![Sample our Social Sciences journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110822/TOC-Subject-Sample-2021-Social-Sciences.jpg"})
Abstract
Officers are not equally supportive of community policing despite its potential for improving police–citizen relationships. Research has yet to identify and explain variations in officer support for community policing with racial minorities. Using roll-call surveys with 741 officers in three departments, this project addressed two questions: Do officers differ in their support for community policing across racial groups? And, if so, why? Officers are less supportive of community policing with racial minorities and perceive greater social distance from minority groups. General support for community policing and lower perceived social distance from a minority community are linked with greater support for community policing with that group. Community policing experience is not related to support for the practice across racial groups. By understanding differences at the officer-level, departments can build support for community policing—particularly with minority communities—through reducing perceived social distance. Additionally, department-level differences highlight the importance of comparative research.
Disclosure Statement
No potential conflict of interest was reported by the author.
Acknowledgements
I would like to thank Katie Hail-Jares, Gary LaFree, Belen Lowrey-Kinberg, Ed Maguire, Natalie Todak, Joseph K. Young, and Thomas Zeitzoff for providing feedback on this project at various stages. Emma Ashooh provided valuable research assistance.
Notes
1 Path dependency refers to the notion that once a policy is set into motion, it is more difficult and sometimes impossible to change course (Schneider & Ingram, Citation2005).
2 In his work on categorical terrorism, Goodwin (Citation2006, p. 2041) describes social distance as “the weakness or absence of political alliances between revolutionaries and their presumed constituents and complicitous civilians.” In this paper, I refer to these alliances between law enforcement and community members.
3 Six departments were contacted and asked to participate: three agreed, two declined, and one did not respond.
4 At the time of data collection, 69.8% of patrol officers in Department 1, 84.4% of patrol officers in Department 2, and 71.7% of patrol officers in Department 3 were given the opportunity to participate.
5 Response rate by department: 94.9% in Department 1, 98.5% Department 2, and 97.6% in Department 3.
6 Paoline and Terrill (Citation2013) had a list of officers in each department and reached out to officers who were absent from roll-call to increase participation. Unfortunately, the police chiefs in these three departments would not give me access to their personnel information to employ this same method.
7 My survey asked officers for an honest assessment of what they do at their jobs, how they feel about it, and how they perceive various minority communities in their jurisdiction. These are sensitive topics that can increase dishonesty. To build rapport with officers and show that questioning authority is okay, the first question in the survey asks participants’ gender and gives the following response options: male, female, and other. As expected, the “other” option got a reaction out of officers in every roll-call. I responded that the research ethics board made me phrase the responses this way and made a joke about bureaucracy. By starting with a critical statement of authority, I hoped this would help participants be less suspicious of my intentions and increase honesty in responses.
8 Racial groups were chosen to reflect the population in the Washington DC metropolitan area. The correlation among these variables ranges from .79 to .91.
9 Averaging these scores generates a variable with decimal points so ordered logistic regression would not be an option. Since the variable is not normally distributed, OLS could produce biased estimates. To address these concerns, I added scores together. As robustness checks, I estimated all models using OLS and results were fundamentally unchanged. Ordered logistic regression models are reported for consistency across tables.
10 There are often differences between a person’s general and specific views across a range of topics. The correlation between general support for community policing and group-specific support ranges from .50 to .53. This demonstrates a positive relationship between general support and group-specific support for community policing, but also shows that these variables are not measuring the same thing.
11 Correlations among these variables range from −.37 to .48.
12 Cordner (Citation1995) summarized the common elements of community policing into three dimensions: philosophical, strategic, and programmatic. Skogan and Frydl (Citation2004) built on Cordner’s conceptualization to argue that community policing has four main elements: police functions, decentralization, community engagement, and problem orientation. Maguire and Wells (Citation2009) focused on the organizational element for implementation and argue that community policing has three main facets: problem solving, community engagement and partnerships, and organizational adaptation. The Community Policing Self-Assessment Tool (CP-SAT) uses these three overarching elements. Cordner (Citation2014) later split the programmatic dimension of his original conceptualization into two parts: tactical and organizations.
13 Ideally, community policing experience would be measured through direct observation. Unfortunately, it is infeasible to systematically monitor officers across departments over time. Additionally, officers may behave differently when they know they are being watched (the Hawthorne effect). Rather, survey methods are necessary to measure community policing experience for individual officers across a department. Since surveys have space constraints, I was not able to ask about all behaviors or actions that could fall under the community policing framework. I winnowed the items to measure fundamental elements of community policing practices.
14 Correlations among the dimensions of community policing experience range from .13 to .63. All models are reported using an additive index of community policing experience. Models were also estimated with each dimension of community policing included separately rather than as an additive index. The results were fundamentally unchanged and none of the dimensions of community policing had a significant impact on support for community policing across racial groups.
15 Correlations among perceived social distance from each racial group range from .31 to .62.
16 I coded the first and second items as 1 if the response suggests more social distance and −1 if the response suggests less social distance so that each question would carry weight more equivalent to that of the other measures.
17 Results are reported using average support across minority groups, but are the same when comparing support for community policing with Caucasians to each minority group individually. Results also hold across departments.
18 Support for community policing with Caucasians does not vary across departments, F(2, 705) = 1.38, p = .25. In contrast, support for community policing with minorities does vary: officers in Department 3 are the most supportive (M = 5.98, SD = 1.09), followed by Department 2 (M = 5.89, SD = 1.13), then Department 1 (M = 5.72, SD = 1.18); F(2, 704) = 3.25, p = .04.
19 Honesty in responding is a concern, particularly when asking officers about sensitive topics in the workplace. It is possible that some participants provide the same responses across all racial groups due to social desirability bias or fear of retaliation if their true views were made public. To account for this explanation, I estimated the models removing officers who “straight-lined” responses in one of two ways. I excluded participants who (a) straight-lined responses across the majority of questions about minority communities and (b) straight-lined the dependent variable. Across all models, the statistical or substantive results were unchanged. Models reported include all observations.
20 An alternative would be to estimate models with interaction effects between the department dummy variables and each predictor. The main drawback with this approach is that one department is necessarily excluded as the reference category. Estimating models by department and comparing the coefficients bypasses this concern.
21 As robustness checks, I treated the dependent variables as continuous and estimated models with OLS. I also transformed the dependent variables to improve homoscedasticity. Results were unchanged across models.
22 Models were also estimated without officer demographics included and the results are unchanged.
23 Minority officers perceive greater social distance from minority communities in general, (t(625) = −2.96, p = .002). This finding holds for perceived distance from African American (t(634) = −2.73, p = .003) and Middle Eastern (t(634) = −3.64, p = .001) communities specifically. There were no differences in experience with (t(668) = .96, p = .17) or general support for (t(711) = .90, p = .18) community policing between minority and Caucasian officers.
Additional information
Notes on contributors
Erin M. Kearns
Erin M. Kearns is an assistant professor in the Department of Criminology & Criminal Justice at the University of Alabama. Her primary research seeks to understand the relationship among law enforcement, the public, and violent organizations. Her work has been funded through multiple sources, including the National Consortium for the Study of and Responses to Terrorism (START).