable (i.e., what someone should have been like). But obituaries contain evaluative information on how much leaders were liked and appreciated. Reverting to the established association between the violation of prescriptive gender stereotypes and dislike (e.g., Heilman and Okimoto, 2007), correlations of descriptions (i.e., as agentic) with evaluative information on likability (e.g., popular, esteemed) can be used to infer changes in prescriptive stereotypes.

Regarding descriptive stereotypes we hypothesized that deceased female leaders were described as increasingly agentic (e.g., assertive, independent) (H1a) and competent (e.g., experienced, expert) (H1b), but as decreasingly communal (e.g., kind, caring) over time (H1c). Concurrently, we hypothesized that increases in agency and competence of female leaders were associated with decreases in likability (e.g., popular, esteemed) (H2) expressing the assumption of stable prescriptive stereotypes. Obituaries of male leaders were included for comparison, allowing us to infer that the hypothesized changes were specific to obituaries of female leaders. Furthermore, we explored changes in obituaries of male leaders.

From 1974 to 2016, 1415 obituaries were identified for analysis. 1082 obituaries were about male and 333 about female leaders. For the latter, only 64 were found using the sampling procedure described above and 269 additional obituaries were found in the second extended search that included all issues of the respective year. As a noticeable trend, the number of obituaries dedicated to female leaders roughly doubled after the year 2000 (Table 1). An exploratory correlation between number of female obituaries and time reveals r = 0.82, p = 0.014.

Reconciling divided theory, we suggested that stereotypes about how women are might change, but stereotypes about how women should be might not. Results from the (re-)analysis of newspaper obituaries dedicated to female and male leaders of private and public organizations in Germany, Austria, and Switzerland supported this notion. Consistent with theories claiming changeability of stereotypes (e.g., Social Role Theory), female leaders were described as increasingly agentic over time; however, not as increasingly competent. Female leaders were not described as decreasingly communal over time. In contrast, male leaders were described as decreasingly competent, and in recent years as increasingly communal. Overall, these results indicate change in descriptive gender stereotypes.

Simultaneously, our results supported theories assuming stability of stereotypes over time (e.g., Backlash Hypothesis). Increased proportions of agency words were associated with decreased proportions of likability words in obituaries of female leaders. Furthermore, explorations revealed that proportions of communion words and likability words correlated negatively in obituaries of male leaders. Reverting to the established association between gendered prescriptions and (dis)likeability (e.g., Heilman and Okimoto, 2007), we concluded that prescriptive gender stereotypes remained unchanged over time.

Alternatively, a general decline of trust in leadership might explain decreases in likability of leaders over time. But opinion polls e.g., in Germany, showed overall high levels of trust in leaders (Nink, 2014; Haufe Akademie, 2015). Moreover, in this case, we should have found positive relations between communion and likability. After all, trust was linked to “feminine” leadership styles (Dirks and Ferrin, 2002; Eagly et al., 2003). In our data, however, communion and likability were unrelated for female leaders, and negatively related for male leaders.

Ultimately, our results raise questions about change mechanisms of prescriptive stereotypes which might be quantitatively different from those of descriptive gender stereotypes. The observation of women in agentic roles might lead to slower changes in prescriptive stereotypes indicating more pronounced “cultural lag” (Diekman et al., 2010). However, change mechanisms might differ qualitatively with prescriptive gender stereotypes following different change mechanisms altogether. Future research should consistently distinguish between descriptive and prescriptive stereotypes and clarify questions in this regard.

One limitation of this study was the re-analysis of previously aggregated data. Individual data would have allowed more elaborate and robust inferential analysis. Strengths of the present study were the long time-span observed and using unobtrusive measures. In opinion polls and questionnaire studies, research participants’ awareness of being observed might alter response-behavior (Podsakoff et al., 2003). Unobtrusive methods are therefore particularly suited for the research of sensitive issues such as stereotypes.

Last, our results implicate that merely increasing numbers of female leaders (i.e., by quotas), although important, might be insufficient. To support female leadership permanently, understanding and breaking prescriptive norms for women and men is essential.

All datasets generated and analyzed for this study are included in the manuscript and the Supplementary Files.

EK und MZ contributed to the conception and design of the study. EK and JO organized and supervised data collection. JO performed the statistical analysis. MZ wrote the first draft of the manuscript. JO wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

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.