Staffing composition in large, US police departments: benchmarking workforce diversity

Abstract

Purpose

Policymakers have long suggested diversifying US police forces, which typically have white male majorities among officers. This article explores to what extent police diversity has changed over time in large agencies, as well as whether different diversity benchmarks may matter for agencies.

Design/methodology/approach

The authors draw data from the Law Enforcement Management and Administrative Statistics (LEMAS) survey for 358 agencies that had at least 100 full-time sworn officers in 1997 and 2016 and that reported officer demographic data to the LEMAS in both years. For a selection of 12 communities – three randomly chosen in each of the four US Census regions – the authors compare officer diversity to Census data on population diversity for different benchmarks.

Findings

There has been some increase in diversity but policing largely remains a white male profession. The authors find only limited variation in diversity by type of benchmark – e.g. total population, working population or recruiting-age population – a community considers. This suggests communities may wish to choose a benchmark they can best measure and seek to increase diversity by it, and research on workforce representation may not be sensitive to benchmark choice. The authors also suggest communities and their police organizations consider other ways to assess diversity, including those that illustrate a broader range of attributes and representation throughout the organization, and that they research and test alternative forms of measurement to gauge whether these findings hold for different modeling approaches.

Research limitations/implications

Our analysis is limited to the largest police agencies and to overall staffing level diversity metrics pertaining to sex, race and Hispanic origin. Still, we find many police agencies have room for greater diversity, which could draw more qualified workers and lead to better policing.

Originality/value

While there has been much attention to police diversity in recent decades, there have been few efforts to compare alternative measurement approaches. This research provides guidance to some initial measures, as well as further considerations communities may wish to make.

Citation

Wilson, J.M. and Grammich, C.A. (2022), "Staffing composition in large, US police departments: benchmarking workforce diversity", Policing: An International Journal, Vol. 45 No. 5, pp. 707-726. https://doi.org/10.1108/PIJPSM-12-2021-0175

Introduction

US police forces have traditionally and primarily comprised white, male officers. More than a half-century ago, the largest police departments in the nation had more than ten white officers for every nonwhite one (Report of the National Advisory Commission on Civil Disorders, 1968). In 1993, more than three-fourths of full-time sworn employees of local US police departments were white males (Reaves, 1996). In 2016, nearly two-thirds of full-time sworn officers were white males (Hyland and Davis, 2019).

Policymakers have long sought to diversify police forces. The President's Commission on Law Enforcement and Administration of Justice (1967, p. 101) noted that a police “department can show convincingly that it does not practice racial discrimination by recruiting minority-group officers, by assigning them fairly to duties of all sorts in all kinds of neighborhoods, and by pursuing promotion policies that are scrupulously fair to such officers.” Nearly 50 years later, The President's Task Force on 21st Century Policing (2015) was charged “with identifying best practices and offering recommendations on how policing practices can promote effective crime reduction while building public trust” (p. 1). The task force report suggested law enforcement agencies incorporate “a broad range of diversity including race, gender, language, life experience, and cultural background to improve” their operations (p. 16). Similarly, the Commission on Accreditation for Law Enforcement Agencies (2017) standards include one that agencies recruit full-time sworn police personnel reflecting the local workforce in ethnic, racial and gender composition. New Jersey in 2021 required the state's Civil Service Commission to help make law enforcement more reflective of the communities it serves (Associated Press, 2021). More than 120 police departments have joined the national 30 × 30 initiative to increase the representation of women in police recruit classes to 30% by 2030 (New York University School of Law, n.d.).

The issue of workforce diversity achieved greater prominence with the May 2020 death of George Floyd, an unarmed African American killed by a white police officer, and subsequent demonstrations against police brutality and systemic racism. This and previous incidents served as a catalyst for national discourse on reforming the police. Calls for change vary significantly by focus, complexity and depth. Regardless of the form such initiatives take, they will, given the bureaucratic complexity of police organizations, take time to realize.

This may be particularly true for any efforts to diversify the police workforce. Given that (1) the treatment of minority residents by the police is at the heart of the police-legitimacy crisis, (2) policing is largely a white, male profession while many large communities are highly diverse and (3) human resources are the most important aspect of any organization, any major police reform should consider the diversity within and among US police workforces.

This article seeks to lay a foundation for discourse on police workforce diversity by examining such diversity among large US police agencies. It focuses on two research questions. First, how, if at all, has the diversity of police staffing changed over time? Second, does the level of community representation vary by different benchmarks? We begin with a review of the desire for agencies to demographically represent their communities and what it means for an agency to do so. After we summarize our research approach, we illustrate long-term trends in police diversity and community representation using various benchmarks. We conclude with a discussion of key findings and their implications.

Background

Support for diversity and community representation

The belief that a police force should represent the characteristics of the community it serves rests on several principles. Among these is that police departments should not discriminate against any community members but provide equal opportunity to all. Departments can demonstrate this by recruiting among all members of the community and providing equal opportunities within the department. Policymakers have also long assumed that more diverse forces could better serve diverse communities. In its internal deliberations, the President's Commission on Law Enforcement and Administration of Justice suggested “that increasing racial diversity could discourage at least overt expressions of racial intolerance by White officers, and [reduce] stereotyping and prejudice” (Skogan, 2018, p. 384). Scholars have also suggested that the presence of minority and female officers can change the predispositions of police subculture (Skogan and Frydl, 2004; Walker, 1998).

More generally, research on representative bureaucracy suggests police forces that are representative of their community can reduce community conflict (Trochmann and Gover, 2016). Such benefits are likely to be greater when the proportion of minority officers is greater and when they are dispersed across all ranks (Hong, 2021) and can be best drawn out with accompanying organizational strategy that seeks to leverage representativeness in better performance (Andrews et al., 2005).

The “business case” for recruiting more minority and female officers includes expanding the pool of potential qualified and talented workers and hence enhancing organizational performance (see, for example, Starheim, 2019). Policymakers may assume that there are certain circumstances in which minority and female officers perform better than white male officers. For example, African American officers may be less aggressive toward, and better relate to, African American citizens, and women may be more receptive to community needs and use less coercion when dealing with the public. A recent analysis of Chicago found that, relative to white officers, “Black and Hispanic officers make far fewer stops and arrests, and they use force less often, especially against Black civilians,” with such effects being greatest in majority-black areas and stemming “from reduced focus on enforcing low-level offenses” (Ba et al., 2021, p. 696). Another study suggests that while white officers are more coercive toward black suspects, use of force by black officers does not differ by race (Paoline et al., 2018). A study of English and Welsh police forces suggests that greater ethnic minority representation on a police force can reduce complaints of police misconduct, particularly from black residents, but that residents may be less satisfied with the service that minority officers provide (Hong, 2016).

Others contend that the general scholarly evidence for such claims may be lacking (see, for example, discussions in Skogan and Frydl, 2004, Brunson and Gau, 2015, Ozkan et al., 2016 and Carpenter, 2019). One analysis of US jurisdictions with at least 50,000 residents found more diversified police workforces led to reduced crime control and more personnel turnover, while also suggesting that diversity management practices could offset these adverse effects (Hur, 2012). And an analysis of traffic stops by more than 150 local agencies in Illinois and Missouri found that greater minority representation on a force did not reduce disparities in traffic stops, raising questions of whether representative bureaucracy—by which fair representation of different groups in government agencies will ensure that public interests are reflected in an agency—can be achieved solely through diverse police organizations (Shjarback et al., 2016).

Research on the effects of more women on a force has been more positive and consistent. One recent study contends that residents believe police forces with more women officers will be less prone to corruption (Barnes et al., 2018). Higher proportions of female officers may also lead to improved police quality, particularly in reporting and prevention of domestic violence incidents (Miller and Segal, 2014). Some scholars find that female officers are less likely to use force in general and severe force in particular than male officers are, though these differences may depend on the type of roles female officers serve as well as the presence of supervising officers (Bolger, 2015; Novak et al., 2011), and others claim little difference between men and women officers in the use of coercion (Paoline and Terrill, 2004). Other research suggests that increasing female representation on police forces leads to greater adherence to norms and practices that, in turn, help reduce complaints about inappropriate use of force (Schuck and Rabe-Hemp, 2016). Female officers may also be particularly adept at assessing the policing needs of diverse communities (Black and Kari, 2010; Schuck, 2017). At the same time, women face many longstanding challenges to being accepted in all aspects of police work (Archbold and Schulz, 2012) and particularly in accessing leadership positions where they might sustain their gains within a department (Archbold and Schulz, 2008; Archbold and Hassell, 2009).

Citizens themselves have also expressed a desire for the makeup of police departments to reflect the composition of their communities. One initiative to “end police violence” contends, “First responders should reflect and be responsive to the cultural, racial and gender diversity of the communities they are supposed to serve” (Campaign Zero, 2021, p. 6). Indeed, when police forces do not reflect the population, officers may find community trust erodes, which can make it harder recruit diverse officers, particularly in smaller jurisdictions with less turnover that are undergoing demographic transition (Maciag, 2015).

Previous research suggests that diversification can be self-reinforcing. Rather than producing a “zero-sum” effect, agencies that hire from one under-represented group may be more likely to hire from other under-represented groups, with agencies hiring a larger share of women officers more likely to hire still more women applicants (Nowacki et al., 2021a). Previous research also suggests agencies with existing workforce diversity are more likely to make more subsequent non-white hires (Nowacki et al., 2021b). Previous research using Law Enforcement Management and Administrative Statistics (LEMAS) data finds agency size, vertical differentiation and accreditation by the Commission on Accreditation for Law Enforcement Agencies all have positive influences on proportions of officers who are female, non-white and African American (Morabito and Shelley, 2015). More recent research analyzing LEMAS data through 2013 finds slowing growth in representation of women among full-time sworn law-enforcement officers, but that organizational size and percentage of black full-time sworn officers are among the most consistent predictors of the presence of female officers (Matusiak and Matusiak, 2018).

Gauging community representation

Despite widespread claims that the composition of police organizations should reflect the composition of the communities they serve, there has been little comparative examination of benchmarks that could be used to assess progress toward representativeness or advance theoretical development and modeling. Initial reactions may be that the police force of a community should reflect the composition of its total population. Communities that draw in large numbers of individuals from elsewhere (e.g. commuters) might consider having agencies that reflect this broader population. At the same time, there are many individuals, such as children, the elderly or those who are institutionalized, whom the police cannot recruit. This has led some to contend that police diversity should match the working population (Commission on Accreditation for Law Enforcement Agencies, 2017). Yet even among the working population there are many whom the police cannot recruit because they do not meet the minimum standards to become sworn police officers, suggesting yet another benchmark might be the qualified workforce.

Different components relevant to police recruiting may change over time. There are, for example, compositional differences between the total population and that by age groups, which can affect the pool from which police departments may draw recruits. Police agencies typically seek candidates who are, roughly, between 20 and 40 years of age (Taylor et al., 2005). At the same time, many police departments are seeking more educated candidates who may have fewer disciplinary problems and incidents with use of force (Aamodt, 2004; Paoline and Terrill, 2007). Other restrictions that police departments place on recruits can further distort the pool of potential officers.

Other recruitment policies may adversely affect recruits needed for diversity. The use of veterans' preferences in hiring law enforcement officers may disproportionately disadvantage women (U.S. Department of Justice and US Equal Opportunity Employment Commission, 2016). Use of criminal background checks may disproportionately impact minority applicants, as may admissions of illegal drug use (Matthies et al., 2012). Credit checks may exclude potential officers from low-income or predominantly minority communities, where residents are more likely to have lower credit scores (U.S. Department of Justice and US Equal Opportunity Employment Commission, 2016). Citizenship requirements can further constrain officer diversity, including in communities with large numbers of residents who speak foreign languages (Matthies et al., 2012). Application costs and lengthy hiring processes can deter applicants from underrepresented populations (U.S. Department of Justice and US Equal Employment Opportunity Commission, 2016).

Our point here is not the appropriateness of such restrictions. Rather, it is that such qualifications may affect pools of eligible applicants. Communities may wish to balance diversity goals of their police departments with what their qualifications yield. They also may wish to benchmark their goals with what their qualifications yield.

Benchmarks may fall on a continuum from the very broad to the very specific, such as

1. The entire community served (e.g. including commuters)

2. All residents in the jurisdiction (e.g. all ages)

3. All adult residents

4. Residents of police hiring age (e.g. 20–40 years old)

5. Residents of hiring age meeting certain qualifications (e.g. education).

The choice of benchmarks, and how to measure them, may vary by community preferences (and, of course, some may prefer not to use any specific benchmark at all). The ability to acquire timely and useful data will also affect how well an agency can be assessed relative to its diversity goals.

Methods

Research questions

Given the longstanding calls to increase police diversity as well as the lack of clarity on how to measure such diversity, this research seeks to answer two questions. First, how, if at all, has the diversity of police staffing changed over time? Second, does the level of community representation vary by different benchmarks? These questions are difficult to answer given the sporadic, limited and inconsistent data available on US law enforcement workforces (Wilson and Heinonen, 2011). Fortunately, we were able to draw meaningful information on police diversity from the LEMAS survey and on community diversity from the US Census.

Data/approach

The LEMAS survey, conducted periodically since 1987 by the Bureau of Justice Statistics, collects data from a sample of more than 3,000 US law enforcement agencies, including all those that employ at least 100 full-time sworn officers (US Bureau of Justice Statistics, 2020). In addition to administrative, structural, capacity and policy data, it collects demographic data on officers. We assessed LEMAS data for 1997 and 2016, the most recent year available for these data. That the most recent LEMAS survey is six years old (at the time of this writing) is itself a significant indicator of the challenges of conducting timely, policy-relevant research regarding the administration of police organizations. The variables LEMAS collects and the ways it operationalizes them, especially in terms of personnel composition, vary by wave and inhibit tracking detailed changes by wave. Fortunately, the 1997 and 2016 LEMAS surveys collected staffing data consistently, facilitating longitudinal comparison of staffing demographics in terms of race/ethnicity and sex.

By using the 1997 and 2016 LEMAS surveys, we can assess any change in workforce demographics over a relatively recent two-decade span. We focused on large municipal police organizations given LEMAS data for all agencies with 100 or more full-time equivalent officers. There is greater variability in community composition in larger communities, which is an important consideration in police staffing. Many high-profile police-community incidents occur in communities policed by large, municipal agencies. Because of their size, larger agencies are also more likely to exhibit change and are responsible for policing the largest proportion of the US population.

The LEMAS collected data on 475 large municipal agencies in 1997 and on 537 in 2016. Large agencies fall into or out of the LEMAS over time for many reasons, including nonresponse, changes in staffing size and organizational changes including consolidation and disbanding. To best assess change over time, we merged the LEMAS surveys, matching agencies between the two waves. This yielded a sample of 395 large municipal agencies in both the 1997 and 2016 waves. Of these, 358 had both staffing and demographic subgroup data available in both years and at least 100 sworn full-time (rather than full-time equivalent) officers. Our final sample for gauging workforce diversity over time included all 358 agencies with at least 100 full-time officers that provided workforce data in both the 1997 and 2016 LEMAS surveys.

To demonstrate the effects of different benchmarks, we randomly selected three departments in each of the four Census regions, for a total of 12 departments. These were.

1. Northeastern departments of

   • Trenton, NJ

   • Allentown, PA

   • Pittsburgh, PA

2. Southern departments of

   • Shreveport, LA

   • Suffolk, VA

   • Hagerstown, MD

3. Midwestern departments of

   • Elkhart, IN

   • Muncie, IN

   • Springfield, OH

4. Western departments of

   • Everett, WA

   • Rialto, CA

   • Spokane, WA.

For each of these, we compiled demographic data from the 2016 Census Bureau American Community Survey (ACS) to form three sets of community benchmarks. These reflect 1) the entire residential community of all ages, 2) the residential community of working age (18–64) and 3) the residential community of a typical recruit age (20–34). For each benchmark, we gathered data to calculate the proportions of non-Hispanic single-race white males, minorities (i.e. those who are not single-race non-Hispanic whites) and females.

Unfortunately, while LEMAS categories of officers are exhaustive and mutually exclusive, Census Bureau ACS population estimates by community for 2016 are not. For example, these estimates provide data by age and sex for the single-race African American population, but not for the non-Hispanic single-race African American population. These estimates do, however, provide estimates by age and sex for the single-race non-Hispanic white population. Hence, we limit benchmark calculations to categories for the total population, the single-race non-Hispanic white population, and the remaining minority population—which, in some of these 12 communities, comprises most of the population.

For each benchmark in each community, we compare the proportion of each demographic attribute represented among the police staff to the corresponding attribute proportion represented among the residential population. Assuming the “goal” is for the proportion of each attribute's representation among police staff to match that in the community, we divide the former by the latter to estimate a “percentage to goal” (PTG) achievement metric. For example, if minorities represent 20% of police staff and 40% of residents in a community, the percentage to goal is 50 (20/40). If the proportions are equal between the agency and community (e.g. 35%), the agency has achieved full representation and the percentage to goal is 100% (35/35).

Limitations

There are some limitations to our approach. While we include most large, municipal police organizations that existed between 1997 and 2016, some such agencies were missing in our data. We also do not account for agencies that fell below our threshold of 100 full-time sworn officers from 1997 to 2016, nor do we account for other types of policing agencies such as sheriff/county departments, state agencies, tribal agencies and special jurisdictions. The failure of LEMAS to consistently measure demographic characteristics or to make more recent data available prevents us from assessing any variation in trends over time or from describing a more current picture of police staffing.

While the benchmark analysis is meant to be illustrative and it is randomly drawn, it is based on a small sample. While it accomplishes its goal to exemplify how the benchmarks work and their variation in practice, this analysis should not be used to characterize the nature of all large municipal police organizations. Limitations in the Census data also prevented us from examining more refined aspects of race and ethnicity in the sampled communities, which future research may seek to do.

Finally, our analysis is limited to the two aspects of community composition available in both the LEMAS and the ACS: race/ethnicity and sex. While these are the most commonly discussed and collected attributes of diversity, future research should consider other characteristics such as gender identity, cultural identity, primary language and life experience.

Results

Workforce composition

We begin by examining workforce composition. Table 1 illustrates the demographic characteristics among large agencies and their change in recent decades. It does so by showing the lowest, highest and mean proportions that each group had among officers in the 358 departments that we examined in 1997 and 2016. In 1997, for example, the male percentage of full-time sworn officers ranged from 71% (Pittsburgh, PA) to 100% (Dearborn, MI, and West New York, NJ). The average proportion male across all these agencies was 90%. By 2016, only one department (Dover Township, NJ) had more than 98% male, while two departments (Bessemer, AL, and Ann Arbor, MI) were more than one-third female. Across all departments, however, the average proportion male decreased only slightly, from 90% in 1997 to 88% in 2016.

Similar results are evident for other demographic groups: dramatic changes in some communities, but more incremental change across all. In 1997, three departments (Warren, MI, Cheektowaga, NY, and Troy, NY) had no minority officers. That is, all their sworn officers were non-Hispanic white. Most departments had fewer than 20% minority officers. By 2016, all 358 departments had at least 2% minority officers, and nearly half had at least 25% minority officers. The average minority share across all departments increased from 19% to 25%, a substantial increase but not matching that of the overall minority population across the nation. Among minority groups, increases for males were greater than for females, and increases for Hispanics were greater than for others. The increase in average Hispanic share across departments, from 7% to 12%, occurred at a time when the Hispanic population of the nation nearly doubled (Noe-Bustamente et al., 2020). This raises questions of how changes in department representation reflect changes in community representation, which we will discuss below.

As minority and female representation on departments has increased, white and, particularly, white male representation has decreased, but both groups still remain dominant among large-agency officers. From 1997 to 2016, the average percentage that whites represented of full-time sworn officers decreased from 81% to 74%. White males, the only race/origin and sex group to have officers in every department in 1997, saw their average percentage by department decrease from 74 to 66.

Across all 358 agencies, diversity did increase for nearly every population group we examined. We calculate a paired-sample t-test score of 13.909 (p < 0.001) for the mean minority percentage of officers in 2016 (26%) and 1997 (19%). We get similar results indicating increasing diversity for every other officer group in Table 1 across the 358 agencies. That is, we get positive and statistically significant results for nearly all groups outside whites and white males.

We did not find statistically significant changes for males and females among American Indian and Alaska Natives and males among African Americans. The minimal change for American Indian and Alaska Natives may be due to low numbers among both officers and population of these places. Of the 358 agencies we examined, 154 had no American Indian or Alaska Native officers in 1997 or 2016. Only seven of these communities had single-race American Indian and Alaska Native populations that exceeded 3% in 2010, and only one (Anchorage, 8%) had such populations exceed 5% (Norris et al., 2012). Among the seven communities whose American Indian and Alaska Native populations exceeded 3% in 2010, the American Indian and Alaska Native proportion of officers increased from 2.8 to 3.9%. The three communities where this increase was greatest were all in Oklahoma: Tulsa, where the number of such officers increased from 51 (6%) to 78 (11%), Oklahoma City, where it increased from 22 (2%) to 35 (3%), and Norman, where it increased from 0 (0%) to 4 (3%).

African American males held positions in more departments than American Indians and Alaska Natives did. Of the 358 agencies we examined, only 14 did not have African American male officers in either year. In most of the 358 agencies, the proportion of African American male officers increased. In six of them (East Point, GA, Pine Bluff, AR, Montgomery, AL, Hempstead, NY, Memphis, TN, and Mount Vernon, NY), the proportion of African American male officers increased more than 10% points. Nevertheless, these gains were nearly offset by decreases elsewhere. Some of these decreases coincided with African American population decreases. In Charleston, SC, for example, the African American male share of officers decreased 15% points (and the African American female share decreased 8% points), following a period when the population of the city changed from two-thirds African American to two-thirds white (Slade, 2020). In Inglewood, CA, the African American male share of officers decreased 12% points (and the African American female share decreased 16% points) while the African American share of the city population decreased 6% points (Southern California Association of Governments, 2019). In Spokane, the African American male share of officers decreased 10% points (and the African American female share decreased 14% points) while the non-Hispanic single-race African American share of the city population grew modestly (from 2.0% in 2000 to 2.2% in 2010 according to decennial Census counts). Future research may wish to explore why African American shares of officers did not increase as other measures of diversity did, and why it decreased further than African American changes of population in some locations.

There is considerable variation elsewhere worth exploring. This is evident in the standard deviations shown in Table 1. These are particularly large for minority, minority male, white, white male, African American and Hispanic officers in both years. Furthermore, the standard deviations have grown for each group shown in Table 1. In other words, while department diversity is growing, so is the variation in these measures. This may indicate varying approaches to and success in promoting officer diversity by communities across the nation. It may also indicate substantial and growing swings in officer diversity.

Community benchmarking

Next, we assess benchmarks for illustrating the extent to which the diversity of police staffs represents the communities they serve. Figure 1 shows the 12 departments we selected, three randomly chosen in each Census region, for further analysis are much like other departments with a predominance of white male officers. White males comprise 47 to 92% of sworn officers in these 12 jurisdictions, but only 5 to 39% of the total population. If white males had similar proportions of sworn officers than they have of the total population, then the 12 dots of the scatterplot would lie closer to the dotted diagonal across the figure, rather than near the solid trend line for these observations in the upper left quadrant of the figure.

Figure 2 also shows these departments are like other departments in having minority percentages of officers below their corresponding numbers in the total population. Minority members of these communities comprise 4 to 48% of sworn officers, but 18 to 90% of their populations. If minorities had similar proportions of officers than they have in the total population, then the 12 dots of the scatterplot would lie closer to the dotted diagonal across this figure as well, rather than the solid trend line for these observations in the lower half of the figure.

Finally, Figure 3 shows these departments are like others in that women, while comprising about half the population, account for only 5 to 25% of the sworn officers in these cities. If women accounted for the same proportion of sworn officers as they have in the population, then the 12 dots of this figure would cluster in the middle, near the 50% spot on both axes.

Beyond this graphical interpretation, for each of these 12 cities we compare the percentage that each of eight groups—males, females, minorities, minority males, minority females, single-race non-Hispanic whites, single-race non-Hispanic white males and single-race non-Hispanic white females—has in the department with the percentage they have in three benchmark populations (total, working-age and recruit-age). Table 2 presents our results.

For example, in Trenton, NJ, the LEMAS indicates there were 223 male officers on a force of 254 officers in 2016. That is, males were 87.80% of the force. That same year, the five-year sample estimate of the Census Bureau's ACS shows the Trenton population was 51.05% male. Dividing 87.80 by 51.05, we find a ratio of 171.99%. That is, Trenton had 172% of the male officers it needed to ensure that male representation among sworn officers matched that in the population. As noted earlier, we call this figure the “percentage to goal” (PTG) for a community. If the Trenton department had 130 (or, more precisely 129.65) male officers, then its PTG for males in the whole community would have been 100%, rather than 171.99%. We made similar calculations for the working-age population (ages 18 to 64) and the recruit-age population (ages 20 to 34). As shown, the proportion of males among Trenton's officers was 163% of their proportion among the Trenton working-age population, and 162% of their proportion among the Trenton recruit-age population. We also made similar calculations for females, the minority population (i.e. the population that is not single-race non-Hispanic white), minority males and females, the white population and white males and females.

Four agencies—Shreveport, LA, Suffolk, VA, Rialto, CA, and Spokane, WA—show minority male representation among officers nearly equal to that in the population. In Rialto and Shreveport, minority male gains in proportion of officers outstripped that of the population between 1997 and 2016. In Suffolk, minority male share of officers grew slightly while that of the population decreased sharply. In Spokane, minority male share of officers decreased while that of the population increased.

Suffolk, Shreveport and Pittsburgh had the highest shares of female officers. In Suffolk and Shreveport, the female share of officers has increased in recent decades, while that in Pittsburgh has decreased. Pittsburgh may offer a cautionary tale regarding efforts to diversify police forces. A 1970s court order “demanding that for every white man hired, the force had to hire a woman and two members of minorities, one a man and one a woman,” sharply increased force diversity, to the point where women were 25% of agency officers, the highest proportion in the nation, in the 1990s (Janofsky, 1998). After expiration of the court order, however, the department underwent what some called a “re-whitening,” with the proportion of female officers also decreasing to about half their previous level (Dudkiewicz, 2016). Local attempts to again increase diversity included new courses on policing in city high schools, job fairs and Church meetings, though local officials noted the need to do more, including through nationwide recruiting.

The Pittsburgh experience roughly parallels that of Spokane. In Spokane, concerns in the 1990s about department relations with the black community led to a special hiring initiative for a department that previously had no black officers (Sheingold, 1999). This effort also included a community-wide forum on race as well as documentation of past and future efforts. By 2016, as noted, however, Spokane had lost this progress.

The two agencies with the greatest overrepresentation of white males, Rialto, CA, and Trenton, NJ, have small and shrinking white male populations. In 1997 in Rialto, white males had a 7.7% share of the population and a 59.5% share of officers; by 2016, these shares had decreased to 4.7% of the population and 46.7% of officers. In 1997 in Trenton, white males had a 10.6% share of the population and 73.9% of officers; by 2016 these shares had decreased to 7.0% of the population and 50.8% of officers. In both cities, both the total population and the total number of officers had decreased, while the number of officers in categories besides white males had increased. Still, white males remained greatly overrepresented. Furthermore, in both cities the white male percentage-to-goal value by the community recruit age benchmark was even higher. In other words, these departments look even less diverse relative to their recruit-age population. This may present both a challenge and an opportunity to these communities: these departments may need to diversify considerably more to match their younger populations, but the greater diversity of the younger populations may help them do so.

Our random sample of three agencies within the Midwest included three agencies that are all within 200 miles of each other, as well as two within the same state. There are some notable differences among these agencies, which may illustrate how different agencies within the same state approach the question of diversifying their workforces. Among these three, Springfield has the highest percentage-to-goal values for minority officers while Elkhart has the highest values for white officers. Elkhart and Springfield have similar minority shares of the population, but Elkhart has a higher share of white male officers, which has increased over time. In Muncie, both the number of sworn officers and the number of white male officers have decreased. Several sources of variation could be affecting these differences. For example, the Elkhart population is more likely to be foreign-born, more likely to not speak English at home and less likely to have completed high school or college than that in Muncie and Springfield (U.S. Census Bureau, 2022). All these characteristics may affect the pool of eligible officers in each community and the composition of that pool. All in turn may also affect whatever benchmark a community wishes to use to gauge diversification.

Discussion

In many communities, sworn officers do not reflect broad community demographics. In few communities do those who are not non-Hispanic white males have representation on the force reflecting their representation in the community. There are historical and possible policy explanations for this. What diversity goals agencies should have, and how they should meet these and myriad other demands on police qualifications, is a decision best left to communities. Our goal here has been to illustrate whether progress has been made and standards that might be considered in gauging how well a given police agency reflects its community.

We found some modest increase in diversity over time in large agencies, but also wide variation in this diversity. Our more detailed sample of 12 large agencies helped illustrate some of the issues here. Our analysis of these agencies show, as previous research has found, increasing diversity, but diversity that still does not match the diversity of the communities they serve. Furthermore, in some agencies, there has been regression from previous advances.

Our work also illustrates, particularly in the experiences of Pittsburgh and Spokane, that diversity is not always self-reinforcing. This helps to contextualize recent research that indicates current workforce diversity predicts future hiring diversity (Nowacki et al., 2021a, b). Perhaps it is the genesis of change that is important for understanding the sustainability and cumulative effects of workforce diversity over time. This is an important area for future research to explore.

Diversification over time and between and within organizations may depend on several influences. These may include community influences such as population composition, organizational influences such as needs to meet specific standards or recruiting and other policies. All these may affect what benchmarks a community chooses and how it meets them, or whether to use a quantitative benchmark at all.

While diversity in allocation is a primary goal, so is diversity in deployment, that is, assigning officers to work in specific locations where their attributes align with the areas they are serving. Residents may desire police in their neighborhoods who have similar backgrounds and attributes to themselves. While having diversity in allocation may help agencies achieve diversity in deployment, it does not by itself ensure the community sees the diversity it desires. Achieving diversity in deployment is more of an operational than a strategic feat. Further complicating such a goal, officers may resist assignments based on their demographic attributes. Diversity in deployment is substantially more difficult to assess than diversity in allocation. It requires detailed data on police officer attributes and assignments as well as demographic data on residents in specific neighborhoods or patrol areas. While beyond the scope of this article, such analysis is necessary for assessing the effects of personnel diversity on relations between police and the community.

Given the importance of diversity to building effective police organizations, the number of ways to approach its assessment, and the current challenge of building and maintaining diverse workforces, it is important for agencies and communities to work together to define how workforce composition should be gauged relative to the local community, build capacity to measure it (should they choose to do so), define criteria to evaluate changes over time and formulate strategic plans for promoting workforce diversity. Key to this process is agreement by the police and community on the proper yardstick to evaluate workforce composition. While our work focused on staffing composition of overall workforce levels, there are many additional areas of workforce diversity to consider, even beyond deployment, such as staffing composition in hiring and selection, assignment, promotional consideration, rank, leadership, occupation (e.g. sworn vs civilian), work scheduling and still others.

Implications

We sought to answer two research questions in this work. First, how, if at all, has the diversity of police staffing changed over time? Second, does the level of community representation vary by different benchmarks?

We found that diversity has increased somewhat, but that police agencies largely still do not reflect the diversity of the communities they serve. Our findings for large agencies reinforce that of recent research, showing some growing diversification over time but at slow rates. Furthermore, the rates of change have varied widely. And in two communities that we studied more in-depth, Pittsburgh and Spokane, there has been some regression from the advances evident in the 1990s, as court orders or more detailed plans expired.

As for assessing diversity and its change over time, our findings suggest that the choice of a specific metric, should a community or police agency wish to gauge diversity quantitatively, may not matter greatly for measuring diversity by sex, race and Hispanic origin. Regardless of the benchmark chosen—whole community, working-age population or recruit-age population—we found little variation by associated metrics. Future research should explore whether other refinements might matter more in gauging diversity, such as basing a benchmark on the composition of residents who meet the minimum qualifications of becoming a police officer in the community or demonstrate an interest in the profession, or using other intervals of Census estimates (e.g. five year) of community composition that may have less standard error. Our present work suggests a community may, if seeking to gauge diversity in its force by sex, race and Hispanic origin, choose the metric it can most easily and reliably measure and assess its progress by that metric. Consistent use of a single metric that a community can easily and reliably measure over time may also help communities identify where they are making progress, where progress may be lost and where progress may still need to be achieved. Likewise, the limited variation in the metrics suggests future theoretical development and empirical modeling regarding the relationship between workforce and community composition may not be sensitive benchmark choice. Future research should examine a larger sample of communities to benchmark and test alternative forms of measurement to discover whether these findings hold for other communities and different ways of comparing workforce to community diversity.

Figures

Figure 1

Non-Hispanic single-race white male percentages of total population and full-time sworn officers in 12 select cities 2016

Figure 2

Minority percentages of total population and full-time sworn officers in 12 select cities 2016

Figure 3

Female percentages of total population and full-time sworn officers in 12 select cities 2016

Table 1

Diversity in large municipal police Agencies, 1997 and 2016 (n = 358)

	1997				2016				Difference
	Lowest (%)	Highest (%)	Mean (%)	Std. Dev. (%)	Lowest (%)	Highest (%)	Mean (%)	Std. Dev. (%)	Paired sample t-test	p-value
Male	70.68	100.00	90.45	4.69	58.62	98.13	87.84	5.29	−10.808	0.000
Female	0.00	29.32	9.55	4.69	1.88	41.38	12.16	5.29	10.808	0.000
Minority	0.00	99.63	18.69	14.74	1.52	97.04	25.99	18.45	13.909	0.000
Minority Male	0.00	97.03	16.17	12.72	0.51	85.84	21.77	15.25	12.844	0.000
Minority Female	0.00	23.02	2.52	2.85	0.00	31.40	4.22	4.15	10.454	0.000
White	0.37	100.00	81.31	14.74	2.96	98.48	73.54	18.55	−14.46	0.000
White Male	0.37	97.60	74.27	14.78	2.96	95.00	65.65	17.40	−16.147	0.000
White Female	0.00	24.46	7.04	3.65	0.00	30.14	7.88	4.16	4.69	0.000
Black	0.00	68.60	10.16	11.23	0.00	84.06	10.81	12.31	2.352	0.019
Black Male	0.00	59.65	8.43	8.99	0.00	55.07	8.68	9.11	1.234	0.218
Black Female	0.00	22.03	1.73	2.62	0.00	28.99	2.13	3.55	3.091	0.002
Hispanic	0.00	99.63	6.86	10.47	0.00	96.83	12.26	15.14	13.336	0.000
Hispanic Male	0.00	97.03	6.21	9.79	0.00	85.84	10.51	13.29	12.324	0.000
Hispanic Female	0.00	8.15	0.65	1.03	0.00	10.99	1.75	2.11	12.489	0.000
Am Indian/Alaskan	0.00	10.08	0.44	0.93	0.00	12.24	0.39	1.00	−0.907	0.365
Am Indian/Alaskan Male	0.00	10.08	0.39	0.85	0.00	11.56	0.33	0.89	−1.109	0.268
Am Indian/Alaskan Female	0.00	2.13	0.05	0.17	0.00	1.99	0.05	0.20	0.577	0.564
Asian/Haw/Pac Island	0.00	14.16	1.23	1.97	0.00	22.28	2.18	2.86	9.346	0.000
Asian/Haw/Pac Island Male	0.00	13.26	1.14	1.86	0.00	20.25	1.96	2.59	8.607	0.000
Asian/Haw/Pac Island Female	0.00	1.45	0.09%	0.22	0.00	2.88	0.23	0.44	6.653	0.000

Table 2

Diversity percent-to-goal by benchmark in sample agencies, Northeastern and Southern Regions, 2016

	Community whole (all ages) (PTG) (%)	Community working age (PTG) (%)	Community recruit age (PTG) (%)	Community whole (all ages) (PTG) (%)	Community working age (PTG) (%)	Community recruit age (PTG) (%)	Community whole (all ages) (PTG) (%)	Community working age (PTG) (%)	Community recruit age (PTG) (%)
Northeastern Region	Trenton Police (NJ)			Allentown City Police (PA)			Pittsburgh City Police (PA)
Male	172	163	162	197	196	197	172	168	162
Female	25	26	27	9	9	9	31	32	33
Minority	52	53	51	19	20	18	42	45	45
Minority Male	84	81	76	39	40	35	65	70	67
Minority Female	20	21	20	0	0	0	20	22	23
White	402	387	512	244	233	298	132	127	127
White Male	725	630	857	501	459	627	229	213	206
White Female	59	65	83	24	24	28	37	37	38
Southern Region	Shreveport Police (LA)			Suffolk Police (VA)			Hagerstown Police (MD)
Male	169	168	166	156	156	154	192	190	199
Female	39	39	40	48	48	48	16	16	16
Minority	65	65	63	52	52	50	15	17	16
Minority Male	95	98	91	91	93	87	30	35	39
Minority Female	39	39	38	17	17	16	0	0	0
White	156	155	166	147	146	152	140	132	135
White Male	284	270	289	214	210	218	270	251	256
White Female	39	40	43	80	81	85	23	22	23

	Community whole (all ages) (PTG) (%)	Community working age (PTG) (%)	Community recruit age (PTG) (%)	Community whole (all ages) (PTG) (%)	Community working age (PTG) (%)	Community recruit age (PTG) (%)	Community whole (all ages) (PTG) (%)	Community working age (PTG) (%)	Community recruit age (PTG) (%)
Midwestern Region	Elkhart Police (IN)			Muncie Police (IN)			Springfield Police (OH)
Male	194	195	190	202	198	194	190	189	197
Female	11	11	11	9	9	9	19	19	19
Minority	13	14	13	21	23	22	39	42	41
Minority Male	26	28	26	44	48	43	66	71	70
Minority Female	0	0	0	0	0	0	13	14	13
White	168	161	169	117	115	116	122	119	120
White Male	334	318	330	237	228	226	236	228	242
White Female	20	19	20	11	11	11	22	21	20
Western Region	Everett Police (WA)			Rialto Police (CA)			Spokane Police (WA)
Male	178	174	174	179	180	178	187	184	182
Female	18	18	18	21	21	21	18	18	18
Minority	31	34	30	53	53	53	45	48	40
Minority Male	53	58	52	94	96	93	90	97	79
Minority Female	6	7	6	11	11	11	0	0	0
White	138	132	141	513	486	562	112	110	115
White Male	247	228	242	1001	912	1059	208	200	207
White Female	24	24	26	103	101	116	21	21	23