Gender diversification in police agencies: is it a zero-sum game?

Abstract

Purpose

The article first examines whether police hiring decisions represent a zero-sum game where hires from one under-represented group (e.g. White women) reduce the number of hires made from other under-represented groups (non-White men and/or non-White women). Second, we explore whether agencies that hire more members of underrepresented groups achieve more diverse applicant pools in future hiring cycles. Negative binomial regression techniques are used in both analyses.

Design/methodology/approach

Data for this study come from the Commission on Accreditation for Law Enforcement Agencies (CALEAs) from 2011 to 2016. These data are divided into two periods: Period A (2011–2013) and Period B (2014–2016). The two periods are combined to assess a zero-sum effect. Then, Period A data on hiring decisions is used to estimate the diversity in applicant pools in Period B.

Findings

Results from this study provided little evidence of a zero-sum effect. It does not seem that agencies that hire from one under-represented group are less likely to hire from others. Instead, agencies that have shown a commitment to diversification are more likely to make additional hires from under-represented groups. We also found evidence of a relationship between Period A hires and Period B applicant pools for Hispanic women, but not for other groups. Broadly, we found that agencies where a larger share of officers are women were more likely to hire more women applicants.

Originality/value

Previous research examining zero-sum effects in hiring rely on officer rosters rather than specific applicant and hiring data. The data used in this study allows for a more precise examination of hiring decisions, and allows us to link hiring decisions to future applicant pool composition.

Citation

Nowacki, J., Schafer, J. and Hibdon, J. (2021), "Gender diversification in police agencies: is it a zero-sum game?", Policing: An International Journal, Vol. 44 No. 6, pp. 1077-1092. https://doi.org/10.1108/PIJPSM-02-2021-0033

Introduction

Recent calls for police reform include diversification of police ranks. Though the focus of diversification seems primarily centered on racial and ethnic diversity, there has been increased advocacy for increasing the representation of women in policing (e.g. Schumaker, 2020) [1]. The racial and ethnic diversity of policing has long been criticized (e.g. President's Commission on Law Enforcement and Administration of Justice, 1967), however the percentage of Black police officers has approximated the US population for several decades and Hispanic representation has advanced with the growth of Hispanics in the US population (Hyland and Davis, 2019). In contrast, representation of women has been stagnant since the 1990s (Cordner and Cordner, 2011) with women representing around twelve percent of sworn officers (Hyland and Davis, 2019) [2].

Police reform efforts across the US continue to highlight the importance of racial/ethnic and gender diversification. The first pillar in the President's Task Force on 21st Century Policing reflects “Building Trust and Legitimacy.” Recommendation 1.8 specifies that “Law enforcement agencies should strive to create a workforce that contains a broad range of diversity including race, gender, language, life experience, and cultural background to improve understanding and effectiveness in dealing with communities” (2015, p. 16). US policing has been encouraged to diversify for years, often with limited success. Even where initial improvements have been made, this progress tends to level after initial advances (Cordner and Cordner, 2011).

Diversification is a two-step process. First, agencies need to develop diverse applicant pools; organizations can only hire from individuals who have acted (completing an application) to become a police officer. If members of underrepresented groups do not apply, they cannot be hired and departments cannot improve diversity. Second, even where applicant pools are diverse, departments make hiring decisions from among the most qualified candidates. The selection process entails traveling to locations where written and physical tests are held, participating in interviews and screening assessments, and passing all requisite steps in the often-lengthy selection process. Applicants then must complete a training program before becoming police officers. Given the length and rigor of these processes, some applicants will drop out or be removed. The rate at which attrition occurs is likely racialized and gendered (Reaves, 2016).

The limited research exploring gender diversification in policing generally approaches this issue from a broad perspective (Jordan et al., 2009; Matusiak and Matusiak, 2018; Morabito and Shelley, 2015). Studies frequently treat women as a single group, with few studies approaching police diversification from an intersectional perspective, acknowledging the oppression and marginalization women experience might vary by race or ethnicity (Collins, 2000; Crenshaw, 1989). The multiple, interlocking features that contextualize differing experiences between White, Black, and Hispanic women illustrate that the characteristics that influence diversification for White women may look quite different for Black and Hispanic women.

Even absent other constraints, only a limited number of police officer positions are available at any one time. Some of these positions may be offered to members of underrepresented groups, while others may not. Some evidence points to the existence of a zero-sum game where attempts to diversify an agency through targeted hires can result in reductions in the positions available for members of competing groups (Alozie and Ramirez, 1999). Essentially, the “gains” of hiring officers from underrepresented groups are simply trade-offs where advances in diversification are off-set. For instance, gains in racial or ethnic diversity may come at the expense of advancing gender diversity in an agency.

Successful diversification at one point in time may have implications for future hiring cycles. Agencies that show a willingness and ability to hire members of underrepresented groups may see future increases in the volume of applications from those groups. As officer diversity becomes more visible, those who were previously reluctant or disinterested in policing may become more likely to pursue policing positions (Rossler et al., 2020; Schulz, 2004). Conversely, agencies who emphasize hiring from underrepresented groups in one cycle may be less likely to hire applicants from underrepresented groups in subsequent cycles if they perceive the need for diversification was met with prior hires.

This article explores agency characteristics conducive to gender diversification in policing, accounting for racial/ethnic differences among female applicants. We examine the argument that police hiring decisions reflect a zero-sum game where hiring from one underrepresented group (e.g. non-White men) may reduce the opportunities and resources available to other groups (e.g. women applicants). Moreover, we explore whether hiring outcomes in one period affect the composition of application pools in later periods. We argue that agencies that show an effort to diversify will be rewarded with more diverse applicant pools going forward.

Women in policing

The limited literature regarding racial/ethnic diversification in policing is growing (Alozie and Ramirez, 1999; Gustafson, 2013; Nowacki et al., 2021; Zhao et al., 2005), with even less research focused on gender diversification [3]. Gender diversification studies focus on how organizational characteristics of police agencies may influence hiring women. Studies have found that larger agencies (Matusiak and Matusiak, 2018; Morabito and Shelley, 2015; Schuck, 2014b; cf. Schulz, 2003) and agencies with larger operating budgets (Jordan et al., 2009) achieve greater gender diversity. Gender representation is related to targeted recruiting strategies (Jordan et al., 2009), occupational benevolence (Morabito and Shelley, 2015), and jurisdictional characteristics, such as racial composition (Morabito and Shelley, 2015; Schuck, 2014b), accreditation (Morabito and Shelley, 2015), and unionization (Morabito and Shelley, 2015; Schuck, 2014b). More complex agencies tend to employ smaller proportions of women officers (Matusiak and Matusiak, 2018).

Other research has examined gender representation among police supervisors (Murray, 2021; Shjarback and Todak, 2019; Silvestri, 2018; Todak et al., 2021). The proportion of police supervisors who are women remains low (Hyland and Davis, 2019; Schulz, 2003, 2004), yet the ratio of male-to-female supervisors has increased over time (Guajardo, 2016). Organizational characteristics that influence gender diversification among supervisors include agency size, community-oriented policing strategies, and accreditation (Shjarback and Todak, 2019). Women who have achieved promotion identify supportive colleagues, mentors, family support, credentials, and litigation as factors conducive to promotion (Morabito and Shelley, 2018; Schulz, 2004).

Benefits of a gender diverse police force

Women may bring traits to policing that make diversification appealing (Murray, 2021). These “soft” policing skills (McCarthy, 2013) include greater communication, less use of force, and more efficacy working with vulnerable populations. One of the frequently highlighted benefits of women in policing is a stronger ability to empathize with another person, which implies an ability to place oneself in another person's circumstances. Women officers perceive themselves as more compassionate (Morash and Haarr, 2012; Schuck, 2014a), more empathetic (McCarthy, 2013), and better at communicating (McCarthy, 2013; Rabe-Hemp, 2009).

Several empirical studies indicate that women officers are less likely to engage in forceful behaviors or misconduct (Schuck and Rabe-Hemp, 2005). Women officers in Australia received fewer initial and repeat complaints for excessive force (Porter and Prenzler, 2017). A study of New York City officers finds that women who were married, college-educated, and belonged to specialized units were less likely to experience termination via misconduct (Gaub, 2020). The gender composition of officer dyads is noteworthy, as research indicates that officers participating in woman-woman dyads are less likely to use force (Schuck and Rabe-Hemp, 2005). Other studies find that women in gender-balanced workgroups are just as likely to make arrests and issue citations as men, though women are less likely to engage in those activities in unbalanced workgroups (Kern and Lundman, 2012).

Finally, research indicates that women officers are more likely to work with victims, women, and children, are more likely to work with victim advocates and take sexual assault education, and are less likely to accept rape myths (Rich and Seffrin, 2014). Agencies with larger proportions of women officers showed higher rates of reporting and clearances for rape offenses (Schuck, 2018), suggesting victims may be more likely to report and cooperate with police when more women are available. Conversely, women officers are held to a higher standard of competency when responding to domestic violence calls compared to men (Grant, 2000). Taken together, this literature indicates that there are many benefits for agencies when they improve gender diversification.

Barriers to gender representation in policing

One of the key challenges in creating a diverse police department is developing a diverse applicant pool. In order to do this, police agencies need to highlight policing as an attractive profession, yet several barriers make it difficult for police agencies to improve gender diversity. These barriers include: lack of opportunities; harassment and resistance from male officers; gendered policy; pervasive masculinity; and, perceptions about the selection process. Women may be reluctant to pursue sworn positions in police departments; historically, women were relegated to social work roles within police organizations (Martin, 1999; Schulz, 2004). Compared to men, women may be more likely to work with victims of crime, particularly women and children (Martin and Jurik, 2007). Surveys indicate that female undergraduates perceive lower levels of fulfillment, respect, acceptance, and opportunity for women in policing careers compared to men (Rossler et al., 2020). Students associate policing with masculinity; aspirations to policing have been linked with masculine self-conceptions among both male and female students (Clinkinbeard et al., 2020).

Harassment and discrimination toward women working in policing serve as another barrier. Women are more likely to experience sexual harassment, originating from both male colleagues and citizens. The “boy's club” aspect of policing may resist diversification efforts. Research in the US (Helfgott et al., 2018; Rief and Clinkinbeard, 2020), Latin America (Colvin, 2017), and Australia (Ward et al., 2020) have identified gender discrimination as an obstacle facing women officers. Women working in policing frequently indicate deploying strategies to neutralize, dismiss, or even participate in hypersexualized behavior (Brown et al., 2020; Cordner and Cordner, 2011; Rabe-Hemp, 2009; Rossler et al., 2020). Some women officers characterize hypersexualized banter and joking in the workplace as a sign they are accepted by their male peers (Brown et al., 2020), yet the price of achieving such acceptance for some might drive other women out of policing.

Gendered policy within police agencies is an additional issue, with female police officers often citing the lack of family-friendly policies (Murray, 2021) and hair length standards (Kringen and Novich, 2018) as a barrier to policing careers. Perceptions of physical fitness standards may steer women away from policing (Rossler et al., 2020; Schulze, 2012). Agencies with physical fitness criteria as a component in their selection process have lower levels of female representation (Lonsway, 2003; National Institute of Justice, 2019; Schuck, 2014b). Conversely, women may perceive physical fitness standards to be higher than they actually are; when standards are spelled out, women might see them as achievable and be more willing to pursue opportunities in policing (Aiello, 2019). Available evidence suggests workplace policies, particularly those that appear to be gendered, may constrain women applications, thus reducing ultimate hires and challenging retention among women.

Police departments may not be doing enough in their efforts to recruit women. In surveys of women police officers and male police chiefs in Pennsylvania, each group offered divergent explanations for the lack of gender representation in policing (Cordner and Cordner, 2011). Those leading agency recruitment efforts (male police chiefs) do not see the same gender and diversity issues as those they seek to recruit and retain, which speaks volumes about the problem. Female applicants identify being motivated to apply for police positions because of a desire to help others (Gibbs, 2019). Agencies may not leverage this as completely as they could. Starting pay was positively correlated with applications from women (Jordan et al., 2009), suggesting efforts to raise entry-level compensation may prove fruitful (Giblin and Galli, 2017; Yu, 2018).

Perhaps the largest barrier to gender equity in police agencies is the male-dominated culture that persists in policing (Brown et al., 2020; Murray, 2021). Women working in policing often report feeling marginalized by a culture that amplifies and exemplifies hegemonic masculinity (Brown et al., 2020; Matusiak and Matusiak, 2018; Murray, 2021; Shelley et al., 2011). This system of social and cultural practices reinforces men's power and advantage across social institutions through fortifying patriarchal arrangements (Martin and Jurik, 2007; Messerschmidt and Tomsen, 2015; Prokos and Padavic, 2002; Shelley et al., 2011). Police culture also constructs a masculine image of the ideal police leader (Silvestri, 2018), creating challenges unique to women in leadership roles (Schulz, 2004). Even where women have made significant strides in terms of numeric representation, policing as a profession has not been feminized (Brown and Silvestri, 2020), with women identifying male resistance as a significant obstacle to their full integration (Burdett et al., 2018; Cordner and Corner, 2011; Schulz, 2004).

Gender diversification in an intersectional framework

Much of the literature regarding women in policing treats gender as a singular concept, providing little or no consideration of within group variation in the experiences of women in policing [4]. As such, extant literature may be more applicable to White female officers, with less clear applicability to female officers of color. Examinations of Black and Hispanic officers are important because their experience may differ from White women, supporting the utility of approaching gender dynamics in policing through an intersectional lens. The intersectional approach suggests that inequities are best understood through multiple, interlocking features that compose an individual's position in the social structure (Collins, 2000; Crenshaw, 1989). From this framework, “race, class, gender, sexuality, and other locations of inequality are dynamic, historically grounded, socially constructed power relationships that simultaneously operate at both the micro-structural and macro-structural levels” (Burgess-Proctor, 2006, p. 37). Thus, to not only understand the experiences of White women in policing, an intersectional approach is ideal in analyzing how race/ethnicity, sex, and other social characteristics jointly affect experiences (Hassell and Brandl, 2009) [5].

In reviewing the literature regarding non-White women in policing, Todak and Brown (2019) found that studies focused on either racial/ethnic or gender diversification, but rarely both. Studies they located that examine non-White women officers used data from 2000 or earlier [6]. The studies that were reviewed suggest that Black women represent less than 4% of all sworn positions, while Hispanic women represent less than 2%. Todak and Brown (2019) also found the research that does exist often focuses on Black rather than Hispanic women in policing. Given the lack of research in this area, a study of characteristics that predict hiring outcomes among Black and Hispanic women applicants is needed.

The culture of policing emphasizes masculine traits and attributes in ways that might result in alienating and isolating females (see Acker, 1990), impeding women from entry into policing and isolating them if they are working in policing (Matusiak and Matusiak, 2018; Murray, 2021). This can extend into less overtly apparent policies and practices that disadvantage women as primary care givers (Cordner and Cordner, 2011; Murray, 2021) and that have compounding effects on non-white women. For example, Kringen and Novich (2018) found evidence of masculine embodiments of gendered positions in a large southwestern police academy's policy regarding hair length for female recruits. The policy was a substantial barrier to entry for Black women, who felt it would reduce their femininity and would have lasting effects after graduating the academy. The policy was grounded in what was convenient for the academy, not an officer safety justification, and with no agency recognition that an arbitrary rule was working against efforts to diversify the academy, especially among Black women. Viewing this study through an intersectional lens emphasizes the importance of recognizing differing experiences among women.

Inter- and intra-group competition in police hiring decisions

While police departments may aim to diversify, one of the key challenges that they face is that they are working with a finite set of resources. Police agencies are limited by the number of positions that they can fill. Given these limited resources, it is likely that advances in diversification among one group may come at the expense of another. For example, an agency that hires many women applicants may be less likely to hire non-White applicants in the same cycle. This presents a zero-sum dilemma. Thus, it is likely that agencies that hire more non-White men will hire fewer women, and agencies that hire more White women are less likely to hire non-White women.

In this situation, members of different underrepresented groups compete for the same scarce resource. Essentially, these applicants are competing for the same “piece of the pie”. Unless the organization pivots away from historic practices of hiring mostly White men, the overall share of positions occupied by members of underrepresented groups will not change. In the context of police hiring, Alozie and Ramirez (1999) examined this situation using data from the 1990 Equal Employment Opportunity (EEOC) special files dataset. They found that: Black employment is negatively correlated with Hispanic employment in police positions; that White women compete with Hispanic candidates, especially in Hispanic minority cities; and that Black and Hispanic men and women often compete for the same share of resources. This study provides positive evidence for a zero-sum relationship. While informative, the Alozie and Ramirez study is limited by the absence of data regarding applicant pool composition.

Current study

This study seeks to better understand the characteristics that predict success in diversifying police department employee rosters by hiring women applicants, including Black and Hispanic women. We explore whether a zero-sum relationship exists, such that departments that hire from one underrepresented group are less able to hire from other underrepresented groups. Lastly, we explore whether hiring decisions in earlier periods influence the composition of applicant pools in later periods.

Data and methods

The data for this study comes from the Commission on Accreditation for Law Enforcement Agencies (CALEAs). Every three years, CALEA member agencies submit a re-accreditation report [7]. As part of this report agencies submit applicant and hiring data. We use data from the accreditation reports received in 2012/2013 and 2015/2016. We treat two reports as Period A (2011–2013) and Period B (2014–2016). The reports include the total number of hires (in each 3-year period), the total number of applications, and racial/ethnic information for the applicants and hires [8] This dataset includes information regarding jurisdictional and organizational characteristics, specifically: service population, available workforce, and the racial/ethnic and gender composition of the officers in the department during the reporting period. Complete data were available for 127 agencies.

Dependent variables

For the hiring models, we examine three separate dependent variables: the total number of female hires, Black female hires, and Hispanic female hires from periods A and B. Three additional dependent variables measure the number of women, Black women, and Hispanic women applicants in Period B. The latter variables are used to determine if the proportion of women hired in Period A relates to the representation of women in applicant pools during Period B.

Independent variables

To predict our outcomes, we include measures of jurisdictional characteristics, agency characteristics, and hiring outcomes from other racial/ethnic and gender groups. The jurisdictional and agency variables are measured during Period A (2011–2013). The timing of the hiring and application variables are specified in the appropriate sections below.

Jurisdictional variables

In each model, we include the percentage of the service population that is Black and the percentage that is Hispanic. We calculate this by taking the size of the service population from each group, dividing by the total service population, and multiplying by 100. Agencies may be more successful in hiring more non-White applicants in jurisdictions where non-White populations are larger.

Agency variables

We include four measures of agency characteristics: department size, percentage of Black officers, percentage of Hispanic officers, and percentage of female officers. Department size is measured as the total number of full-time sworn officers. This variable is log-transformed due to skew. Smaller agencies may have fewer positions and opportunities for promotion available, which can be a deterrent to applying. We calculate the variables for percent Black officers, percent Hispanic officers, and percent women officers by dividing the number of officers from each group by the total number of officers and multiplying by 100. Agencies with more diverse staff compositions may be more likely to hire more non-White and women applicants. These variables are measured in Period A.

Hiring variables

We include measures of the number of off-group hires that the agency made in each model. Each of these measures are log-transformed in order to normalize the distribution [9] For example, in the model examining all women that were hired, we include measures of the number of Black and Hispanic male applicants that were hired by the agency. This allows us to examine whether a zero-sum relationship exists. All of the independent variables described in this section are drawn from Period A.

Application variables

Last, we examine the number of applications in Period B for women, Black women, and Hispanic women. In these models, we include the jurisdiction and agency variables described above from Period A, with the addition of the percentage of hires and the percentage of applications from Period A from each corresponding group. These models allow us to examine whether agencies that hire more diversely in Period A receive more diverse applicant pools in Period B.

Descriptive statistics are presented in Table 1. These statistics illustrate that departments in our sample hired, on average, 13.76 women, including 2.85 black women and 1.68 Hispanic women. Women represented 13.43% of applications during Period A, with Black women representing 3.69% and Hispanic women 1.37%. The mean service population in those jurisdictions is 14% black and 11% Hispanic. The average department size is 281.82 officers. Lastly, these agencies are composed of (on average), 8.60% Black officers, 6.15% Hispanic officers, and 12.40% female officers.

Analytic strategy

We use negative binomial regression to analyze our models. Both hires and applications represent discrete counts and feature non-symmetric distributions that violate ordinary least squares (OLSs) assumptions. We use negative binomial techniques because these models allow for over-dispersion (see Osgood, 2000) [10]. In our hiring models, we include jurisdictional and agency characteristics, along with measures of off-group hiring outcomes to examine the possibility of zero-sum relationships [11]. In our application models (where the outcome is measured during Period B), we include jurisdictional and agency variables, along with hiring and application measures from Period A [12].

Results

Hiring outcome models

We begin by examining the multivariate models for hiring outcomes (see Table 2) [13]. The first model includes all female hires from both periods. In terms of the zero-sum argument, neither Black male hires nor Hispanic male hires was significant. This finding does not support a zero-sum argument. On the other hand, the variable for White male hires was statistically significant (b = 0.471), suggesting that agencies that hired more White men also hired more women. Additionally, in this model, department size is significant, suggesting that larger departments hire fewer women (b = −0.516). Percent Black officers (b = 0.041), percent Hispanic officers (b = 0.033), and percent women officers (b = 0.035) reach significance, all suggesting that more diverse agencies hire more women.

We now turn to the model where we examine Black female hires. In this model, the measure of Black male hires is not statistically significant, but the measure of Hispanic male hires is (b = −0.173). Neither measure of female hires (White or Hispanic) reached statistical significance. Thus, only the measure of Hispanic male hires supports the zero-sum argument. In addition, percentage of women officers (b = 0.092) is statistically significant. Thus, agencies with more women already on staff are more likely to hire a larger number of Black women applicants.

Our final hiring model predicts the expected number of Hispanic female hires. In this model, the number of Black (b = 0.232) but not Hispanic male hires was significant, however the direction of the coefficient does not support a zero-sum argument. Moreover, this model suggests that agencies in places with larger Hispanic service populations (b = −0.078) hire fewer Hispanic women applicants. Lastly, agencies with a larger proportion of Hispanic officers (b = 0.076) and a larger proportion of women officers (b = 0.073) hire a larger number of Hispanic women applicants.

Applications in Period B models

The next set of results examine how hiring outcomes during Period A might affect the number of applications received from women in Period B (see Table 3). We begin with the model for all female applications in Period B. According to the results from this model, the percentage of women hires in Period A is not related to female applications in Period B. However, this model does suggest that agencies located in jurisdictions with larger Black populations (b = 0.072) and agencies where a larger proportion of Hispanic officers (b = 0.142) received more female applications in Period B. Conversely, agencies with a larger proportion of Black officers (b = −0.131) received fewer women applications in Period B.

Our next model examines applications from Black women during Period B. Here the percentage of hires that were Black women during Period A does not influence Black female applications during Period B. Additionally, results from this model suggest that agencies located in places with larger Hispanic populations received more applications from Black women during Period B.

Our final model examines applications from Hispanic women during Period B. The results from this model suggest that there is a statistically significant (and positive) relationship between the percentage of Hispanic female hires in Period A (b = 0.209) and the number of Hispanic female applications in Period B. Additionally, we found that more applications were submitted by Hispanic women to larger agencies (b = 0.606). The model examining applications from Hispanic women does support our argument that more diverse hires in Period A will correspond with more diverse application pools during Period B, however, our models examining all women and Black women applications provide no such evidence.

Discussion

The President's Task Force on 21st Century Policing (2015) recommend that police departments emphasize diversification in future recruitment and hiring efforts. While several studies have examined efforts to diversify across race/ethnicity (Alozie and Ramirez, 1999; Gustafson, 2013; Zhao et al., 2005), far fewer have examined gender diversification. In this study, we explore characteristics related to gender diversification in policing both through hiring decisions and application pools. Specifically, we investigate whether a zero-sum relationship exists between racial/ethnic and gendered hiring, and whether diversification in one period influences the composition of applicant pools in a later period. Results from our study provided little evidence of a zero-sum relationship, however, the composition of officers within the agency, in terms of both race/ethnicity and gender, influenced the volume of hires among women. Moreover, while we did not find broad evidence of a relationship between hires from 2011 to 2013 and applications from 2014 to 2016, we did find that agencies where Hispanic women represented a larger share of hires in Period A did see more applications from Hispanic women in the following period.

In terms of hiring outcomes, we found that departments whose officer rosters are represented by larger shares of Black, Hispanic, and women officers hired more women during the timeframe included in our data. More specifically, agencies with larger shares of Hispanic officers hired more Hispanic women. These findings illustrate a trend where departments that have committed to diversification in the past continue to make hiring decisions that reflect that commitment. It appears there is an underlying trait associated with these agencies that fosters further diversification. For a police agency to diversify, the agency needs to embrace the idea that diversification is beneficial to the organization.

The argument that hiring from underrepresented groups (e.g. women, Black, and Hispanic applicants) takes on a zero-sum relationship, where hiring from one group would mean that fewer hires were made from competing groups, was tested in our analysis. The results provided little evidence in support of this argument. Instead, agencies that hired more non-White men often also hired more women, both White and non-White. These findings suggest that different underrepresented groups are not competing for the same “piece of the pie” (Alozie and Ramirez, 1999). Instead, agencies that are committed to diversification are expanding the volume of positions dedicated to these underrepresented groups. Rather than limiting access to scarce resources, these agencies appear to be expanding the resources available. That is, these agencies are offering more pieces of pie, not asking competing groups to share what little has been available historically.

Last, we investigated how hiring decisions in 2011–2013 might influence the composition of applicant pools in 2014–2016. Overall, a relationship did not materialize, but in the case of Hispanic women, more hires in 2011–2013 corresponded with more applications in 2014–2016. This may suggest that greater representation among Hispanic women in police positions may motivate more Hispanic women to seek out those positions in the future. A similar relationship for Black women did not emerge. It may be that legitimacy issues between communities and police agencies are greater for Black women, and thus it takes more effort to convince Black women that policing is an appealing career. Moreover, gendered policies that regulate, for example, hair length and styles may be more of a deterrent for Black than Hispanic women (Kringen and Novich, 2018). Lastly, with so few applications from Hispanic women (a mean of only 35.09 even in the 2014–2016 period), there is more room for growth than there is for White and Black women.

Our study had some limitations. First, our sample consisted of agencies that were accredited by CALEA. We may have encountered a selection effect where accredited agencies are more likely to embrace gender diversification. Second, our data were grouped into two 3-year periods. Our results may have been more precise if we had complete data from each agency for each year from 2011 to 2016. Finally, our data did not include information regarding departmental policies or recruitment practices. Agencies that implement more “women-friendly” policies (e.g. maternity leave, day care, etc.) may have better success developing gender-diverse applicant pools (Murray, 2021). We were unable to measure specific recruitment strategies and workplace policies and practices, leaving unanswered questions about what precise steps agencies might take to achieve more diverse applicant pools and more successful hiring outcomes.

Future research should continue to explore gender diversification in policing. Scholars would do well to examine more characteristics internal to police organizations. Examining policies, organizational structures, and other elements of gendered organizations could prove useful (Cordner and Cordner, 2011; Rossler et al., 2020; Todak, 2017). Research should also further explore how the culture of police workplaces might be perceived as unwelcoming to women (Brown et al., 2020) or might “create facades of inclusion” (Matusiak and Matusiak, 2018, p. 327). Acker's (1990) foundational work is instructive in understanding how professions linked with one gender may develop organizational structures and assumptions that contribute to the maintenance of gender segregation.

Future scholarship should examine not only hiring decisions, but also retention of female officers. Some evidence points to stressors in the workplace like gender discrimination (Haarr, 2005; Haarr and Morash, 2013) and violent encounters on the job (McCarty et al., 2007) as contributors to the premature departure of female officers. Less is known about the impact of external strains, such as work-family conflict, on the retention of women in policing. Relatedly, future research should examine promotional outcomes for female officers. While the proportion of women patrol officers is small relative to men, the proportion of women in leadership positions is even smaller (Hyland and Davis, 2019; Schulz, 2003; Shjarback and Todak, 2019). Exploring predictors of women in leadership positions would help advance this literature. Third, examining how structural factors external to the organization (such as aggregate measures of gender inequality) may help understand variations in hiring outcomes for women. Fourth, the continued accumulation of research evidence supporting the need for women in policing to develop and deploy “coping strategies” (Yu, 2018) to “gain status” (Rabe-Hemp, 2009) within the work environment. This suggests that recruitment and hiring alone are not enough to overcome the underrepresentation of women. Until culture and workplace policies change, broad and sustained advances in workplace diversity are likely to remain elusive; future research might better inform the pathways to reform. Lastly, research that explores the relationship between diverse officer rosters and a range of outcomes, including police legitimacy and use of force is needed.

The results from this study suggest that diversification is related to characteristics of the organizations making hiring decisions. Agencies that have committed to diversification by creating diverse officer rosters are the ones most likely to hire from underrepresented groups in the future. Interestingly, committing to diversification may serve as a catalyst for change in department cultures, especially on issues where there are conflicting beliefs of problems within the department between majority and underrepresented officers (e.g. gender discrimination) (Dowler and Arai, 2008). At least in some cases, hiring decisions at one point can have implications for the composition of applicant pools at a later point. Police agencies should understand the benefits of a diverse police force. This is the first step in making such an organization a reality.

Table 1

Descriptive statistics (n = 127)

Variable	Mean	SD
Women hires (all periods)	13.76	22.12
Black women hires (all periods)	2.85	8.09
Hispanic women hires (all periods)	1.68	6.52
Black men hires (all periods)	10.63	37.53
Hispanic men hires (all periods)	6.74	18.67
White women hires (all periods)	9.23	14.86
White men hires (all periods)	52.71	83.07
Women applications (Period B)	209.63	443.67
Black women applications (Period B)	68.69	171.87
Hispanic women applications (Period B)	35.09	97.74
Percentage of women hires (Period A)	14.06	14.17
Percentage of Black women hires (Period A)	2.21	5.70
Percentage of Hispanic women hires (Period A)	0.99	3.22
Percentage of women applications (Period A)	13.43	6.24
Percentage of Black women applications (Period A)	3.69	4.68
Percentage of Hispanic women applications (Period A)	1.37	1.61
Percent Black	14.00	16.26
Percent Hispanic	11.01	12.42
Total officers	281.82	591.28
Percentage of Black officers	8.60	12.42
Percentage of Hispanic officers	6.15	10.76
Percentage of policewomen	12.40	6.21

Table 2

Hiring outcome negative binomial regression models

Variable	Women hires	Black women hires	Hispanic women hires
Black men hires (ln)	0.033 (0.049)	0.155 (0.109)	0.232* (0.113)
Hispanic men hires (ln)	0.015 (0.042)	−0.173* (0.081)	0.239 (0.138)
White women hires (ln)	–	−0.252 (0.138)	−0.198 (0.144)
Black women hires (ln)	–	–	0.072 (0.095)
Hispanic women hires (ln)	–	0.104 (0.077)	–
White men hires (ln)	0.471*** (0.150)	0.503 (0.358)	0.008 (0.368)
Percent Black	−0.010 (0.009)	−0.003 (0.013)	0.0002 (0.020)
Percent Hispanic	−0.015 (0.013)	−0.039 (0.024)	−0.078* (0.030)
Department size (ln)	−0.516** (0.152)	−0.253 (0.282)	−0.247 (0.273)
Percent Black officers	0.041*** (0.011)	0.025 (0.017)	−0.062 (0.040)
Percent Hispanic officers	0.033* (0.015)	0.013 (0.022)	0.076** (0.028)
Percent policewomen	0.035 (0.015)	0.101*** (0.026)	0.073* (0.034)
Constant	−2.569*** (0.633)	−4.917*** (1.152)	−2.361 (1.459)
Alpha	0.674 (0.098)	0.804 (0.230)	0.717 (0.292)
Log likelihood	−409.529	−182.988	−130.256

Note(s): *p < 0.05; **p < 0.01; ***p < 0.001

Table 3

Application (Period B) negative binomial regression models

Variable	Women applications	Black women applications	Hispanic women applications
Percent women hires (Period A)	−0.009 (0.029)	–	–
Percent women applications (Period A)	0.013 (0.057)	–	–
Percent Black women hires (Period A)	–	−0.015 (0.109)	–
Percent Black women applications (Period A)	–	0.285 (0.148)	–
Percent Hispanic women hires (Period A)	–	–	0.209* (0.098)
Percent Hispanic women applications (Period A)	–	–	0.001 (0.231)
Percent Black	0.072* (0.033)	0.0005 (0.046)	−0.016 (0.031)
Percent Hispanic	−0.037 (0.057)	0.109* (0.045)	0.009 (0.036)
Total officers (ln)	−0.230 (0.212)	0.242 (0.296)	0.606* (0.244)
Percent Black officers	−0.131** (0.045)	−0.071 (0.062)	−0.031 (0.043)
Percent Hispanic officers	0.142* (0.060)	−0.047 (0.060)	−0.007 (0.060)
Percent policewomen	−0.058 (0.072)	−0.062 (0.066)	−0.004 (0.052)
Constant	−2.886* (1.142)	−5.599** (1.694)	−7.71*** (1.254)
Alpha	4.879*** (0.516)	5.478*** (0.636)	4.462*** (0.581)
Log-likelihood	−789.429	−561.445	−456.457

Note(s): *p < 0.05; **p < 0.01; ***p < 0.001

Notes

1. This includes the launching of the 30 × 30 initiative (https://30x30initiative.org/ in March, 2021), which asks agencies to pledge that by 2030, 30% of new recruits will be female.

2. Disparities persist within individual agencies and tend to operate as a function of jurisdiction size. Representation by non-White and women officers continues to lag in smaller communities and both groups are underrepresented in leadership positions (Hyland and Davis, 2019; Schulz, 2003).

3. Literature does consider the experiences of women in policing (Morabito and Shelley, 2018; Morash and Haarr, 2012; Rabe-Hemp, 2009), however these studies tend to focus on those working within policing, with less consideration of barriers to entry.

4. Many studies of women in policing rely on small samples, making it difficult to draw meaningful inferences about Black female or Hispanic female officers. Alternatively, research relying on large, multi-agency data sets (e.g. the LEMAS program) often do not allow consideration of the intersectionality because of how the data were collected.

5. Sexuality and gender identity might also condition officers' experiences with police culture, with some research finding hierarchies within policing based on sexual orientation (Brown et al., 2020; Colvin, 2009; Mennicke et al., 2018). Data available in this study did not account for sexual orientation or gender identity.

6. Perhaps one reason for the sparse research in this area is the small percentages of non-White women in these positions. In our study, Black women represented only 2.21% and Hispanic women only 0.99% of hires in Period A (2011–2013).

7. Starting in 2016, CALEA moved to a 4-year accreditation cycle.

8. In this study, we limit our analyses to White, Black, and Hispanic applicants and hires.

9. We performed the Shapiro–Wilk test of normality to inform the decision to log-transform these hiring variables. In each case, the null hypothesis of normality was rejected (p < 0.001).

10. To ensure that the negative binomial model was the most appropriate choice, we compared each negative binomial model to a Poisson model. In each case, the chi-square test was significant (p < 0.001) suggesting that the negative binomial model is a better fit for the data.

11. We treat the number of applications as an exposure variable in the hiring models.

12. We treat available workforce as an exposure variable in the application models.

13. Because many agencies did not hire women (particularly non-White women) in either period, we also ran our models using zero-inflated negative binomial methods. These models produced similar results to the non-inflated models. Moreover, the AIC and BIC scores were nearly identical across models. For these reasons, we present the non-inflated negative binomial models.