Transference and interrupted transference: How a predecessor affects female senior manager dismissal

LCT theory

Among the theories that attempt to explain why and how female senior managers are treated differently from male senior managers, LCT posits that individuals rely on schemas (i.e. prototypes) from their belief system that categorize leaders using self-determined features that distinguish good leaders from bad leaders (Lord et al., 1982, 1984; Scott and Brown, 2006). Van Quaquebeke et al. (2011) argued that leader categorization stems from individuals’ perceptions when comparing their leaders to an “ideal leader,” in which the traits of an ideal leader range from demographic traits to self-perceptions to benchmarks associated with successful leaders in business, politics, and sports (see also Lord et al., 1984). Stereotypes regarding women are pervasive and widely held by both women and men (Nye and Forsyth, 1991), with women typically thought to be interpersonally sensitive and concerned with the welfare of others (Scott and Brown, 2006; Williams and Best, 1990). Men, on the other hand, are believed to be agentic and instrumental, more task-oriented, ambitious, competent, and competitive—traits that evaluators are more likely to associate with their prototypes of effective leaders (Eagly and Karau, 2002; Lord and Maher, 1993; Scott and Brown, 2006). Thus, women are seen as a categorical misfit for senior positions, affecting their ability to advance to the senior position and be considered legitimate if they do advance.

Evaluators’ attribution of good and bad performance is central to LCT. When leaders are perceived to fit the leadership stereotype—that they possess the necessary skills and competencies expected of a leader—positive outcomes are attributed to their leadership abilities regardless of the validity of a causal relationship (Wangrow et al., 2023). However, when they are not perceived to fit the leadership stereotype, causal relationships between their abilities and positive outcomes are discounted. Furthermore, leaders who do not fit the leadership stereotype are more likely to be penalized for poor performance or missing expectations (via dismissal, reduced compensation, etc.) than those who fit the stereotype (Obenauer and Langer, 2019). For female leaders, this is especially noteworthy, as prior research has shown that the likelihood of female leaders being considered more effective leaders than men is only greater when they are in roles that are perceived to be salient to their gender (e.g. nursing and flight attendants; Hogg et al., 2006).

Prior research supports LCT’s relevance to female leader stereotypes. Heilman et al. (2004) found that women who are successful are less liked and face greater negative reactions than men, especially in roles that are considered to be distinctly masculine. This suggests that evaluations of successful women are subject to stereotypes, even when these women have demonstrated competence and achieved positive outcomes (Heilman et al., 2004). Relatedly, Scott and Brown (2006) found that perceptions of female leaders continue to be based on stereotypes even when they exhibit more agentic and instrumental behaviors, providing further support that female leaders are categorized instead of being evaluated by their behaviors, actions, and outcomes.

Role congruity theory

Eagly and Karau (2002) proposed the role congruity theory as an extension of the social role theory of gender differences and similarities (see also Eagly, 1987), which further explains how individuals categorize leaders. It has been argued that prejudice occurs when the gender of an actual or proposed occupant of a leadership role is incongruous with key stakeholders’ views of leadership (see Del Carmen Triana et al., 2024, for a review). When women are in roles viewed to be incongruous by evaluators and other key stakeholders, prejudice against them manifests itself in two forms: (1) less-favorable evaluation of their leadership ability because they are not consistent with stereotypes and (2) less-favorable evaluation because behaviors deemed necessary for leadership are more acceptable from men than they are from women (Eagly and Karau, 2002; Johnson et al., 2008). Less-favorable evaluation from prejudice in either form is likely to lead to reduced opportunities for advancement, lower compensation, and a greater likelihood of dismissal (Del Carmen Triana et al., 2024). Eagly and Karau (2002) note that role congruity theory is most applicable at senior managerial levels since highly agentic qualities (e.g. acting as a change agent, managerial courage, and results-oriented) are deemed necessary for senior managers to be successful.

Subsequent studies have offered evidence of how role incongruity may disadvantage women in the workplace. Gupta et al. (2020) found that, while controlling for performance, female CEOs of publicly traded firms are more likely to be dismissed than male CEOs. While some studies do not find a significant relationship between a CEO’s gender and the likelihood of dismissal (e.g. Elsaid and Ursel, 2018), this finding is noteworthy because it points to potential contexts in which female senior managers may be more likely to be dismissed. Dwivedi et al.’s (2021) study of new female CEOs suggests that leadership endorsements (usually from the board of directors) of past achievements and competence penalize new female CEOs, resulting in shorter tenures unless the endorsements are for a new insider female CEO ⁴ or the firm has a substantial number of female leaders. Dwivedi and colleagues theorized that stakeholders’ view of the new female CEO through a stereotypical lens is reinforced because the endorsements run contrary to the stereotypes they hold.

Scholars have also found that role incongruity problems can extend to investor perceptions of mature firms (Del Carmen Triana et al., 2024). Lee and James (2007) found that investor reactions to CEO appointments are significantly more unfavorable when the new CEO is female, while Cook et al. (2024) provided evidence that female CEOs are more likely to be dismissed when performance is poor and that firms are rewarded by investors when they replace them with white men. Other studies found that investor expectations of firm growth (Yacus et al., 2019) and shareholder returns (Jeong and Harrison, 2017) are lower in firms led by women. Further, challenges associated with role incongruity can extend to women in other leadership roles. For instance, Lee and Huang’s (2018) findings suggest that entrepreneurial ventures led by women may be perceived less favorably than ventures led by men because of incongruity between expectations of female and entrepreneurial roles, while other studies show that incongruity between expectations of gender and founder may negatively impact an entrepreneurial venture’s commercialization (Uzuegbunam et al., 2021), financing (Eddleston et al., 2016), and funding (Cowden et al., 2021). Overall, these studies suggest that female leaders, including senior managers, can suffer from prejudice because their gender does not match stereotypes held by evaluators and other key individuals, which likely leads to negative outcomes (e.g. less support, lower compensation, reduced tenure, and greater likelihood of dismissal) for female senior managers than their male counterparts.

Direct succession of a female senior manager (leader transference)

The premise of LCT and role congruity theory in gender research—that all female leaders are viewed and treated the same by their organization’s evaluators—is built on the assumption that evaluators who form perceptions of current female leaders likely have had no prior experience with women in senior leadership roles. However, if the predecessor leader has been a woman, “evaluators are likely to view the predecessor as a highly relevant point of comparison . . . because this person occupied the same role” as the current female leader (Preston and Carton, 2024: 332). The predecessor is highly salient because psychological processes imprint the most recent observations as key reference points (see Tversky and Kahneman, 1973), and evaluators tend to overemphasize recent observations when judging the next observation. Prior research has shown that when evaluators focus on a known person, stereotypes and prototypes are less likely to affect judgment (Andersen and Baum, 1994; Andersen and Berk, 1998; Chen, 2001). As a result, evaluators are less likely to be affected by preconceived biases and stereotypes about female leaders and, instead, are likely to employ the process of transference. Ultimately, this process should impact how evaluators assess and treat the current female leader, which could ultimately affect the likelihood of her dismissal.

Leader transference is a psychological process where evaluators falsely assume that two leaders who share a common characteristic, such as gender, are also similar in other ways. In such cases, evaluators make assumptions about the success of the leader based on transferring the traits of the predecessor (Kelley and Michela, 1980). As evaluators come to know the female predecessor and obtain concrete information about her skills and behaviors, they are likely to prioritize this information (Napier et al., 2018; Rim et al., 2013) and use it to make comparisons between the predecessor and current female leader (Smith and Zarate, 1992). Furthermore, as the female predecessor represents the most recent experience imprinted on evaluators, their episodic memory is likely to affect their judgment (Tulving, 1972, 1984) and lead them to generalize that the current female leader has similar skills and competencies (Ritter and Lord, 2007). Prior research provides evidence that similarity between a current leader and a predecessor is positively related to generalizations that other characteristics of the two leaders are also similar (Preston and Carton, 2024).

Translating these arguments to our context, we expect to observe a transference process based on gender. Evaluators could use gender similarity between the current female leader and her female predecessor to perceive other similarities between them (Andersen and Berk, 1998; Chen and Andersen, 1999). For example, the appointment of the predecessor to a senior manager position serves as evidence that she has overcome “more barriers than men on their way up” (Gupta et al., 2020: 563) and is “particularly gifted and/or especially good at learning and/or dealing with adversity” (Gupta et al., 2018: 2039). Evaluators are likely to ascribe these skills and competencies of the female predecessor to the current female leader. Such attributions should significantly reduce or even overcome inherently held biases and stereotypes on the incongruity between a woman and the senior position (Andersen and Baum, 1994; Andersen and Berk, 1998). Thus, evaluators are more likely to view the current leader as a strong fit (or less of a misfit) for the position. This reduced bias should lead to a lower probability of mistreatment and discriminatory behavior toward the current female leader, resulting in a lower likelihood of dismissal. Formally stated,

Female senior manager tenures separated by male leadership (interrupted gender transference)

The underlying argument in our first hypothesis is that an organization’s evaluators change their perceptions of women in leadership roles when they are familiar with the woman who previously served in the senior role. Here, we consider cases where one or more men served as the senior leader between female senior manager tenures, which may reintroduce LCT, role incongruity bias, and stereotyping.

Regardless of the performance and perceptions of the woman who most recently served in the senior leadership position, a few factors point to the potential reintroduction of gender bias and stereotyping when one or more men served in the senior role between female senior manager tenures. First, the accepted status quo of the senior position may, over time, revert back to a man serving in the senior position. One or more years of a man at the organization’s helm can generate recency bias, where evaluators overemphasize recent experiences and overlook more distant experiences (Tversky and Kahneman, 1973). As the observation of and interaction with the male leader represent the most concrete and relevant information for evaluators, his skills and behaviors are likely to become imprinted on evaluators’ minds and judgments (Tulving, 1972, 1984). As a result, memory of the legitimacy and leadership skills of the female predecessor may erode with longer periods of male leadership and be replaced with more recent impressions of the male leader. With longer periods, evaluators’ ability to observe and interact with the female predecessor, an essential part of reducing gender bias, declines (Dovidio et al., 2017). Thus, gender stereotyping may resurface among evaluators, leading to enhanced perceptions of a misfit between a woman and the senior manager position. Second, as more time elapses since the female predecessor was in charge of the organization, there is a greater chance that the organization’s evaluators and other key stakeholders will change, with new individuals likely lacking contact and experience with the female predecessor. In such a case, it is more likely that a new female senior manager will be considered incongruent for the leadership role and evaluated less favorably.

Ultimately, we posit that the presence of male leadership between female senior manager tenures can undermine gender transference effects and lead to what we term as “interrupted gender transference.” In particular, as the time between female senior manager tenures lengthens, during which one or more men lead the organization, the gender transference effects are likely replaced by the reintroduction of gender bias and stereotyping that results in an increased probability of dismissal for the current female leader. We therefore offer the following hypothesis:

Boundary conditions of female senior manager dismissal

Given that senior managers do not exist in a vacuum, it is important to examine contingency conditions that are likely to impact their likelihood of dismissal. We focus on predecessor performance, as performance serves as a valid indicator of senior managers’ skills and reflects their ability to handle complex and challenging situations (Hubbard et al., 2017; Wiersema and Zhang, 2011), and predecessor performance is a critical factor affecting gender bias and stereotyping (e.g. Gupta et al., 2020).

Performance triggers cognitive processes of evaluating leaders’ accountability and role congruity (Weber and Wiersema, 2017). In line with research on attribution and inference-based approaches to leadership (e.g. Calder, 1977; Kelly and Michela, 1980; Meindl et al., 1985), a female predecessor’s strong (weak) performance can instill (reduce) confidence among evaluators regarding her competence and effectiveness as a leader. Furthermore, strong (weak) performance can weaken (strengthen) inherent biases that a woman is a misfit for a senior manager position. Favorable predecessor performance should lessen misfit evaluations, attribute the source of performance to the current female leader instead of external factors, and instill that the current female leader is capable of performing as her female predecessor. In contrast, when a male predecessor performs well (poorly), evaluators’ gender biases could increase (decrease), leading to a higher (lower) probability of dismissal for the current female leader. As a result, a female predecessor who led the organization to strong performance compared to a male predecessor who generated strong performance should increase transference effects and ultimately lower the probability of dismissal for the current female senior manager. Thus,

For situations when one or more men serve as senior managers between female senior manager tenures, we argue that strong performance of the male senior manager(s) can enhance recency bias (Tversky and Kahneman, 1973) and ultimately gender bias. For example, if one or more of these male leaders were successful in the senior position, evaluators are likely to disproportionately attribute positive performance to their skills (e.g. Park and Westphal, 2013), strengthen their inherent beliefs that men are better leaders than women, and forget or ignore the legitimacy and leadership skills of the female predecessor. These resurfaced biases should lead evaluators to unfairly cast blame on current female leaders for poor performance and diminish expectations of the female leader’s ability to secure future performance. The longer that successful male leaders have served in the senior position, the stronger the gender bias against them and the higher the likelihood of dismissal for the current female leader. In contrast, when one or more of these male leaders performed poorly, evaluators are less likely to perceive men as superior leaders compared to women. In these cases, evaluators are more likely to reflect on the leadership competence and expertise of the female predecessor, resulting in a lower likelihood of gender bias toward the current female senior manager.

We argue that the male predecessor’s strong performance confirms male-centric biases, such that the strong performance may suggest a male leadership advantage and supplant information about the prior female leader’s effectiveness. The result strengthens interrupted gender transference, reducing favorable evaluator judgments toward female leaders and increasing the likelihood of their dismissal. Formally stated,

Dependent variable

For each season (i.e. spell) in which a coach was dismissed or forced to resign, Dismissal was assigned “1.” If the coach continued into the following season, voluntarily resigned or died, Dismissal was assigned “0.” Causes of voluntary resignation include retirement, poor health, leaving to coach a Women’s National Basketball Association (WNBA) team, leaving to serve as a WNBA executive, or leaving to coach another Women’s NCAA college basketball team.

As organizations generally do not publicly disclose reasons for dismissing senior managers, it is difficult to determine the cause of turnover (Shen and Cannella, 2002). To assess and determine the cause of each turnover in our sample, one of the authors and a research assistant independently reviewed press releases and media reports associated with each turnover event. Such methods have been previously used to distinguish dismissals from other types of turnover (e.g. Shen and Cannella, 2002; Wangrow et al., 2018). The author and research assistant agreed on 310 of the 330 turnover events (93.9%), which yielded 0.878 for Cohen’s kappa and suggested strong alignment between the two raters (McHugh, 2012). For the 20 turnover events in which they disagreed, the two researchers discussed the university and media reports surrounding each turnover and reached an agreement on whether each of these events were associated with dismissal.

This method was repeated, with an almost perfect level of agreement (i.e. Cohen’s kappa greater than 0.90, see McHugh, 2012), for the following variables: Gender, Prior Coach Gender, Person of Color, Prior Coach Dismissed, and Prior Coach Person of Color.

Independent variables

The binary variable, Gender, was assigned “1” if the current coach was a woman and “0” otherwise. Approximately 66% of the spells were associated with female coaches. ⁶ Prior Coach Gender is a binary variable that was assigned “1” if the prior coach was a woman and “0” otherwise.

Female Coach Gap is the number of years following a prior female coach’s departure that one or more men served as coach before the current female coach. For example, at the University of Illinois, a female coach (Nancy Fahey) was preceded by a male coach (Matt Bollant) who coached for six seasons. Matt Bollant was preceded by a female coach (Jolette Law). Female Coach Gap is coded “6” for each season in which Nancy Fahey was the coach. This variable is set to “0” when the current female coach replaces a female coach. For example, at the University of Minnesota, a female coach (Lindsay Whelan) replaced a female coach (Marlene Stollings). Female Coach Gap is coded “0” for each season in which Lindsay Whelan was the coach.

Prior Coach Winning Percentage (for testing H3) is the percentage of total games that the team won during the prior coach’s tenure. This variable also serves as a control variable for tests of H1 and H2, as it can be an indicator of a glass cliff scenario in which the female coach was hired during a period of crisis (Ryan et al., 2007; Ryan and Haslam, 2005). Prior Male Coaches’ Winning Percentage (for testing H4) is the combined winning percentage of all male coaches during the period between the beginning of a female coach’s tenure and the end of the tenure of the most recent prior female coach. For example, at the University of Oklahoma, two male coaches (Burl Plunkett and Gary Hudson) held the position between the current coach (Sherri Coale) and the team’s most recent female coach (Valerie Goodwin-Colbert). The combined winning percentage of these two male coaches, 52.9%, is assigned for each season in which Sherri Coale was the coach.

Control variables

We controlled for additional variables at the individual, team, and university levels that might affect coach dismissal. The binary variable, Person of Color, was assigned “1” if the current coach represented a non-White racial or ethnic group and “0” otherwise. Approximately 18% of the observations are associated with non-White coaches. Binary variables were included for whether the current coach was a Former Division 1 Player and whether he or she was an Alumnus of the university whose team they led. Previous Head Coaching Experience is the number of years the current coach led other NCAA Division 1 teams prior to beginning his or her tenure as the focal team’s coach. Years Assistant at School is the number of years that the current coach served as one of the team’s assistant coaches earlier in his or her career. We also included two variables associated with a coach’s reputational capital and power. Following prior research (e.g. Allen et al., 2022), we counted each time the current coach had achieved the following accomplishments: reaching the Women’s NCAA Final Four, winning the Women’s NCAA national championship, recognition as the conference’s coach of the year, and recognition as the Naismith National Coach of the Year. Prestige when Hired is a coach’s cumulative accomplishments prior to being hired, while Prestige at School is a coach’s cumulative accomplishments since beginning their tenure as the team’s coach.

University-level control variables account for a university’s structural or institutionalized norms as sources of potential bias. Private University was coded “1” if the university was privately owned. New Conference was coded “1” if, for that season, the team was competing in a different conference than the prior year. We also included a binary variable, AD Change, that was coded “1” if the current athletic director was different than the athletic director who hired the current coach. Financial Support was obtained from the US EADA database (https://www2.ed.gov/finaid/prof/resources/athletics/eada.html) and is the total dollars spent (in millions) by the university for that season’s women’s basketball program. ⁷

Three variables account for expectations or challenges prior to a season. Preseason Ranking is a team’s ranking in the Associated Press’ Top 25 prior to the season, with a team ranked number one assigned 25, a team ranked number two assigned 24, and so on. Teams not included in the preseason rankings were assigned 0. Draft Value measures the team’s cumulative talent level for a season (e.g. see Wangrow et al., 2023). For each player who eventually played in the WNBA, the variable is incremented by 4 if the player was drafted in the first round, 3 if drafted in the second round, 2 if drafted in the third round, and 1 if not drafted. Scandal or Sanctions is a binary variable that was coded “1” if there was an ongoing scandal during the season or if the NCAA had penalized the team in some way for that season.

Winning Percentage (t) is the percentage of total games that the team won during the most recent season, while Winning Percentage (t − 1) is the percentage of total games that the team won during the prior season. As prior research has shown that the relationship between performance and dismissal is nonlinear (e.g. Wangrow et al., 2018), the linear and squared terms of these variables are used in the analysis models.

Three additional variables associated with the prior coach are included. Prior Coach Dismissed was coded “1” if the prior coach had been dismissed. Prior Coach Person of Color was assigned “1” if the prior coach represented a non-White racial or ethnic group. Prior Coach Tenure is a count variable representing the number of years the prior coach led the team.

For models 1 through 6 in our analysis, 19 dummy variables were included for year. All models included 12 dummy variables for the conference in which the team competed. ⁸

Analysis

We used a Cox Proportional Hazards event-history model (Cox, 1972) to examine whether the likelihood of dismissal for a female coach was affected by whether the coach had succeeded a female or male coach. A Cox model estimates the hazard rate h(t), which represents the probability of dismissal following a season, in which a season is a distinct period (i.e. spell) in which the coach is at risk of being dismissed (see Wangrow et al., 2023, for an example). A Cox model was chosen because it does not assume the hazard function’s characteristics but instead uses data to determine the underlying hazard function (Benner and Tripsas, 2012; Blossfeld et al., 2007). In addition, as many variables change during a coach’s tenure, Cox models are ideal for this analysis because they allow for time-varying covariates and, thus, can handle fluctuations of variables over a coach’s tenure (Blossfeld et al., 2007; Box-Steffensmeier and Jones, 2004). Further, this study’s analysis method needed to effectively prevent right-censored data from biasing parameter estimates since many coaching tenures end voluntarily or extend into the 2022–2023 season. Cox models are appropriately suited for this.

Results

Table 1 provides descriptive statistics and correlations. To verify that collinearity is not a problem, we determined the variance inflation factor (VIF) for each independent variable in each model. For models 1 through 6, as VIF did not exceed 3.6 for any variable, multicollinearity was not indicated, and all variables were included in the models (VIF < 10; see O’Brien, 2007). For model 7, which uses a small subset of the main sample, collinearity was indicated, and three variables (New Conference, Scandals or Sanctions, and Preseason Ranking) were removed from this model to eliminate collinearity issues.

OPEN IN VIEWER

Table 1.

Descriptive statistics and correlations.

	N	Mean	Std. dev.	Min	Max	1	2	3	4	5	6	7	8	9	10
1. Dismissed	2548	0.07	0.25	0	1
2. Tenure	2548	8.00	7.55	1	42	0.005
3. Gender	2548	0.66	0.47	0	1	0.053**	−0.025
4. Person of color	2548	0.18	0.38	0	1	0.051*	−0.149**	0.081**
5. Former D1 player	2548	0.56	0.50	0	1	0.050*	−0.043	0.474**	0.155**
6. Alumni	2548	0.11	0.31	0	1	−0.001	0.047	−0.006	−0.043*	0.217**
7. Previous head coaching experience	2548	4.59	6.23	0	33	−0.018	−0.046	−0.014	−0.136**	−0.006	−0.082**
8. Years assistant as school	2548	0.66	2.09	0	17	0.004	−0.087**	−0.130**	0.087**	0.053**	0.202**	−0.182**
9. Prestige when hired	2548	0.74	1.50	0	13	−0.030	−0.027	−0.055**	−0.057**	−0.017	−0.098**	0.651**	−0.119**
10. Prestige at school	2548	1.67	4.76	0	51	−0.059**	0.620**	−0.031	−0.090**	−0.056**	−0.076**	0.003	−0.069**	0.004
11. Scandals or sanctions	2548	0.01	0.09	0	1	0.097**	0.021	−0.027	−0.009	−0.045*	−0.006	−0.015	0.010	0.012	0.003
12. Private university	2548	0.37	0.48	0	1	0.001	0.088**	−0.093**	−0.009	0.077**	0.098**	−0.140**	0.113**	−0.140**	0.031
13. New conference	2548	0.03	0.16	0	1	0.004	0.022	0.002	0.020	−0.029	0.004	−0.014	−0.007	−0.022	0.035
14. AD change	2548	0.51	0.50	0	1	0.038	0.544**	0.043*	−0.055**	−0.040*	−0.037	0.017	−0.136**	−0.037	0.252**
15. Financial support	2525	2.69	1.44	0.22	20.78	−0.016	0.296**	0.041*	0.005	0.001	−0.044*	0.186**	−0.110**	0.261**	0.383**
16. Preseason ranking	2548	2.41	5.94	0	25	−0.090**	0.374**	−0.006	−0.077**	−0.040*	−0.059**	0.077**	−0.069**	0.148**	0.586**
17. Draft value	2548	2.16	4.20	0	32	−0.075**	0.331**	−0.014	−0.040*	−0.047*	−0.044*	0.099**	−0.079**	0.158**	0.523**
18. Winning percentage (t)	2548	55.20	19.12	0	100	−0.265**	0.351**	−0.073**	−0.128**	−0.064**	−0.018	0.025	−0.040*	0.098**	0.380**
19. Winning percentage (t − 1)	2548	55.18	19.07	0	100	−0.188**	0.405**	−0.065**	−0.123**	−0.061**	0.000	0.005	−0.026	0.065**	0.395**
20. Prior coach dismissed	2468	0.50	0.50	0	1	0.020	−0.152**	0.020	0.063**	−0.025	0.023	0.121**	−0.174**	0.062**	−0.065**
21. Prior coach person of color	2468	0.16	0.37	0	1	0.008	−0.117**	−0.030	−0.001	−0.091*	−0.013	0.094**	−0.060**	0.050*	−0.102**
22. Prior coach winning percentage	2468	49.93	15.75	12.80	86.20	0.015	−0.023	0.142**	0.023	0.111**	0.127**	−0.051*	0.112**	0.017	−0.036
23. Prior coach gender	2468	0.75	0.43	0	1	−0.034	0.073**	−0.013	−0.084**	−0.039*	−0.035	0.050*	−0.114**	0.074**	0.080**
24. Prior coach tenure	2468	7.90	6.10	1	39	0.022	−0.173**	0.192**	0.108**	0.130**	−0.011	0.004	0.137**	0.089**	−0.088**
25. Female coach gap	1600	2.05	4.76	0	28	0.066**	−0.128**	n/a	0.060*	−0.036	−0.051*	−0.001	0.246**	−0.082**	−0.078**
26. Prior male coaches’ winning percentage	384	50.46	11.28	23.00	78.10	0.071	−0.024	n/a	0.069	0.131**	−0.085	−0.040	0.203**	−0.091	0.085
	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25
12. Private university	−0.051*
13. New conference	−0.016	0.012
14. AD change	−0.003	−0.022	0.008
15. Financial support	0.050*	−0.116**	−0.042*	0.147**
16. Preseason ranking	0.045*	−0.058**	−0.008	0.196**	0.477**
17. Draft value	0.057**	−0.118**	0.020	0.175**	0.454**	0.732**
18. Winning percentage (t)	0.028	−0.049*	−0.001	0.220**	0.328**	0.511**	0.517**
19. Winning percentage (t − 1)	0.029	−0.040*	0.025	0.252**	0.327**	0.539**	0.472**	0.639**
20. Prior coach dismissed	0.005	−0.034	−0.014	−0.093**	−0.029	−0.132**	−0.114**	−0.160**	−0.218**
21. Prior coach person of color	−0.004	0.083**	−0.025	−0.035	−0.052**	−0.083**	−0.098**	−0.095**	−0.114**	0.223**
22. Prior coach winning percentage	0.004	−0.034	−0.002	−0.068**	0.185**	0.134**	0.154**	0.143**	0.204**	−0.402**	−0.128**
23. Prior coach gender	0.013	−0.079**	0.002	0.033	0.025	0.064**	0.047*	0.053**	0.051*	0.108**	0.076**	−0.102**
24. Prior coach tenure	−0.011	−0.014	−0.020	−0.135**	0.104**	0.017	0.048*	−0.018	−0.019	−0.107**	−0.116**	0.391**	0.039*
25. Female coach gap	−0.001	0.157**	−0.028	−0.065**	−0.071**	−0.114**	−0.135**	−0.111**	−0.092**	−0.146**	−0.012	0.060*	n/a	0.090**
26. Prior male coaches’ winning percentage	0.034	−0.131*	−0.048	−0.109*	0.230**	0.139**	0.152**	0.222**	0.329**	−0.472**	−0.231**	n/a	−0.076	0.249**	0.324**

*

p < 0.05. **p < 0.01.

Table 2 provides results from the event-history analysis. Exponentiated hazard coefficients are shown, with a significant and positive (negative) coefficient indicating that the likelihood of a coach being dismissed increases (decreases) as the associated variable increases. A variable’s numerical effect on dismissal is determined by taking the exponential function of the coefficient (i.e. numerical effect = exp(exponentiated hazard coefficient)). For example, model 1 of Table 2 reports a significant hazard coefficient of 0.263 for Financial Support. Since exp(0.263) equals 1.301, the results suggest that, with other factors held equal, the likelihood of any coach in the sample being dismissed increases by 30.1% for each additional million dollars of annual spending by the university on its women’s college basketball team.

OPEN IN VIEWER

Table 2.

Cox survival hazard models.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
	Full sample	All male coaches in the sample	All female coaches in the sample	All female coaches in the sample	All male coaches in the sample	All female coaches in the sample	Female coaches in the sample with one or more male predecessors
Variables
Gender	0.017(0.203)
Person of color	0.094(0.181)	0.180(0.512)	0.129(0.227)	0.099(0.238)	0.229(0.498)	0.036(0.236)	-0.430(0.550)
Former D1 player	0.295(0.193)	0.802 +(0.438)	0.140(0.225)	0.163(0.238)	0.773 +(0.440)	0.143(0.216)	0.320(0.441)
Alumni	−0.077(0.230)	−1.583* (0.677)	0.163(0.274)	0.193(0.264)	−1.636* (0.692)	0.270(0.267)	0.615(0.421)
Previous head coaching experience	−0.014(0.021)	0.035(0.052)	−0.032(0.024)	−0.034(0.024)	0.028(0.056)	−0.036(0.024)	0.050(0.045)
Years assistant as school	−0.001(0.035)	−0.084(0.068)	0.089(0.065)	0.078(0.068)	−0.080(0.070)	0.085(0.061)	−0.008(0.102)
Prestige when hired	−0.024(0.086)	−0.346(0.232)	0.077(0.078)	0.081(0.081)	−0.298(0.251)	0.097(0.082)	0.079(0.261)
Prestige at school	−0.096(0.107)	0.623** (0.222)	−0.253* (0.117)	−0.233 +(0.125)	0.606** (0.217)	−0.226* (0.115)	−0.378(0.327)
Private university	−0.089(0.182)	−0.758(0.476)	−0.031(0.231)	−0.085(0.256)	−0.799(0.492)	−0.042(0.228)	−0.175(0.467)
New conference	−0.246(0.401)	−1.792(1.322)	−0.098(0.440)	−0.155(0.518)	−1.784(1.303)	−0.217(0.450)
AD change	−0.119(0.171)	−0.011(0.497)	−0.119(0.215)	−0.159(0.217)	0.017(0.508)	−0.146(0.212)	−0.151(0.583)
Scandal or sanctions	2.221*** (0.578)	3.877** (1.319)	2.562** (0.973)	2.413* (1.003)	3.957** (1.300)	2.430* (0.980)
Financial support	0.263* (0.108)	0.232(0.322)	0.253 +(0.129)	0.225(0.142)	0.293(0.325)	0.197(0.130)	0.154(0.162)
Preseason ranking	0.023(0.037)	−0.279(0.209)	0.048(0.044)	0.048(0.045)	−0.261(0.195)	0.048(0.047)
Draft value	0.091* (0.042)	0.316 +(0.163)	0.085 +(0.048)	0.086 +(0.048)	0.329* (0.165)	0.090 +(0.049)	−0.115(0.103)
Winning percentage (t)	−0.007(0.019)	0.055(0.048)	−0.010(0.028)	−0.013(0.027)	0.054(0.049)	−0.017(0.028)	−0.063 +(0.033)
Winning percentage (t) squared	−0.001** (0.000)	−0.002** (0.001)	−0.001 +(0.000)	−0.001 +(0.000)	−0.002** (0.001)	−0.001(0.000)	−0.000(0.000)
Winning percentage (t − 1)	0.020(0.018)	0.084(0.054)	0.010(0.021)	0.005(0.022)	0.094 +(0.053)	0.001(0.021)	0.052(0.044)
Winning percentage (t − 1) squared	−0.001* (0.000)	−0.001* (0.001)	−0.000 +(0.000)	−0.000(0.000)	−0.002* (0.001)	−0.000(0.000)	0.000(0.000)
Prior coach dismissed	−0.124(0.164)	−0.107(0.589)	−0.241(0.227)	−0.184(0.226)	−0.128(0.605)	−0.133(0.227)	0.059(0.429)
Prior coach person of color	0.191(0.178)	1.213* (0.612)	0.062(0.216)	0.054(0.223)	1.221* (0.571)	0.144(0.219)	0.984(0.761)
Prior coach winning percentage	0.013* (0.006)	0.029(0.024)	0.010(0.008)	0.009(0.008)	0.014(0.034)	0.046** (0.016)
Prior coach tenure	0.003(0.012)	0.027(0.063)	0.003(0.015)	−0.002(0.016)	0.027(0.066)	0.005(0.015)	0.018(0.045)
Prior coach gender	−0.468** (0.146)	−0.794(0.595)	−0.516** (0.167)		−1.948(1.508)	1.626* (0.804)
Female coach gap				0.038** (0.013)			−0.228(0.197)
Prior coach gender × prior coach winning percentage					0.023(0.031)	−0.041** (0.015)
Prior male coaches’ winning percentage							0.058(0.043)
Female coach gap × prior male coaches’ winning percentage							0.004(0.003)
Coaches	396	120	276	271	120	276	71
Dismissals	169	42	127	124	42	127	36
Observations	2445	831	1614	1580	831	1614	366
Log pseudolikelihood	−690.72	−104.14	−467.35	−454.76	−103.97	−464.69	−96.91
Wald chi-squared	429.43***	649.46***	380.75***	389.99***	664.41***	392.98	127.14***

Dummy variables for conference and year are included in models 1 through 6. Year dummies are not included in model 7, as multiple years do not have dismissals. Robust standard errors in parentheses. For model 7, New Conference, Scandal or Sanctions, and Preseason Ranking were collinear with the failure variable and were removed from the model.

*

p < 0.05, **p < 0.01, ***p < 0.001, ⁺p < 0.10050010001.

Model 1 of Table 2 reports results for the entire sample. Prior Coach Gender is negatively and significantly related to the likelihood of dismissal (p = 0.001). The exponentiated hazard coefficient for Prior Coach Gender is −0.468, which suggests that the likelihood of any coach being dismissed when the prior coach is female (with all other factors held equal) is reduced by 37.4%. We also note that, for the entire sample, we do not find a significant relationship between the current coach’s gender and the likelihood of dismissal.

H1 argues that a senior female manager will be less likely to be dismissed when she succeeds a female in the senior position. To test this hypothesis, we divided our sample into male and female coaches and repeated the analysis from model 1. As shown in model 2 (male coaches only), Prior Coach Gender is not significantly related to the likelihood of dismissal. However, model 3 (female coaches only) shows a negative and significant relationship between Prior Coach Gender and the likelihood of dismissal (p = 0.002). The exponentiated hazard coefficient for Prior Coach Gender in model 3 is −0.516, which suggests that the likelihood of a female coach being dismissed (with all other variables held equal) is reduced by 40.3% when the prior coach is female. H1 is supported.

H2 argues that the number of years that one or more male senior managers succeeded a female senior manager prior to the current female senior manager beginning her tenure is positively related to the likelihood of the current senior female manager being dismissed. To test this hypothesis, we replaced Prior Coach Gender with Female Coach Gap. As shown in model 4, Female Coach Gap is positively and significantly related to the likelihood of dismissal (β = 0.038; p = 0.004), which can be interpreted as each year that one or more male coaches succeeded a female coach is associated with a 3.9% increase in the likelihood of the current female coach being dismissed. ⁹ H2 is supported. ¹⁰

H3 argues that the likelihood of dismissal for female senior managers will be smaller when a prior female senior manager performed well compared to a prior male senior manager who performed well. To test this hypothesis, we again used our subsamples of male and female coaches. Model 5 shows that, for male coaches, the interaction of Prior Coach Gender and Prior Coach Winning Percentage is not significantly related to the likelihood of dismissal. However, model 6 shows that, for female coaches, the interaction of Prior Coach Gender and Prior Coach Winning Percentage is significantly and negatively related to the likelihood of dismissal (β = −0.041; p = 0.006).

Using methods from Lamina et al. (2012) and Wang and Ware (2013), this significant interaction is illustrated in Figure 1. If the prior coach had a winning percentage of 67.2% (one standard deviation larger than the mean for all prior coaches), the increase in the likelihood of dismissal for the current female coach was 107.1% if the prior coach was male. However, the increase was only 8.2% if the prior coach was female, suggesting that, when performance during the prior coach’s tenure was strong, the increase in the likelihood of the current female coach being dismissed is 98.9% less if the prior coach was female compared to the prior coach being male. Alternatively, if the prior coach had a winning percentage of 35.5% (one standard deviation less than the mean for all prior coaches), the decrease in the likelihood of dismissal for the current female coach was 51.7% if the prior coach was male and 7.6% if the prior coach was female. This suggests that, when performance during the prior coach’s tenure was weak, the decrease in the likelihood of the current female coach being dismissed is 44.1% less if the prior coach was female compared to the prior coach being male. To further validate our findings, we conducted simple slope tests. For male predecessors, the regression line is upward sloping and significant (p = 0.004). This indicates that male predecessors with strong performance (one standard deviation above the mean for all prior coaches) lead to a higher likelihood of dismissal for current female coaches compared to male predecessors with poor performance (one standard deviation below the mean for all prior coaches). However, for female predecessors, the regression line is slightly upward sloping, but not significant (p = 0.312). This result indicates that there is no significant difference in the likelihood of dismissal for current female coaches regardless of the female predecessor’s performance. Collectively, these empirical findings indicate that gender transference effects are stable and not dependent on the performance of the prior female coach. H3 is supported.

OPEN IN VIEWER

Figure 1.

Effect of prior coach winning percentage by gender on the likelihood of dismissal for female coaches.

H4 argues that when one or more men serve as a senior manager between two female senior manager tenures, the positive relationship between the number of years these men have served in the senior position and the likelihood of the current female senior manager being dismissed will be strengthened under stronger performance by these men. To test this hypothesis, we used a subsample comprised of the 71 female coaches in our main sample who had one or more male predecessors following the tenure of the team’s most recent female coach. As shown in model 7, the interaction of Female Coach Gap and Prior Male Coaches’ Winning Percentage is positive but not significant. H4 is not supported.

Supplemental analyses

To test whether our models violated the proportional hazards assumption, we executed two tests (see stata.com/manuals14/ststcoxph-assumptions.pdf; also, Allison, 2014; Garrett, 1997). First, Kaplan-Meier plots were generally parallel for different instances of the main effect in each model. Second, STATA’s phtest indicated that all variables in each model were not significant for violating the assumption. Thus, we are confident that our models did not violate the proportional hazards assumption.

We replaced the performance measure used in our primary analyses, Winning Percentage (linear and squared terms for the current and prior year), with Conference Winning Percentage (linear and squared terms for the current and prior year) and repeated our analyses. There were no meaningful differences in the results. We then repeated our analyses using two alternative measures of performance: a binary variable indicating whether the team made that year’s NCAA tournament and a count variable of the number of wins in that year’s NCAA tournament. Again, there were no meaningful differences in the results.

Given the possibility of a quadratic relationship between the number of years in which one or more men served as coach between the current and most recent female coach and the likelihood of dismissal, we repeated our H2 analysis using linear and squared terms for Female Coach Gap. The squared term for Female Coach Gap in this model was not significant.

As a coach may depart for a variety of reasons not associated with dismissal, we followed methods from Gentry et al. (2021) and completed a supplemental analysis using a subsample that excluded the 17 coaching tenures that ended with retirement, illness, or death. University announcements and media reports were used to determine if a coach’s tenure ended with illness or death, while a coach was assumed to have retired if he or she voluntarily left the position without taking another position at the age of 65 years or older. The results from this supplemental analysis did not meaningfully differ from the primary results. ¹¹

We completed a supplemental analysis using an alternative measure of the prior coach’s experience as a control variable. The primary results did not meaningfully change when Prior Coach Age at Departure replaced Prior Coach Tenure in the models.

To test whether an omitted variable may be influencing our results (see Bliese et al., 2020, for a review), we used STATA’s konfound command to assess the strength necessary for an omitted variable to invalidate our results (Frank, 2000). For the explanatory variable, Prior Coach Gender, at least 591 observations would need to be excluded due to bias to invalidate the results. For the explanatory variable, Female Coach Gap, 492 observations would need to be excluded due to bias to invalidate the results. For the explanatory variable, Prior Coach Winning Percentage, 712 observations would need to be excluded due to bias to invalidate the results. For the explanatory variable, Prior Male Coaches’ Winning Percentage, used only in model 7, 200 observations would need to be excluded due to bias to invalidate the results. As our models have a suitable and large set of control variables, the substantial number of cases needed to sustain an inference of omitted variable bias suggests that the results are not influenced by one or more omitted variables (Hubbard et al., 2017; Wangrow et al., 2023).

To examine whether the effects found in our analysis were consistent over time, we split the sample into two time periods (2004–2013 and 2014–2022) and repeated our analysis. ¹² Models 2 and 6 in Table 3 show that Prior Coach Gender is negatively and significantly associated with the likelihood of dismissal for female coaches for both periods (model 2: β = −0.635, p = 0.034; model 6: β = −0.773, p = 0.007). This suggests that, for the period of 2004 to 2013, immediately succeeding a female coach reduced the likelihood of dismissal for the current female coach by 47.0%, while, for the period of 2014–2022, immediately succeeding a female coach reduced the likelihood of dismissal for the current female coach by 53.8%. Models 3 and 7 in Table 3 show that the Female Coach Gap is positively and significantly associated with the likelihood of dismissal for female coaches for the two time periods. For the period from 2004 to 2013, the coefficient for Female Coach Gap is 0.057 (p = 0.042), while the coefficient for the Female Coach Gap is 0.038 (p = 0.022) for the period from 2014 to 2022. This suggests that the likelihood of female coach dismissal for each year in which one or more male coaches preceded a female coach increased by 5.9% during the period from 2004 to 2013 and by 3.9% during the period from 2014 to 2022. While model 4 in Table 3 shows no significant interaction, model 8 shows a significant interaction of Prior Coach Gender with Prior Coach Winning Percentage (β = −0.045; p = 0.025) for the period from 2014 to 2022. ¹³

OPEN IN VIEWER

Table 3.

Supplemental results with the sample split into two periods.

	2004–2013				2014–2022
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
	Full sample	All female coaches in the sample	All female coaches in the sample	All female coaches in the sample	Full sample	All female coaches in the sample	All female coaches in the sample	All female coaches in the sample
Variables
Gender	−0.108(0.340)				−0.117(0.328)
Person of color	0.337(0.329)	0.505(0.404)	0.476(0.413)	0.537(0.412)	−0.077(0.253)	0.361(0.403)	0.290(0.438)	0.316(0.414)
Former D1 player	0.160(0.284)	−0.169(0.356)	−0.357(0.357)	−0.186(0.353)	0.330(0.314)	0.407(0.418)	0.489(0.447)	0.339(0.396)
Alumni	−0.084(0.390)	0.555(0.455)	0.747(0.505)	0.647(0.468)	−0.151(0.338)	0.008(0.448)	−0.094(0.464)	0.206(0.449)
Previous head coaching experience	−0.044(0.028)	−0.040(0.033)	−0.020(0.032)	−0.036(0.033)	0.014(0.028)	−0.035(0.035)	−0.042(0.035)	−0.043(0.035)
Years assistant as school	−0.0170.045)	0.084(0.100)	0.078(0.108)	0.080(0.101)	0.041(0.076)	0.126(0.158)	0.122(0.159)	0.126(0.132)
Prestige when hired	0.016(0.113)	0.041(0.104)	−0.029(0.093)	0.049(0.104)	−0.122(0.120)	0.196(0.157)	0.209(0.166)	0.247(0.163)
Prestige at school	−0.056(0.160)	−0.214(0.192)	−0.331(0.207)	−0.194(0.193)	−0.057(0.164)	−0.595** (0.215)	−0.493* (0.225)	−0.593** (0.203)
Private university	−0.063(0.276)	−0.059(0.367)	0.012(0.381)	−0.068(0.363)	−0.207(0.275)	−0.148(0.367)	−0.244(0.422)	−0.101(0.351)
New conference	−0.965(1.111)	−1.083(1.215)	−0.929(1.135)	−1.169(1.249)	0.067(0.560)	−0.034(0.790)	0.505(0.832)	−0.350(0.728)
AD change	−0.239(0.256)	−0.116(0.311)	0.031(0.290)	−0.150(0.310)	−0.168(0.263)	−0.213(0.376)	−0.392(0.404)	−0.174(0.360)
Scandal or sanctions	3.525 +(1.984)	2.530(2.341)	2.421(2.542)	2.493(2.403)	2.307*** (0.508)	3.240*** (0.757)	3.010*** (0.757)	3.016*** (0.756)
Financial support	0.402(0.278)	0.533* (0.261)	1.084*** (0.326)	0.426 +(0.250)	0.399* (0.161)	0.490* (0.244)	0.325(0.247)	0.487* (0.242)
Preseason ranking	0.001(0.055)	−0.033(0.074)	−0.028(0.072)	−0.048(0.076)	0.021(0.059)	0.123 +(0.065)	0.122 +(0.065)	0.137* (0.067)
Draft value	0.074(0.061)	0.090(0.071)	0.088(0.071)	0.091(0.070)	0.122* (0.062)	0.103(0.098)	0.122(0.092)	0.111(0.101)
Winning percentage (t)	−0.051(0.033)	−0.068 +(0.039)	−0.057(0.043)	−0.075 +(0.039)	0.018(0.027)	0.093* (0.044)	0.092* (0.043)	0.083 +(0.046)
Winning percentage (t) squared	0.000(0.000)	0.000(0.000)	−0.000(0.001)	0.000(0.000)	−0.001** (0.000)	−0.002** (0.001)	−0.002** (0.001)	−0.002** (0.001)
Winning percentage (t − 1)	−0.023(0.033)	−0.048(0.038)	−0.069 +(0.041)	−0.059(0.037)	0.033(0.022)	0.021(0.026)	0.018(0.026)	0.013(0.026)
Winning percentage (t − 1) squared	0.000(0.000)	0.000(0.000)	0.000(0.000)	0.000(0.000)	−0.001** (0.000)	−0.001* (0.000)	−0.001* (0.000)	−0.001 +(0.000)
Prior coach dismissed	−0.160(0.277)	−0.212(0.420)	−0.268(0.425)	−0.170(0.409)	0.070(0.274)	−0.268(0.436)	−0.171(0.428)	−0.217(0.428)
Prior coach person of color	0.424	−0.455	−0.851*	−0.373	0.157	0.561 +	0.513	0.697*
	(0.380)	(0.425)	(0.431)	(0.461)	(0.249)	(0.334)	(0.336)	(0.349)
Prior coach winning percentage	0.020 +(0.010)	0.016(0.013)	0.011(0.014)	0.048(0.030)	0.003(0.008)	−0.005(0.014)	−0.004(0.014)	0.031(0.020)
Prior coach tenure	0.015(0.028)	0.015(0.027)	0.018(0.029)	0.014(0.028)	0.003(0.020)	0.004(0.028)	−0.003(0.028)	0.009(0.028)
Prior coach gender	−0.490 +(0.255)	−0.635* (0.300)		1.180(1.568)	−0.507* (0.225)	−0.773** (0.286)		1.497(1.029)
Female coach gap			0.057* (0.028)				0.038* (0.017)
Prior coach gender × prior coach winning percentage				−0.033(0.029)				−0.045* (0.020)
Coaches	241	168	165	168	262	176	172	176
Dismissals	81	65	64	65	88	62	60	62
Observations	1173	811	785	811	1272	803	795	803
Log pseudolikelihood	−270.38	−196.79	−191.22	−196.10	−293.23	−179.05	−172.74	−177.79
Wald chi-squared	444.65***	691.16***	740.99***	690.62***	253.91***	231.81***	295.28***	283.05***

Dummy variables for conference and year are included in each model. Robust standard errors in parentheses.

*

p < 0.05, **p < 0.01, ***p < 0.001, ⁺p < 0.10050010001.

Finally, as this study examines transference’s relationship with the likelihood of dismissal, we considered a subsequent question of whether transference is associated with future performance. This supplemental analysis found that, for new female coaches, the prior coach being female (i.e. Prior Coach Gender) was significantly and positively related to performance (i.e. winning percentage) in the first year (β = 5.171; p = 0.035). This supplemental analysis further found that, for new female coaches, the number of years in which one or more men served as coach prior to a female coach beginning her tenure (Female Coach Gap) was significantly and negatively related to performance in the first 3 years of her tenure (year 1: β = −0.395; p = 0.050; year 2: β = −0.433; p = 0.080; year 3: β = −0.504; p = 0.044). These results offer some evidence that stereotyping and bias may be reduced under transference. It is possible that transference may also facilitate a more favorable environment (i.e. more supportive) for incoming female leaders that fosters success early in their tenure. However, perceptions of a senior female manager’s performance and her actual performance may diverge. Research has found that women may need to outperform their male counterparts to be viewed as competent, and such attributions can lead to evaluators and other key stakeholders providing less support to female leaders (Cowen et al., 2022; Ma et al., 2022). However, this prior literature does not take into account transference effects or prior experience with women in senior positions. We encourage future research that delves deeper into performance consequences and also examines other outcomes from transference, such as turnarounds and the ability to access, develop, and retain human capital.

Detailed results from the supplemental analyses are available from the authors upon request.

Practical implications

Our study’s findings point to situations that can increase or decrease career disadvantages associated with gender bias and stereotyping. Hence, we offer some guidance to practitioners on potential strategies to ensure gender equality in organizations, with recommendations for potential remedies that can prevent career disadvantages for female leaders. First, boards and other governance bodies should be cognizant that while appointing women to leadership positions could foster reductions in subsequent gender bias in organizations, this is not a permanent remedy, and gender bias can reemerge (via appointment of a male successor). Thus, succession from one female senior manager to another can be viewed as a long-term and intentional process toward gender equality. Second, when boards and other governance bodies (in this study’s context, a university’s president and its board of trustees) select their new members, they could consider whether candidates have recent experience and interaction with women in senior positions. While these candidates should be assessed by their potential to effectively provide counsel and support as well as monitor the organization and its leaders, boards and other governance bodies could also assess candidates’ motivations, abilities, and experiences to support the advancement of high-potential leaders from all demographic groups. Such support would be evident if candidates, within their current or former organizations, have continuously engaged with and supported women who led key businesses and functions. Third, organizations could continuously support a managerial pipeline, in which high-potential female leaders (as well as high-potential candidates from other occupational minority groups) are given opportunities to lead essential businesses, functions, and initiatives within the firm. As part of this initiative, organizations could ensure that the work of high-potential female leaders is visible and that they are given access to evaluators and other key stakeholders (Mah et al., 2023). Relatedly, evaluators and key stakeholders could seek closer relationships with these women to find similarities and common ground, as this will foster the female leaders being part of the in-group (Zhu et al., 2014). Beyond the value of this initiative to high-potential female leaders, it may also reinforce to all stakeholders that women can effectively fill leadership roles in the organization. Finally, organizations could ensure female presence in leadership roles by including successful female predecessors on search and hiring committees tasked with finding the organization’s next leader (Dwivedi and Paolella, 2024), especially in cases of interrupted gender transference. These predecessors could actively promote and advocate for future female leaders and thus facilitate greater gender equality across the upper echelons of organizations.

Limitations

It is important to note that our findings are in a context, women’s college basketball, that differs from the typical male-dominated corporate setting. While the sports context has been successfully utilized in prior research to examine leader bias (e.g. Carton and Rosette, 2011; Obenauer and Langer, 2019; Wangrow et al., 2023) and enabled empirical testing of our article’s hypothesized relationships, we acknowledge that the extent and intensity of gender stereotypes for female basketball coaches may be different than what female leaders experience at the upper echelons of organizations. We believe that, as the number of women in top corporate positions continues to increase, it is necessary for future research to examine the applicability of gender transference effects in upper echelon contexts. While weaker effects (than our findings) would suggest the continued existence of bias and hurdles that women face when trying to reach upper echelons, stronger effects could point to reductions in gender bias and more equitable conditions for women.

This study utilizes a variety of archival data sources to develop measures for our variables, and we acknowledge limitations from this approach. For example, Draft Value is a measure of human capital strength that is derived from whether a team’s players in a given season were later drafted into or played in the WNBA. Methods, using a combination of surveys, interviews, and quantitative performance assessments, could assess a wider range of players’ capabilities, experience, and knowledge. In addition, as prior studies that include prestige as a measure of reputational capital and power have suggested, “future research could examine truer values of reputation . . . and examine how reputational capital from major accomplishments may decline over time” (Wangrow et al., 2018: 345; see also Allen et al., 2022). Further, Financial Support measures a university’s annual expenditures on its women’s college basketball program but does not include capital spending (e.g. for new or remodeled facilities)—data that is not made available by most universities. Future research that includes all forms of spending may offer further insights into contexts associated with female senior manager dismissal.

We model dismissal as the dependent variable. Like CEOs, many coaches’ contracts include bonuses for excellent performance and buyout clauses in which they receive substantial payments for several years after dismissal. Future research that includes contractual terms associated with bonuses and buyouts may provide greater nuance for understanding the dismissal of both male and female senior managers.

Our sample was based on US data. We encourage future research to examine whether our findings are generalizable to international settings. As countries vary in gender parity scores and female access to business employment and empowerment (Hausmann et al., 2012), a cross-national study could compare which countries exhibit less (greater) bias toward female senior managers and provide the most favorable (unfavorable) conditions for retention of female leaders in organizations.

While the focus of this article was on behavioral and not structural elements influencing dismissal, the coach’s gender and the number of years between female coaches could be impacted by internal practices (e.g. selection, training, development, and promotion policies; Oakley, 2000) that foster stereotypes or serve as barriers to promotion within the organization. Biases could be inherent in these internal practices and influence evaluators’ dismissal decisions. Future research could examine how these internal practices interact with behavioral explanations to impact succession decisions. For example, in cases of interrupted gender transference, it would be interesting to examine how gender-inclusive behaviors by male predecessors (Dwivedi et al., 2018) and internal diversity practices (e.g. diverse search committees; see Kazmi et al., 2022) jointly impact leader gender succession.

Finally, the underlying mechanism in our theorizing was built on the assumption that evaluators exhibit gender bias and stereotyping toward female senior managers. Yet, we do not directly measure bias and stereotypes, as research designs using archival and secondary data can only proxy for and infer perceptions and cognitive evaluations. We encourage future research that uses surveys and experimental approaches for directly assessing how and why evaluators develop and exhibit stereotypes.