Institutional predictors of campus sexual misconduct reporting: the role of gender in leadership

ABSTRACT

Sexual misconduct remains at crisis levels on American college campuses and is vastly underreported. Most research focuses on individual level risks of assault and perpetration, yet campuses vary significantly in sexual offenses and reporting rates. The current study responds to calls to consider institutional factors that shape campus climates for sexual misconduct and reporting. We consider the role of gender in leadership, analyzing the impact of women's appointment as college president on reporting rates over time. We consider individual and organizational mechanisms that may shape the impact of women's appointment on reporting rates. Our analyses rely on data from U.S. colleges and universities between 2005-2020. We find that the appointment of women presidents is associated with a significant increase in reporting of sexual misconduct. We observe a significant increase in reporting during the first year following the appointment and a continued increase in reporting rates over time.

KEYWORDS:

• Sexual misconduct
• leadership
• gender
• reporting

Sexual offenses remain at crisis levels on college campuses in the U.S. (AAU 2019). Despite renewed efforts to address this crisis, rates of sexual misconduct among American college students remain unacceptably high. By the end of their college careers, one third of women will experience sexual assault (Hirsch and Kahn 2020), and one in four men will self-report having committed sexual violence (Anderson et al. 2021). At highest risk of assault are LGBTQ + students, particularly bisexual women and gender non-conforming individuals, as well as disabled and Indigenous students (Coulter and Rankin 2020; Findley, Plummer, and McMahon 2016; Ford and Soto-Marquez 2016; Patterson Silver Wolf et al. 2016).

Despite its prevalence, sexual assault generally and campus assault specifically is one of the most underreported violent crimes (Cantor et al. 2015; Fischer, Francis, and Turner 2000; Tjaden and Thoennes 2006; Truman and Morgan 2016). Survivors often do not report out of fear that the process will be stigmatizing and punitive and will lead to revictimization and inaction (Ahrens 2006; Krebs et al. 2016; Mancini et al. 2016; Smith and Freyd 2014).

Most research on campus sexual violence focuses on individual and demographic predictors of risk, protection or perpetration (O’Connor et al. 2021). Such studies explore individual attitudes, substance use, peer effects and social behavior to understand who is at highest risk of assault and who is most likely to perpetrate sexual violence. This individual-level focus is also mirrored in common policy approaches to prevention, such as mandated trainings, which focus overwhelmingly on changing individual attitudes and behaviors (McMahon 2015). Research on the efficacy of such efforts suggest they have minimal effect on assault prevention or reporting rates (Anand and Winters 2008; Robb and Doverspike 2001).

The lack of significant progress on prevention combined with increasing recognition of variation in sexual offenses between and among campuses (Cantor et al. 2015; Lee and Wong 2017; Muehlenhard et al. 2017) has strengthened calls to move beyond individual risk assessment and prevention to focus on institutional factors that contribute to campus climates where sexual misconduct is more or less likely to occur. A recent review of the field by O’Connor et al. (2021, 6) identified ‘major gaps in understanding’ with regard to the institutional settings in which sexual misconduct occurs. Only by adopting an institutional lens can we identify factors that explain variation across campuses (Moylan and Javorka 2020), thus enabling more efficacious preventative efforts.

The current study responds to calls to analyze impacts and outcomes related to campus sexual misconduct using an institutional lens (Lee and Wong 2017; Moylan and Javorka 2020; Moylan et al. 2019). While scholars have begun to identify mechanisms that shape institutional responses and contribute to institutional variation in misconduct and reporting, no previous studies have analyzed how the gender composition of college administrators shapes reporting outcomes. We fill this gap by analyzing how women's appointment to the role of president on American college and university campuses impacts reporting rates over time. We rely on data from U.S. colleges and universities over a fifteen-year period (2005-2020). Holding institutions constant, we compare reporting rates during the three years prior to women's appointment and during the three years following the appointment to evaluate whether women's appointment impacts reporting rates and, if so, whether this effect grows over time as women's tenure matures.

By focusing on institutional leadership, gender, and power, our analysis contributes to and advances our understanding of how institutional factors shape campus climates for reporting. Though reporting does not always or necessarily benefit survivors, reporting is critical to efforts to prevent sexual offenses because it serves as a requisite precondition for convicting or otherwise holding perpetrators accountable (Bachman 1998). Thus, by serving as a critical mechanism for transparency and accountability, reporting rates are an important tool in reducing sexual misconduct and aiding prevention efforts.

Gender, leadership and reporting

Reporting campus sexual misconduct¹ is a critical mechanism of prevention and support because it enables transparency and accountability for perpetrators and facilitates support, service provision, and outreach for survivors. Title IX requires universities to connect survivors who report misconduct to campus-based services, including counseling, advocacy, legal assistance, and other services. Research finds that reporting misconduct is critical for survivors’ well-being, leading to better physical and mental health outcomes (Arens, Stansell, and Jennings 2010). Indeed, accessing campus-based supports are associated with reduced self-blame, trauma, and distress for victims of misconduct (Eisenberg et al. 2016). Reporting is also necessary for prevention of misconduct because it allows universities pursue disciplinary action against perpetrators, thereby preventing repeat offenses (U.S. DOE 2021).

Despite the importance of reporting for prevention and support, sexual misconduct remains significantly underreported on college campuses (Fischer, Francis, and Turner 2000; Lindquist, Barrick, and Krebs 2013; Paul et al. 2013; Sabina and Ho 2014; Truman and Morgan 2016; Wolitzky-Taylor et al. 2011). Survivors’ reluctance to report or to seek campus-based services is rooted, in part, in distrust of institutional processes. In interviews and surveys, survivors report concerns regarding confidentiality and accountability; they fear their identities will become known to perpetrators and they lack confidence that perpetrators will be held accountable (Krebs et al. 2016; Sable et al. 2006; Ullman 2000). Survivors also fear that reporting to campus authorities will subject them to skepticism, blame and scrutiny as well as revictimization; survivors fear they will not be believed and will lack autonomy over the investigatory and disciplinary processes (Ahrens 2006; Mancini et al. 2016).

While fear of institutional betrayal can prevent reporting (Smith and Freyd 2014), there is evidence that reporting varies significantly by campus (Cantor et al. 2015). For instance, perceptions of campus climate can suppress or support reporting. Students who perceive that their campus is supportive of sexual violence survivors are more likely to report while students who perceive the campus climate is hostile are less likely to report (Holland and Cortina 2017; Spencer et al. 2017; Walsh et al. 2010). Campus policies can also impact reporting rates. Policies that enhance survivor autonomy by centering their decisions about reporting increase reporting rates (Holland, Cortina, and Freyd 2018).

Leadership composition and accountability are powerful mechanisms for shaping institutional climates related to sexual misconduct (Tenbrunsel, Rees, and Diekmann 2019). Leaders impact organizational cultures by shaping policy and enforcement and by influencing perceptions of risk and safety (Schaubroeck et al. 2012). Women's entrance into leadership ranks can alter organizational cultures by shifting organizational norms, reducing organizational tolerance for misconduct and disrupting hyper masculine beliefs (Acker 1990; Roth 2007; Willness, Steel, and Lee 2007). Indeed, in the wake of the #MeToo movement, calls for greater representation of women in positions of authority have increased, premised on the notion that women are more effective champions of prevention and equity (e.g. Cooper 2017).

To interrogate these claims, we evaluate the impact of the appointment of women presidents on sexual misconduct reporting over time. Our theoretical framework considers two mechanisms that may contribute to higher reporting rates in institutions headed by women. First, we review potential individual mechanisms motivated by role congruity and gender role identity and, second, we consider organizational mechanisms motivated by critical mass and representational leadership.

As individuals, women presidents likely face pressure to address campus sexual misconduct and may be more motivated than men presidents to do so. Women in highly visible senior leadership roles often experience token pressures, including scrutiny and heightened performance expectations (Kanter 1977; Holgersson and Romani 2020; Read and Kehm 2016). Token biases combined with role congruity expectations may pressure women leaders to pursue organizational transformation and champion issues or policies supportive of women and gender equity (Eagly and Carli 2007; Eagly and Karau 2002). Such pressures may be enhanced when it comes to issues of sexual misconduct including sexual harassment and assault – policy areas where women are expected to lead (Kindelin 2017). Women leaders are also more likely than men to be held accountable or blamed for organizational failures, including crises and scandals that predated their tenure (Davidson-Schmich, Jalalzai, and Och 2023; Ryan and Haslam 2005). Combined, these pressures may influence the degree to which women presidents prioritize strengthening sexual misconduct policies, increasing awareness of sexual violence and supporting zero-tolerance policies (Bell, McLaughlin, and Sequeira 2002; Roth 2007; Willness, Steel, and Lee 2007).

Women's gender role identity may also increase their motivation to improve the organizational climate in ways that reduce sexual misconduct and increase reporting. Generally speaking, women leaders are associated with equity policies and practices, including transparency and accountability (Cook and Glass 2018; Eagly and Carli 2007; Fine, Sojo, and Lawford-Smith 2020). Because women are more likely than men to have personal experience with sexual harassment, abuse and assault, they tend to have a broader definition of sexual misconduct (Quick and McFadyen 2017), are more supportive of prevention programs (Worthen and Wallace 2017), are better trained to recognize sexual misconduct (McLaughlin, Uggen, and Blackstone 2012) and, as leaders, are more likely to expand the range of behaviors that are prohibited or subject to sanction (Berdahl and Aquino 2009).This heightened awareness and commitment may lead them to support more efficacious and comprehensive policies over time.

These individual level mechanisms can strengthen help seeking and reporting behaviors in a variety of ways. By supporting more inclusive organizational climates, including by strengthening LGBTQ + friendly and gender equity policies (Cook and Glass 2016; Cohen and Huffman 2007), women leaders can contribute to lower rates of misconduct and higher rates of reporting (Coulter and Rankin 2020). Indeed, previous research finds that the efficacy of many policy approaches is dependent on the commitment of senior administrators. Moylan and Javorka (2020) find that campus administrators who oversee sexual violence prevention efforts significantly influence the efficacy and impact of these programs. Thus, the commitment and willingness of women presidents to implement and enforce anti-misconduct policies and practices can influence the nature of these policies and their impacts, resulting in higher rates of reporting.

Beyond individual motivations to affect campus climate, the appointment of women presidents may also contribute to organizational changes that further advance anti-misconduct efforts and increase reporting. First, women leaders are associated with expanded opportunities for other non-traditional leaders, including women and people of color (Kurtulus and Tomaskovic-Devey 2012; Maume 2011). When women occupy positions of authority, organizations have a stronger track record of appointments in other key roles. A greater representation of women and people of color in various roles (e.g. general counsel, Title IX coordinator, Vice President) can enhance the efficacy of women leaders by providing peer support and other resources necessary to effect organizational change (Higgenbotham and Weber 1999; McGuire 2002; Taylor 2010). In contrast, organizations with male dominated leadership structures are associated with higher rates of sexual misconduct and lower reporting (Mengling et al. 2014). Organizations without women in visible leadership roles are perceived to be more tolerant of sexist behavior, and individuals in such organizations are less likely to report abuse (Buchanan et al. 2014; Willness, Steel, and Lee 2007).

A critical mass of women and people of color in key roles can also increase coordination between and among units on campus and within the community, which can improve survivor outreach and support and, by extension, increase reporting and other help-seeking behaviors (Greeson and Campbell 2013). Research on women's representation in university administration finds that, compared to men, women administrators are more likely to seek collaborative and information-sharing relationships across units (Bobe and Kober 2020), which may enhance their ability to transform their institution's response to sexual misconduct. Previous research also finds that a greater representation of women in university administration reduces harassment claims, which can signal a more inclusive and supportive climate for women overall (Glass, Cook, and Pierce 2020).

Women's representation in leadership can also impact organizational culture in significant ways (Campuzano 2019). When women hold highly visible roles like the president, transparency and accountability across the organization increase (Glass and Cook 2018). For instance, women's leadership is associated with greater perceptions of humane treatment of peers (Jayasinghe 2020), a greater sense of egalitarianism (Herrera et al. 2012), and a greater commitment to collaboration (Fine 2009). Perceptions of campus cultures are critical predictors of reporting and other help seeking behavior. Perceptions of an organization's tolerance for abuse includes not only beliefs that perpetrators will be held accountable but also that the organizational overall is committed to equity and fairness (Willness, Steel, and Lee 2007). We expect that the visibility and leadership priorities of women presidents are more likely than those of men presidents to signal greater intolerance for misconduct and abuse. And, indeed, individuals who perceive a positive campus climate are significantly more likely than others to report their experience of abuse (Spencer et al. 2017).

Overall, we expect that the status of women within an organization, including women's representation in leadership, will be a strong predictor of reporting because survivors feel safe disclosing and more confident in the process (Boyle, Barr, and Clay-Warner 2017). Indeed, women's visible leadership can serve as an ‘accountability cue’ to survivors, signaling that the organization is responsive and will act in accordance with its policies (McMahon 2015). Perceptions of an equitable and fair campus culture also increases institutional trust, which can enhance the intention to report and to seek support and other resources (Holland 2020).

Policy context of prevention in U.S. higher education

While researchers have explored the policy context surrounding campus sexual misconduct beyond the U.S. (Lee and Wong 2017), the American policy context is unique in its standardization at the federal level. American institutions of higher education that receive federal funding are bound by Title IX of the Education Amendments of 1972, a federal civil rights law that prohibits discrimination – including sexual misconduct – in education. Title IX mandates that institutions establish procedures for addressing sexual misconduct, creating (at least in principle) stronger institutional mechanisms for limiting misconduct compared to non-educational institutions. While prevention efforts on college campuses are often more compliance-based than efficacious (Feldblum and Lipnic 2016), these efforts represent an institutional requirement that binds colleges and universities to a common set of reporting and remedial actions.

Prevention efforts in higher education have been subject to increased scrutiny in recent years. In 2011, the U.S. Department of Education's (DOE) Office of Civil Rights issued a Dear Colleague Letter (DCL) to American colleges and universities, reiterating their obligations to prevent and address sexual discrimination, including assault and harassment, under Title IX. The letter's purpose was to provide guidance to college and university administrators regarding the DOE's interpretation of Title IX. Specifically, the letter urged institutions to provide better investigation and adjudication of sexual assault and harassment claims in order to avoid a federal investigation. While many administrators have argued that the guidelines issued in 2011 lacked clarity and consistency (New 2016), the DCL significantly increased campuses’ attention to existing policy enforcement and led to significant amendments of existing policies and procedures (Wilson 2017).

In 2013, the Reauthorization of the Violence Against Women Act (VAWA) included an amendment titled the Campus Sexual Violence Elimination Act (SaVE). The SaVE expanded the Clery Act to include sexual misconduct disclosures at institutions of higher education. The Jeanne Clery Disclosure of Campus Security Policy and Campus Crime Statistics Act (Clery Act) aimed to increase transparency of crime statistics on college campuses in order to hold institutions accountable for ensuring the safety and security of students. It requires institutions receiving federal funding to share crime statistics publicly and with the DOE. When VAWA was reauthorized in 2013, new provisions were added that extended the Clery Act to policies and statistics related to sexual misconduct. Compliance with federal guidelines remains uneven however and, when institutions are found in violation, they may be audited by the Department of Education (Yung 2015). One study found that reports of sexual misconduct reporting increased during the period of an audit but returned to pre-audit levels following the completion of the audit (Yung 2015).²

Data and methods

Procedure

The data necessary to analyze our research questions were gathered from multiple sources. Our primary data source is the U.S. Department of Education OPE Campus Safety and Security Statistics database (https://ope.ed.gov/campussafety/#/datafile/list). This database includes annual submissions of all crime and violence against women statistics for postsecondary schools that receive federal Title IV funding. The reporting of these statistics to the Department of Education is required under the Clery Act, the Campus Sexual Violence Act (SaVE Act) and the Higher Education Opportunity Act. Data for sexual offenses on the Campus Safety and Security website is available starting in 2005, with the availability of the report delayed two years (for example, in January 2023, the most recent report was 2021 which offered reporting from 2020). Our sample uses data gathered from 2005 through 2020, which is the last year for which OPE Campus Safety and Security Statistics are available via the 2021 report.

We first queried a list of all OPE Campus Safety and Security Statistics reporting institutions, which provided a list of over eleven thousand entries. Within that list, multiple campus locations for many schools were included. Once multiple entries were removed, this resulted in 7051 unique institutions and 3004 institutions offering 4-year and graduate programs. At the onset of our collection, we examined each institution to determine if a woman was appointed president during our period of analysis. To be included in our examination, a man had to serve as president or chancellor of the institution prior to the woman's appointment in order to best assess the potential impact of a woman president. If a woman followed a woman as president, she was not included in the analysis. If a woman followed a man as president but his tenure had been shorter than three years and he was preceded by a woman, that appointment was not included in the analysis. For example, Paula Allen Meares served as Chancellor for University of Illinois at Chicago from 2009 until 2015. Her predecessor was Eric Gislason; however, he only served in the position for two years. Prior to his appointment, Sylvia Manning was Chancellor. Given the possibility that the UIC climate had been impacted by Chancellor Manning with limited time to be affected by Chancellor Gislason, this would not be an adequate test of our research question. To offer the most appropriate analyses, we needed clear differentiation between men and women as president. As such, this case was not included. We set the criterion of three years for a man to serve as president before the incoming woman president for two reasons: one, it serves for a balanced dataset since we examined the sexual misconduct in one-, two-, and three-year timeframes; and two, it offers a reasonable timeframe to impact the climate of the school if a woman had been a president prior to the man's appointment. Our data are also limited to institutions offering four-year and graduate programs. Information on two-year institutions was not sufficient for analysis. Additionally, the lack of students living on campus at two-year institutions may cause inaccuracies in the data due to multiple years of zero reported offenses. This resulted in a sample size of 294 women presidents or chancellors.

Our goal was to examine data for three years following the appointments of women presidents. The year of the appointment or the transition year was not included. For example, Colleen Hanycz became President of La Salle University in July, 2015. We analyzed data of sexual offenses for 2016, 2017, and 2018. Of the 294 women, 235 had three years of data available to examine. This was a result of a few factors. One, if the woman was appointed in either 2006 or 2007, we only had data starting in 2005 for our matched sample; hence, we only included the corresponding one or two years of data for that appointment. Two, if the woman was appointed in either 2018 or 2019, we only had data available through 2020. And the last factor was if the woman did not serve for three years beyond her appointment. To include as many women as possible, we opted to present one-year, two-year, and three-year models. For instance, Sheila Bair became President of Washington College in May, 2015 and resigned her post in June, 2017. Only 2016 is used for our analysis. So, our sample sizes are as follows: for the one-year model 588 (294 women and 294 men), for the two-year model 534 (267 women and 267 men), and for the three-year model 470 (235 women and 235 men).

For the matched sample of men presidents, we used the same institutions and analyzed the corresponding number of years (one, two, or three) prior to the appointment year of the woman. As illustrated above, Sheila Bair became President in 2015 and resigned in 2017. As such, we analyzed 2014 (matched men's sample) and 2016 (women's sample) data from Washington College. Because data were limited for her, we kept our sample balanced by only including one year on the matched sample. This method provides us with balanced data for pre and post the transition year. By matching the women with the men from the same institution, we control unobservable factors that may impact the reporting of sexual offenses. Our aim is to keep as many factors constant while discerning the potential impact of the President's gender.

Measures

Dependent variables

Clery Act sexual offenses. The U.S. Department of Education OPE Campus Safety and Security Statistics database was used to find the number of reported incidents of sexual violence on campus. This database includes annual submissions of all crime and violence against women statistics for postsecondary schools that receive federal Title IV funding. The reporting of these statistics to the Department of Education is required under the Clery Act, the Campus Sexual Violence Act (SaVE Act) and the Higher Education Opportunity Act requirements. The crime statistics reported under the Clery Act and gathered for our analyses include both forcible and non-forcible offenses. Forcible offenses defined by the Clery Act are rape, fondling, sexual assault with an object and forcible sodomy. Incest and statutory rape are considered the non-forcible offenses. The Clery Act includes definitions of forcible and non-forcible sexual offenses that may vary from those common to sexual assault law. The Clery Act defines ‘forcible’ offenses as either involving use of force or a nonconsensual act. Criminal statutes often treat force as an element in addition to consent. Our sample includes a tally of all offenses, both forcible and non-forcible, that occurred on campus, including within the residence halls.

Violence Against Women Act (VAWA) offenses. As a supplemental examination to our primary sample, we also analyzed VAWA offenses for the years available. These data are also available through the OPE Campus Safety and Security Statistics database. VAWA offenses include domestic violence, dating violence and stalking. Reporting of these offenses began in 2014. Given our balanced matched sample approach, this resulted in analyzing transitions that took place between 2015 through 2019. One, two, and three-year models are presented. We keep VAWA and Clery Act offenses separate within our data due to the differences in reporting practices.

Independent variable

Woman president or chancellor. This item is dichotomous and reported as 1 if the president or chancellor of the university or college is a woman. Again, we created a balanced matched set which resulted in the same number of men and women represented in the analyses.

Control variables

Urban or rural designation. A variety of sources were used to determine the geographic nature of each institution. First, we searched U.S. News and World Report school statistics. In its overview section, it listed the majority of the schools as urban, suburban or rural. Those listed as city or small city by U.S. News and World Report, were further researched using both individual institutional websites and geographic websites and classified according to population compared to U.S. Census designations. Suburban was then combined with urban, as they were highly similar in their correlation with sexual offenses within the data. Urban is coded as 1, rural as 0.

Division I athletic programs. Schools with Division I athletic programs were coded as 1 in the analyses. Prior research has suggested that Division I schools report higher rates of violence against women as compared to schools in lower divisions (Wiersma-Mosley and Jozkowski 2019). Division I schools were identified through searches in U.S. News and World Report school statistics in its campus life section.

Elite institutions. The acceptance rate at each institution was recorded from U.S. News and World Report school statistics in its admissions section. Schools with less than a thirty percent acceptance rate were coded as elite institutions. We control for elite institutions due to a large number being under investigation for prior violations of Title IX and mishandling sexual misconduct complaints (Anderson 2014).

Percentage of students living on campus. We used institutional websites, U.S. News and World Report school statistics section on campus life, and univstats.com to determine the percent of students that live on campus. This control helps to distinguish between those schools with larger campus presence versus online or commuter schools. Studies suggest that reporting of sexual misconduct statistics may be more problematic for schools without a large on campus living presence (Yung 2015).

Total student population. Student population was queried from the U.S. Department of Education OPE Campus Safety and Security data. We performed a natural log transformation on student population to account for wide variance within the sample.

US Census region. We dummy coded and controlled the four US Census regions of West, Midwest, South, and Northeast. For space purposes, the regions were controlled but not reported in the analyses.

Year of appointment. We dummy coded and controlled the year of the appointment for the woman president. Appointment years ranged from 2006 to 2019. For space purposes, each year was controlled but not reported in the analyses. We also created a time trend variable to determine if that provided a better model fit than dummy coding each year. The results for our independent variable were unchanged from the original model, and the time trend variable's model fit was less efficient.

Analyses

We examined our research question using Welch's ANOVA and negative binomial regression. Welch's ANOVA provides the mean comparison between men and women presidents and the corresponding sexual offenses reported at their institutions. Using Welch's ANOVA rather than a traditional ANOVA is preferable for our data given some models failed the homogeneity of variance test and Welch's ANOVA does not have the base assumption of equal variances. For purposes of consistency, we used Welch's ANOVA for all our mean comparisons. Our Welch's ANOVA is a between-subjects analysis, and it is more appropriate than a within-subjects design given we are investigating differences between men and women and holding the institution in which they serve constant. For our full models with our control variables, we used negative binomial regression. This method is appropriate for our count variable dependent variables. Additionally, the negative binomial regression fit our model better than a Poisson regression given the greater variance and overdispersion present in our dependent variables. For the supplemental analysis of the VAWA data, we used a Poisson regression. It was more appropriate than a negative binomial regression given overdispersion was not present within this smaller dataset.

Results

Our analyses test the research question of whether women, as presidents of universities, impact the reporting rates of sexual offenses, and if so, whether this effect increases over time as their tenure matures. Descriptives and correlations of the examined variables for the primary dataset reporting Clery Act offenses are presented in Table 1. The descriptives and correlations, Welch's ANOVA, and the Poisson regression for the supplementary dataset reporting VAWA offenses are presented in Appendix A. As illustrated in the correlations table (refer to Table 1), women as presidents have a significant and positive relationship with higher rates of reporting of sexual offenses in all examined timeframes (one, two, and three years post their appointment). The control variables of urban campuses, Division I athletic programs, elite institutions, and total student population are also positive and significant across all examined timeframes (refer to Table 1).

Table 1. Descriptives and Correlations. Clery Act Sexual Offenses.

Variable	n	Mean	s.d.	1	2	3	4	5	6	7	8
1. Woman President	588	.50	.50	– –
2. Urban Campus	588	.69	.46	.00	– –
3. Division I Athletic Program	588	.25	.43	.00	.28**	– –
4. Elite Institution	588	.07	.26	.00	.11*	.02	– –
5. Percent Students Living on Campus	588	45.6	26.64	.00	-.15**	-.22**	.30**	– –
6. LN of Total Student Body	588	8.23	1.27	.00	.17**	.64**	.03	-.40**	– –
7. Number of Sexual Offenses for One Year	588	4.48	8.18	.08*	.11*	.39**	.21**	.04	.46**	– –
8. Number of Sexual Offenses for Two Years	534	8.47	11.48	.13**	.11*	.41**	.22**	.08	.46**	.95**	– –
9. Number of Sexual Offenses for Three Years	471	12.76	17.37	.18**	.12*	.42**	.19**	.06	.48**	.91**	.96**

* p < .05 ** p < .01

In the mean comparison analysis, men and women presidents have statistically significant differences in reporting rates of sexual offenses (refer to Table 2). Table 2 reports the findings from the Welch's ANOVA. Specifically, in the first year following the appointment, women presidents have reports averaging 4.91 incidents per year, whereas in the year prior to the appointment, men presidents have reports averaging 3.81 incidents. This difference in the one-year timeframe is significant at p < .05. In examining the two-year timeframe, women presidents have reports averaging 9.98 incidents for those two years whereas men presidents have reports of 6.96 incidents. This difference is significant at p < .01. And in examining the three-year timeframe post appointment, women presidents report 15.91 incidents whereas men presidents report 9.67 incidents for the three years prior to the appointment. This results in a significant difference of p < .001. Consistent with our research question, women presidents have higher reporting rates and those rates tend to increase with the increased tenure of the women (refer to Table 2).

Table 2. Welch's ANOVA. Clery Act Sexual Offenses.

	Woman as President					Man as President
Variable	F-Value	N	Mean	SD	N	Mean	SD
One Year	4.14*	294	4.91	7.21	294	3.81	5.82
Two Years	9.41**	267	9.98	12.83	267	6.96	9.73
Three Years	15.66***	235	15.91	19.87	235	9.67	13.81

*p < .05 **p < .01 ***p < .001

*Welch's test was used to account for unequal variances between groups; asymptotically distributed F-value reported

The negative binomial regressions, after controlling for all variables of interest, also indicate support that women presidents have higher reporting rates than men presidents, and that those rates increase with the tenure of the women presidents (refer to Table 3). As illustrated in Table 3, women presidents are positively and significantly related to reporting rates. In the one-year window, women presidents are positively and significantly (p < .01) related to higher rates of reported sexual offenses with an odds ratio of 1.30. Looking at the two-year window, women presidents are positively and significantly (p < .001) related to higher rates of reported sexual offenses with an odds ratio of 1.40. And in the three-year window, women presidents are positively and significantly (p < .001) related to higher rates of reported sexual offenses with an odds ratio of 1.62. As demonstrated by the increasing odds ratio, the positive and significant relationship between women presidents and sexual offense reporting rates grows stronger over the three-year period (refer to Table 3).

Table 3. DV = Clery Act Sexual Offenses Reported to Campus Security. Negative Binomial Regression.

	One Year Total			Two Year Total			Three Year Total
IVs	B	SE	Odds Ratio	B	SE	Odds Ratio	B	SE	Odds Ratio
Control Variables
	The year of appointment was dummy coded and controlled with years ranging from 2006 to 2019 (data encompasses 2005-2020)
	The US Census Region of the school's location was dummy coded and controlled (West, Midwest, South, Northeast)
Urban Campus	.03	(.11)	1.03	-.04	(.11)	.96	.03	(.10)	1.03
Division I Athletics	.05	(.12)	1.05	.01	(.14)	1.01	.09	(.15)	1.10
Elite Institution	.22	(.16)	1.25	.27	(.16)	1.31	.32	(.17)	1.38
Percent Students Living on Campus	.03***	(.00)	1.03	.03***	(.00)	1.03	.03***	(.00)	1.03
LN of Total Student Body	.97***	(.05)	2.63	1.04***	(.07)	2.83	1.00***	(.09)	2.70
Woman President	.26**	(.08)	1.30	.33***	(.08)	1.40	.48***	(.09)	1.62
Constant	−8.22***	(.61)	.00	−8.43***	(.74)	.00	−7.34***	(.92)	.00

N = 588 for the One-Year Total, N = 534 for the Two-Year Total, and N = 470 for the Three-Year Total

**p < .01 *** p < .001

Supplemental analyses

As our first supplemental analyses, we examined the VAWA sexual offenses data (refer to Appendix A). Given reporting only began in 2014 for the expanded categories of sexual offenses (dating violence, domestic violence, and stalking), our sample size is greatly reduced. In examining one year of data, we have a sample of 133 women presidents and a matched sample of 133 men presidents. In examining two years of data, we have a sample of 93 women presidents and 93 men presidents. And for the three-year examination, we have a sample of 39 women presidents and 39 men presidents. The smaller size does make it more difficult to show statistical significance; however, we believe it is important to also include and report these results that may indicate a broadening understanding and awareness of the scope and prevalence of sexual misconduct on campus.

The correlations table shows a slight positive relationship between women presidents and the one – and two-year examinations of sexual misconduct reporting, yet none reach a level of significance. The mean comparisons in Welch's ANOVA also show similar trends. For the one-year timeframe, the women presidents averaged 9.71 incidents reported and the men presidents averaged 7.95 incidents. In looking at a two-year timeframe, women presidents reported 18.19 incidents to the men's 14.78. And in the three-year timeframe, women presidents reported 23.23 incidents to the men's 23.10. Especially in the one – and two – year timeframes, women presidents have greater reporting of the VAWA sexual offenses, but none of the examinations reach the level of significance. And last, the Poisson regression is consistent with the other tests. A trend is present suggesting a positive relationship between women presidents and sexual misconduct reporting, yet the relationship does not reach the level of significance. In fact, controlling all noted factors, only student body population and percent students living on campus are consistently associated with higher rates of reporting (refer to Appendix A).

The second and third supplemental analyses are included to address whether societal changes during the period under study affected reporting rates. For example, the #MeToo movement came into prominence in 2017 though it had officially started in 2006. Given that the timeframe we examine are presidential appointments between 2006 and 2019, we simply do not have enough cases in the post #MeToo timeframe to differentiate for that social movement. However, we conducted two sets of additional analyses to account for the California Law of Yes Means Yes in 2014 (refer to Appendix B) and the Dear Colleague Letter in 2011 (refer to Appendix C). Although the Yes Means Yes Law began in one state, its presence was widely known and impacted numerous universities and colleges throughout the United States. As illustrated in the Welch's ANOVA in Table B1 of Appendix B, the reporting prior to the Yes Means Yes Law was slightly higher for women than men, but only significant when examining the three-year window. After that law went into effect, the reporting under women presidents increased substantially in each of the examined windows. Reporting also increased for men but by a smaller degree. These findings are also affirmed in the negative binomial regression results. (refer to Appendix B).

The other additional analyses based on societal change is focused on the Dear Colleague Letter (DCL). In 2011, Vice President Joe Biden and the Secretary of Education put forward the DCL to emphasize schools’ legal obligations regarding sexual offenses and to highlight the importance of schools taking action to address sexual misconduct. As shown in the Welch's ANOVA in Table C1 of Appendix C, reporting rates were similar for universities and colleges with men and women presidents. After the DCL, reporting under women increased substantially – over twice as much for the one-year window, 80 percent for the two-year window, and 60 percent for the three-year window. There was minimal increase in reporting in universities with men presidents, and actually a slight decline in the three-year examination. For the negative binomial regression results, women were not significantly related to more reporting prior to the DCL. After the DCL, though, women presidents were positively and significantly associated with increased reporting for all examined timeframes (refer to Appendix C).

Discussion and conclusion

The current study builds on and extends research on the institutional factors that shape campus climates for sexual misconduct (Moylan et al. 2019). Responding to calls by researchers and practitioners, we move beyond individual level predictors to better understand how institutions contribute to variation in reporting. We focus on gender and leadership, analyzing the impact of women's appointment as college president on reporting rates over time. Relying on data from U.S. colleges and universities from 2005-2020, we explore whether women's appointment is associated with changes in reporting rates and whether this impact increases over time.

We find strong empirical support for our research expectations. The appointment of women presidents is significantly and positively associated with higher rates of reporting. At every time point, we find significant differences in reporting rates between men and women presidents, with reporting rates higher for each year following women's appointment. Though the positive impact on reporting increases over time as women presidents’ tenure matures, we see a significant increase in reporting even during the first year following their appointment. This suggests that while women presidents are associated with increased reporting rates over time, the impact of their appointment is measurable immediately following their appointment.

Findings from the supplemental analyses provide further insight into the conditions under which reporting occurs. We find that the rate of reporting sexual misconduct increased substantially under women presidents when the ‘Yes Means Yes’ law was introduced and the ‘Dear Colleague Letter’ was announced. Such public reminders of the pervasive nature of campus sexual misconduct may have prompted women presidents to prioritize this issue within their own institutions.

We find that compared to men, women presidents are more likely to strengthen prevention policies and practices, and that their leadership – in terms of inclusive appointments and coordination between and among units on campus and in the community – improves reporting outcomes. Our findings are consistent with our prediction that the appointment of women to visible and influential leadership positions can affect campus climates by shifting perceptions of risk and safety and increasing confidence and trust in the reporting process.

Several examples from our data illustrate these trends. Dr. Gabrielle Starr was appointed president of Pomona College in 2017 and within a year of her appointment established an advisory council on Sexual Violence Intervention and Prevention. This council was tasked with identifying barriers, challenges, and gaps in prevention, reporting, and response. The council was also tasked with evaluating existing policies as well as suggesting strategies for promoting a climate of safety and awareness. Dr. Linda Livingstone began her tenure at Baylor University amid multiple sexual misconduct scandals. One of her key priorities was to change the culture and ensure that systems were in place to encourage reporting and provide support to survivors. At Iowa State University, Dr. Wendy Wintersteen stated that even one sexual misconduct incident was too much. Iowa State now offers training programs that are continually modified to reflect best practices, and Dr. Wintersteen has prioritized cultivating a supportive climate for survivors, which encourages reporting and bystander intervention. And at University of Minnesota, Joan Gabel, appointed in 2019, launched the President's Initiative to Prevent Sexual Misconduct. This initiative assembled seven subcommittees tasked with specific and actionable items across the university. The overall mission was long-term cultural change and a campus free from sexual misconduct.

These findings have implications for theory and practice. First, our findings underscore the importance of moving beyond individual-level assessments to evaluate institutional factors that shape misconduct and reporting. Institutional variation contributes to the climate for sexual misconduct and survivor response. Research and practice must consider institutional variation when evaluating risk and designing responsive policies. Understanding the institutional conditions that facilitate survivors’ reporting is critical for designing policies that ensure accountability and transparency with regard to sexual misconduct. Creating institutional climates that support and facilitate reporting is a vital step in preventing assault and providing needed supports to survivors.

Our findings also underscore the importance of representation in visible and influential leadership roles. Research finds that the representation of women and people of color in leadership is associated with a range of positive outcomes including stronger community relations, more inclusive policies, greater equity, and stronger governance (Cook, Ingersoll, and Glass 2019; Glass and Cook 2018). The current study underscores the importance of integrating leadership by gender to enhance organizations’ ability to address challenges. Colleges and universities must adopt best practices for fostering inclusive career tracks and for recruiting and hiring presidents and other administrators to maximize organizations’ potential for addressing miscondcut.

Future research can build on these findings to specify the conditions under which leadership composition shapes the context for sexual misconduct reporting. First, the current study focused on gender representation yet impact of leadership is also shaped by other identity positions, including race, ethnicity, sexual identity and disability status, and their intersections (see Rosette et al. 2018 for a review). Research that adopts an intersectional approach to understanding the role of representation can further elaborate the ways in which leadership shapes the context for sexual misconduct reporting. Such an analysis is particularly important in the context of campus sexual misconduct because students of color and LGBTQIA + students are at greater risk of experiencing sexual misconduct and are least likely to report (Brubaker 2019; Tillman et al. 2010). Exploring whether and how intersectional representation of campus leaders facilitates greater institutional trust and reporting among historically marginalized and underrepresented is important for identifying effective mechanisms for supporting survivors.

Our quantitative analysis could also be strengthened by qualitative research that explores individual pressures and motivations of leaders from historically underrepresented groups. Though theory and research predict that women leaders will face pressures to demonstrate a stronger commitment to sexual misconduct prevention, understanding how these pressures and motivations shape decisions ‘on the ground’ would help to further refine theory regarding the mechanisms that contribute to greater reporting rates.

We also encourage researchers to consider institutional mechanisms that shape the context for misconduct reporting and prevention in other organizational settings including work organizations, the military, and K-12 schools, which vary considerably in terms of the representation of women and people of color in leadership. Across contexts where sexual misconduct occurs, sexual misconduct remains underreported. Comparing outcomes across organizational contexts could further elaborate the influence of representation on reporting.

Finally, the current study is limited to the U.S. context. While research finds that more gender equity impacts a range of organizational outcomes (e.g. Otterbach, Sousa-Poza, and Zhang 2021), the impact of leaders may vary in other national or organizational contexts. Comparative research beyond the U.S. can build upon and extend our findings. Comparative research reveals that cultural and legal context contribute to widely varying definitions and perceptions of sexual misconduct and harassment (Paludi 2006; Pryor et al. 1997). Despite this evidence, university-based sexual misconduct research beyond the U.S. remains relatively underexplored (see Lee and Wong 2017 for an exception). A comparative study would reveal whether women's representation in senior administrative roles in other cultural contexts has similar or different impacts on sexual misconduct reporting.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 We use the term sexual misconduct to refer to a range of sexual offenses that use power and control to cause harm to others. Sexual misconduct includes rape, assault, harassment, nonconsensual sexual contact, sexual exploitation, intimate partner violence, dating violence, and stalking.

2 Though new Title IX regulations were issued in 2020 by the Trump Administration, we do not review those here since they are outside of the period under study.

Appendix A

– VAWA Sexual Offenses

Table A1. Descriptives and Correlations. VAWA Sexual Offenses.

Variable	N	Mean	s.d.	1	2	3	4	5	6	7	8
1. Woman President	266	.50	.50	– –
2. Urban Campus	266	.71	.46	.00	– –
3. Division I Athletic Program	266	.19	.22	.00	.26**	– –
4. Elite Institution	266	.05	.07	.00	.04	-.07	– –
5. Percent Students Living on Campus	266	43.35	26.61	.00	-.19**	-.34**	.21**	– –
6. LN of Total Student Body	266	8.18	1.27	.00	.17**	.64**	.04	-.48**	– –
7. Number of Sexual Offenses for One Year	266	8.84	13.90	.06	.16**	.43**	.10	-.09	.51**	– –
8. Number of Sexual Offenses for Two Years	186	16.49	25.70	.07	.13	.47**	.11	-.04	.42**	.96**
9. Number of Sexual Offenses for Three Years	78	23.17	38.12	.00	-.07	.59**	-.11	-.10	.57**	.97**	.99**

* p < .05 ** p < .01

Table A2. Welch's ANOVA. VAWA Sexual Offenses.

	Woman as President					Man as President
Variable	F-Value	N	Mean	SD	N	Mean	SD
One Year	1.07	133	9.71	15.42	133	7.95	12.18
Two Years	.82	93	18.19	29.80	93	14.78	20.82
Three Years	.00	39	23.23	42.69	39	23.10	33.50

*Welch's test was used to account for unequal variances between groups; asymptotically distributed F-value reported

Table A3. DV = VAWA Sexual Offenses Reported to Campus Security. Poisson Regression.

	One Year Total			Two Year Total			Two Year Total
IVs	B	SE	Odds Ratio	B	SE	Odds Ratio	B	SE	Odds Ratio
Control Variables
The year of appointment was dummy coded and controlled with years ranging from 2014 to 2019 (data through 2020)
The US Census Region of the school's location was dummy coded and controlled (West, Midwest, South, Northeast)
Urban Campus	.22	(.28)	1.24	-.07	(.31)	.93	-.51	(.28)	.60
Division I Athletics	.18	(.22)	1.20	.53*	(.22)	1.69	.60	(.31)	1.82
Elite Institution	-.16	(.22)	.85	.09	(.29)	1.09	-.19	(.40)	.83
Percent Students Living on Campus	.02***	(.00)	1.02	.03***	(.01)	1.03	.03***	(.00)	1.04
LN of Total Student Body	.85***	(.10)	2.34	.94***	(.12)	2.55	1.24***	(.14)	3.44
Woman President	.20	(.14)	1.22	.21	(.17)	1.23	.01	(.17)	1.01
Constant	−6.08***	(1.05)	.00	−7.05***	(1.11)	.00	−8.96***	(1.37)	.00

N = 266 for the One-Year Total, N = 186 for the Two-Year Total, and N = 78 for the Three-Year Total

* p < .05 ** p < .01 *** p < .001

Appendix B

– Split File for the Yes Means Yes Bill (2014)

Table B1. Welch's ANOVA. Clery Act Sexual Offenses.

	Pre-Yes Means Yes Bill (2005-2013)							Post-Yes Means Yes Bill (2014-2020)
		Woman as President			Man as President				Woman as President			Man as President
Variable	F-Value	N	Mean	SD	N	Mean	SD	F-Value	N	Mean	SD	N	Mean	SD
One Year	1.11	120	3.03	4.60	120	2.44	3.95	3.24	174	6.21	8.32	174	4.75	6.67
Two Years	2.14	110	7.17	9.98	110	5.35	8.51	7.56**	157	11.95	14.20	157	8.09	10.38
Three Years	4.36*	95	12.60	17.89	95	7.80	13.49	11.62***	140	18.16	20.88	140	10.94	13.92

*p < .05 **p < .01 ***p < .001

*Welch's test was used to account for unequal variances between groups; asymptotically distributed F-value reported

Table B2. DV = Clery Act Sexual Offenses Reported to Campus Security. Negative Binomial Regression.

	Pre-Yes Means Yes Bill (2005-2013)
	One Year Total			Two Year Total			Three Year Total
IVs	B	SE	Odds Ratio	B	SE	Odds Ratio	B	SE	Odds Ratio
Control Variables
The year of appointment was dummy coded and controlled with years ranging from 2006 to 2013 (data from 2005)
The US Census Region of the school's location was dummy coded and controlled (West, Midwest, South, Northeast)
Urban Campus	-.05	(.23)	.96	-.13	(.19)	.88	-.02	(.19)	.99
Division I Athletics	.25	(.21)	1.29	-.05	(.21)	.96	.08	(.20)	1.08
Elite Institution	.71**	(.27)	2.04	.51*	(.23)	1.67	.61**	(.23)	1.84
Percent Students Living on Campus	.02***	(.00)	1.02	.03***	(.00)	1.03	.03***	(.00)	1.03
LN of Total Student Body	.87***	(.10)	2.39	1.13***	(.09)	3.09	1.09***	(.09)	2.98
Woman President	.29*	(.14)	1.33	.34**	(.13)	1.41	.54***	(.14)	1.72
Constant	−7.96***	(1.03)	.00	−9.82***	(.83)	.00	−9.22***	(.94)	.00

N = 240 for the One-Year Total, N = 220 for the Two-Year Total, and N = 190 for the Three-Year Total

* p < .05 ** p < .01 *** p < .001

Table B3. DV = Clery Act Sexual Offenses Reported to Campus Security. Negative Binomial Regression.

	Post-Yes Means Yes Bill (2014-2020)
	One Year Total			Two Year Total			Three Year Total
IVs	B	SE	Odds Ratio	B	SE	Odds Ratio	B	SE	Odds Ratio
Control Variables
The year of appointment was dummy coded and controlled with years ranging from 2014 to 2019 (data through 2020)
The US Census Region of the school's location was dummy coded and controlled (West, Midwest, South, Northeast)
Urban Campus	.05	(.12)	1.05	.02	(.12)	1.02	.07	(.12)	1.07
Division I Athletics	.02	(.15)	1.02	.03	(.17)	1.03	.08	(.18)	1.08
Elite Institution	-.09	(.16)	.91	-.01	(.19)	.99	-.04	(.22)	.96
Percent Students Living on Campus	.03***	(.00)	1.04	.03***	(.00)	1.04	.03***	(.00)	1.03
LN of Total Student Body	.97***	(.07)	2.63	.97***	(.10)	2.64	.91***	(.12)	2.50
Woman President	.26**	(.10)	1.29	.33***	(.10)	1.40	.46***	(.10)	1.59
Constant	−8.32***	(.73)	.00	−7.83***	(1.02)	.00	−6.63**	(1.24)	.00

N = 348 for the One-Year Total, N = 314 for the Two-Year Total, and N = 280 for the Three-Year Total

** p < .01 *** p < .001

Appendix C

– Split File for the Dear Colleague Letter from VP Biden in 2011 (legal obligations on reporting)

Table C1. Welch's ANOVA. Clery Act Sexual Offenses.

	Pre-Dear Colleague Letter (2005-2010)							Post-Dear Colleague Letter (2011-2020)
		Woman as President			Man as President				Woman as President			Man as President
Variable	F-Value	N	Mean	SD	N	Mean	SD	F-Value	N	Mean	SD	N	Mean	SD
One Year	.03	60	2.50	3.58	60	2.62	3.92	4.73*	234	5.53	7.76	234	4.12	6.19
Two Years	.05	52	6.00	7.96	52	6.37	9.41	11.28***	215	10.94	13.59	215	7.10	9.82
Three Years	.03	37	10.57	13.39	37	10.00	17.32	17.58***	198	16.91	20.74	198	9.61	13.10

*p < .05 ***p < .001

*Welch's test was used to account for unequal variances between groups; asymptotically distributed F-value reported

Table C2. DV = Clery Act Sexual Offenses Reported to Campus Security. Negative Binomial Regression.

	Pre-Dear Colleague Letter (2005-2010)
	One Year Total			Two Year Total			Three Year Total
IVs	B	SE	Odds Ratio	B	SE	Odds Ratio	B	SE	Odds Ratio
Control Variables
The year of appointment was dummy coded and controlled with years ranging from 2006 to 2010 (data from 2005)
The US Census Region of the school's location was dummy coded and controlled (West, Midwest, South, Northeast)
Urban Campus	-.08	(.31)	.92	-.45	(.28)	.64	.10	(.30)	1.11
Division I Athletics	.58*	(.27)	1.79	.36	(.28)	1.43	.49	(.28)	1.62
Elite Institution	.50	(.37)	1.65	.52	(.29)	1.68	.68**	(.23)	1.97
Percent Students Living on Campus	.01	(.01)	1.01	.03***	(.01)	1.03	.03***	(.01)	1.03
LN of Total Student Body	.72***	(.15)	2.04	1.04***	(.13)	2.84	1.07***	(.19)	2.91
Woman President	-.01	(.19)	.99	-.02	(.16)	.98	.15	(.17)	1.16
Constant	−6.40***	(1.48)	.00	−8.69***	(1.25)	.00	−8.63***	(1.87)	.00

N = 120 for the One-Year Total, N = 104 for the Two-Year Total, and N = 74 for the Three-Year Total

*p < .05 ** p < .01 *** p < .001

Table C3. DV = Clery Act Sexual Offenses Reported to Campus Security. Negative Binomial Regression.

	Post-Dear Colleague Letter (2011-2020)
	One Year Total			Two Year Total			Three Year Total
IVs	B	SE	Odds Ratio	B	SE	Odds Ratio	B	SE	Odds Ratio
Control Variables
The year of appointment was dummy coded and controlled with years ranging from 2011 to 2019 (data through 2020)
The US Census Region of the school's location was dummy coded and controlled (West, Midwest, South, Northeast)
Urban Campus	.02	(.11)	1.02	.01	(.11)	1.01	.01	(.11)	1.01
Division I Athletics	-.02	(.13)	.97	-.07	(.15)	.93	-.01	(.16)	.99
Elite Institution	.30	(.17)	1.35	.30	(.18)	1.34	.24	(.19)	1.27
Percent Students Living on Campus	.03***	(.00)	1.03	.03***	(.00)	1.03	.03***	(.00)	1.03
LN of Total Student Body	1.00***	(.06)	2.70	1.04***	(.08)	2.82	.99***	(.10)	2.68
Woman President	.31***	(.09)	1.37	.41***	(.09)	1.51	.55***	(.09)	1.74
Constant	−8.55***	(.66)	.00	−8.41***	(.84)	.00	−7.25***	(1.00)	.00

N = 468 for the One-Year Total, N = 430 for the Two-Year Total, and N = 396 for the Three-Year Total

***p < .001