Hiring women into senior leadership positions is associated with a reduction in gender stereotypes in organizational language
Appointing women into top management might mitigate the gendered expressions in language that describe women as less agentic than men.
Women continue to be underrepresented at all levels of management, especially in top organizational leadership including in Chief Executive Officer (CEO) positions and on boards. One of the reasons for this is the general belief that men more frequently display the agentic traits necessary for these positions, such as assertiveness and decisiveness.
Women face a double bind as they aspire to occupy leadership positions: on the one hand, they are less often described using agentic traits and are therefore considered to be less competent than men as candidates for such positions. On the other hand, if women do display agentic traits, they are seen as less likeable because they violate traditional gender stereotypes.
How can organizations change their language in organizational documents, such as public shareholder reports and investor documents, resulting in women being described in more agentic terms and not being penalized for it? The authors of this article explored the hypothesis that hiring a female CEO would be associated with changes in the organization’s language so that the semantic meaning of female words (e.g., she, her, woman) becomes more similar to the semantic meaning of agentic words (e.g., competent, independent, assertive). To do so, they analyzed a variety of public organizational documents including approximately 1.23 billion words from 696 organizations that entered the Standard and Poor’s (S&P) 500 index between 2009 and 2018 and measured the change in the gendering of the language organizations used in shareholder information documents, annual reports, and investor calls before and after hiring a female CEO (Study 1), and within executive boards with female representation (Study 2).
Hiring a female CEO was associated with changes to organizational language, with words used to describe women (e.g., she, her, woman) becoming more closely related to agentic, leadership-congruent traits (such as competent, independent and assertive), without imposing a trade-off with likeability.
- This study used a natural language processing (NLP) algorithm, word2vec, to map the relationship and similarity between words. The findings revealed that 72.7% of the target organizations vs. 27.3% of control organizations saw an increase in the association between female and agentic words, where words determined to be agentic (e.g., capable, independent, dominant) came to be more closely associated with words describing women (e.g., she, woman, her).
- There was no negative effect on the male-agency association when female CEOs were hired, which would have been predicted if the results were purely driven by changes in the gender of the individual who was described performing the role of the CEO.
- The authors observed a much stronger effect for the positive facets of agency (e.g., confident and independent) than the negative facets (e.g., dominant and aggressive). The agentic terms driving the effect were predominantly adjectives (e.g., active, tough, and original) and not action verbs, indicating that the experiment’s results were driven by women’s association with traits, not behaviors.
- Hiring a female CEO had a positive but very small effect on the degree of overlap and closeness between words used to describe women and words used to express communality (e.g., kind, caring), which the authors interpreted as a sign that hiring a female CEO was not associated with backlash on likeability or an agency–communality trade-off in organizational language.
- In a follow-up study to examine if the female-agency association was generalizable beyond female CEOs, the authors randomly selected approximately half of the organizations examined in Study 1 and separated observations into 8 estimates per company based on a 3-year period for each organization from 2009-2018 (i.e., 2009-2011, 2010-2012, etc.). The natural language processing model estimates for each 3-year time period were merged with the average proportion of women on each organization’s board of directors for that period to create a lagged proportion of women and infer the relationship between female-associated words and agency-associated words in the dataset. The researchers found a significant positive effect of the lagged proportion of women on company boards on the measure of the female-agency linguistic association across the following seven 3-year periods.
- The authors found a smaller, but positive and statistically significant, effect of the lagged female–agency association on the proportion of women on organizational boards. This might highlight the potential for a virtuous cycle where the more women get hired into leadership positions, the more embracing organizational language becomes of women’s leadership competencies, and as gender stereotypes dissipate for women more broadly, the more likely organizations are to hire additional female leaders.
- Akin to study 1, the increase in the female-agency association in organizational language did not come at the expense of women becoming less associated with communality.
These findings suggest that women’s presence in senior leadership positions changes gender stereotypes by providing female role models who become associated with leadership-relevant, agentic qualities. The authors hypothesize that such changes to how senior women are perceived might transfer to the wider category of women, conferring greater agency to women in general.
Study 1 aimed to test whether hiring a female CEO was related to an increase in the association of women and agentic words between the periods before and after she was hired. To capture an organization’s language, the authors extracted and analyzed the following types of communications for Standard and Poor’s (S&P) 500 organizations in a 10-year window between 2009 and 2018:
- A total of 13,770 publicly available filings from the U.S. Securities and Exchange Commission (SEC) and forms from the Wharton Research Data Services platform, including:
- DEF 14A filings, which consist of documents that are filed with the SEC when a shareholder vote is required to inform shareholders on a particular issue.
- 10-K filings, which are an organization’s annual reports, used to represent communication between an organization and its shareholders.
- Transcripts of 29,626 investor calls (from Seeking Alpha, a stock market platform for investors), which are verbal communication between high-level executives and stakeholders. The authors consider these calls as a more natural mode of communication.
Subsequently, the authors identified the subset of organizations that fulfilled three conditions:
- A female CEO in the sample period (41 organizations)
- Text data available for the 3 pre-hire and post-hire years
- A male CEO preceding the female CEO
These selection criteria yielded 11 target organizations, for which authors identified 22 propensity matched S&P organizations (two for each target organization) that did not hire a female CEO during the same period. They also performed this process for female CEOs who were replaced by male CEOs in the eligible period, providing an additional two target organizations and four propensity matches. In total, the sample consisted of 33 organizations, with an additional sample of 6 organizations that were analyzed separately.
The nature of the language organizations used before and after hiring a female CEO was analyzed with word2vec, an unsupervised natural language processing algorithm that estimated the semantic meaning of references to gender and references to agency. The authors then studied the distances between gendered words and agency words in vector space, a numerical representation of the relationship between these word categories. This allowed them to capture the strength of association between what it means to be a woman and what it means to be agentic in organizations’ language before and after the hire of a female CEO. The changes in language were then analyzed with regression.
The authors studied the relationship between the proportion of women on companies’ boards and the female–agency association longitudinally across 345 former and current S&P 500 companies. Panel vector autoregression (PVAR) models were used to determine the effect of lagged female representation on the female-agency association.
The authors randomly selected 345 of the 690 eligible organizations that entered the S&P 500 from 2009 to 2018 to pretrain a natural language processing word2vec model using the same text data as in Study 1. Then, they updated the machine learning models for the 345 organizations using a 3-year sliding window (i.e., 2009 to 2011, 2010 to 2012, etc.), to create eight estimates per company. They merged this panel with the average proportion of women on each organization’s board of directors for the relevant period to examine whether the lagged proportion of women predicted female-agency association.