Community-based proactive strategies recognize and promote the community’s active role in the crime-prevention process. They seek to define the relationship or mode of interaction between the police and the community in a way presumed to reduce crime or disorder. As we mentioned in Chapter 5, unlike the other proactive policing approaches considered in this volume, police often employ strategies for a community-based approach with an explicit hope that they will not only reduce crime but also improve people’s assessments of police performance, increase community perceptions of police legitimacy, and enhance cooperation and community engagement to secure public order and safety (Skogan, 2006b). Not surprisingly, then, one might expect to see more research on how community-based strategies affect community outcomes than on how the other three proactive approaches affect community outcomes (the subject of Chapter 5 of this report). This is indeed the case, but even here the research on the community impacts of community-based interventions has concentrated heavily on two strategies: community-oriented policing and procedural justice policing, with much less attention to the community impacts of broken windows policing. Consequently, the bulk of our discussion is skewed to the first two strategies for a community-based policing approach.
While community-oriented policing and procedural justice policing are both strategies that take a community-based approach, their places in the landscape of proactive policing are distinct. The concept of what this report
calls “community-oriented policing”1 has been central to discussions of policing for several decades, and many departments have developed various policing policies that come under the committee’s concept of a community-oriented policing strategy. Consequently, there is a large prior literature on evaluations of community-oriented policing. In contrast, the concepts informing procedural justice policing are comparatively new to the field of proactive policing, at least as policy-level interventions. The broader concept of procedural justice developed within the field of social psychology, in theory-driven studies exploring why people trust authorities, view them as legitimate and entitled to be obeyed, and consequently defer to their authority. Research has subsequently studied procedural justice and perceived legitimacy in work organizations and with court procedures. However, these concepts have only recently been directly applied to policing.
As noted in Chapter 2, community-oriented policing (also called community policing) is widely acknowledged to have many meanings, sometimes as a set of specific tactics, sometimes a set of program-level interventions, and sometimes a general philosophy of how police should relate to the community (Cordner, 2014). Despite its longevity as a reform (at least three decades), as noted in Chapter 2, there is still considerable variation in how community-oriented policing is defined. We follow Gill and colleagues (2014, p. 405) in requiring that, to qualify in this review as community-oriented policing, an intervention must include “some type of consultation or collaboration between the police and local citizens for the purpose of defining, prioritizing, and/or solving problems.” As is the case with other proactive policing strategies, practices typical of a community-oriented strategy are often implemented in combination with practices and tactics typical of other strategies, including strategies that focus on a different policing approach. For instance, some community-oriented policing interventions include practices typical of problem-oriented policing, broken windows policing, hot spots policing, or focused deterrence. As noted many times already in this report, this hybrid character of real-world interventions makes it more difficult to draw conclusions from evaluations of these hybrid interventions regarding the impacts of community-oriented policing, as a distinctive strategy, on community outcomes.
1 The research literature has often used the term “community policing” for what we mean here by community-oriented policing. We have applied our term in reporting on the literature where the topic addressed by the author(s) seemed closer to our strategy of community-oriented policing, as presented in Chapter 2, than to the broader concept of any community-based approach to proactive policing.
As discussed in Chapter 2, it is well established that community-oriented policing became very popular among American police leaders in the 1990s. What is not so well acknowledged is the variable character of community-oriented policing that exists among these police agencies. For the purpose of assessing the community impact of community-oriented policing, it is a significant limitation that the research literature often lacks clear distinctions not only among the different varieties of community-oriented policing but also with respect to their scope and intensity (Cordner, 2014). There currently is no metric for making comparisons across different community-oriented policing programs; therefore, it is difficult to know how appropriate it is to compare results across impact studies.
One indication of the challenges presented in summarizing the effects of community-oriented policing is to consider the difficulties in generalizing about it from the available empirical research. A useful tool in this regard is the data provided in an appendix of a systematic review of 45 studies of the impact of community-oriented policing (Gill et al., 2014). This appendix provides a brief description of each of the community-oriented policing interventions described in the study reviewed. Table 6-1 shows the frequency of those that involve community engagement or collaboration. As is apparent, there are a wide variety of practices used in these interventions, ranging from foot patrols to collaboration with community groups and community newsletters. Clearly this range of practices will influence the nature and intensity of community-oriented policing’s impact on community outcomes. And these 45 studies did not attempt to determine the independent contribution of different program elements in the community-oriented policing interventions they evaluated.
Furthermore, the outcome measures employed in studies are inconsistent, making it even more difficult to draw direct comparisons (Gill et al., 2014, p. 422). To this point, the committee adds that these inconsistencies arise in how given measures are conceptualized, operationalized, or interpreted. An example of this is given in the classification of “legitimacy” outcome measures (measures of what this report calls “perceived legitimacy”). A single research project by Tuffin, Morris, and Poole (2006) accounted for 6 of the 10 comparisons we examined on perceived-legitimacy outcomes. The actual survey question (of residents) used for this item was, “Taking everything into account how good a job do you think the police in your local area are doing?” (Tuffin, Morris, and Poole, 2006, p. 51). Excellent or good responses were interpreted as showing confidence in the police. However, it is not clear why that item has more in common with other outcome indicators used by the meta-analysis to assess perceived legitimacy (e.g., “police are fair,” “trust in police,” “treating people politely”) (Gill et al., 2014, p. 417, Fig. 7) than it does with some of the indicators used for assessing the community outcome of “citizen satisfaction”: “good job
|Intervention Number||Intervention||Number of Studies|
|Decentralization of Police Organization|
|2||Permanent beat assignment||10|
|4||Special community-policing unit created||1|
|5||Change management philosophy||1|
|7||Resident contact (one-on-one): e.g., door-to-door surveys||9|
|8||Collaboration with community groups (unspecified)||7|
|9||Beat/neighborhood organization meetings||18|
|10||Crime-prevention education for citizens||1|
|14||Community relations training for police||1|
|15||Community rallies, unspecified mobilization||4|
|16||Increase positive police–citizen contacts (e.g., on the street, recreation programs)||2|
|17||Problem solving (general)||15|
|19||Broken windows (clean up physical and social disorder)||3|
|20||Household security inspections||1|
|21||Environmental change for crime control, improve neighborhood infrastructure||1|
|22||Enforcement-oriented interventions (crack downs, hot spot patrol)||1|
NOTE: Data from Gill et al. (2014, App. A). The number of defined interventions per study varied from 1 to 4.
to prevent crime,” “evaluation of police,” or “quality of police service.” The last of these, “quality of police service,” is a scale comprising ratings of items that seem good candidates for perceived legitimacy, not satisfaction: police politeness, helpfulness, and perceived fairness (Gill et al., 2014, p. 416, Fig. 6).
Another source of variation across studies to which insufficient attention has been paid is the way in which the targeted community population is defined (Gill et al., 2006, p. 422). Most evaluations of community-oriented policing tend to aggregate “community” as a general population of residents, and this undoubtedly masks what are potentially striking differences. “Community” is most often operationalized as people who live in proximate geographic areas, typically within the boundaries of a police beat or a neighborhood. Residents of a neighborhood presumably have a stake in how their neighborhood is policed, but that stake is not necessarily uniform. The context of how people relate to police—their role—can vary profoundly. Victims and suspected offenders can be expected to hold different concerns or priorities about what they want police to do and accomplish. Those who own and work in businesses may have different priorities from those who reside near them. Those who frequent parks and recreational facilities will have a different framework for evaluating police than those who live near those facilities. And people of different ethnicity may have different histories with the police that produce different evaluation frameworks. Regardless of their role in a particular encounter with the police, people who have frequently been the object of enforcement activity possess a different set of sensitivities from others whose experiences have been as service recipients (see, for example, Brunson and Weitzer, 2007).
Much of the original impetus for community-oriented policing came from groups of citizens who were disgruntled because they felt abused as objects of enforcement or underserved as victims of crime (Kelling and Moore, 1988), and there is currently much interest in community-oriented policing as a way to deal with both of these groups who are more likely to experience contact with the police or to desire their assistance (President’s Task Force on 21st Century Policing, 2015; App. A of this volume). Yet the extant research on community-oriented policing typically fails to distinguish these “high-intensity” populations and thus offers little to enlighten policy and practice for the parts of society that were key to animating the movement for change.
A notable exception to the tendency to ignore high-intensity subgroups within a studied community is the evaluation of Chicago’s community policing program across three different racial groups (Skogan, 2006b). Comparing trends in confidence in the police2 across Blacks, Latinos, and Whites between 1994 and 2003, the researcher noted that improvements were
2 In this study, “confidence” was measured as a composite of three scales, which were constructed from neighborhood resident surveys: perceived demeanor of officers, responsiveness to neighborhood concerns, and perceived effectiveness in controlling crime/disorder and helping victims.
observed across all measures for each racial group but added a caveat that highlights the importance of disaggregating “community” into subgroups:
In the end, however, the contrast between the general optimism of whites and the still-widespread pessimism of African Americans was almost as large in 2003 as it had been in 1994 when CAPS was still in development. Things got better between African Americans and the police, but confidence had also grown among whites, keeping the gap just as wide. “The glass was only half full” when it came to healing the breech between police and the public, for while Chicagoans were more confident in the police, they were still deeply divided by race. (Skogan, 2006b, p. 322)
Finally, most of the studies of community-oriented policing that focus on community outcomes do relatively little to establish the strength of the causal connection between policy and practice. They tend to test the extent to which either policy or practice leads directly to each of the types of outcomes depicted in Figure 5-1 (see Chapter 5 of this report) as stages 3, 4, or 5. The correlations and/or causal links between stages 3 (community evaluations), 4 (community orientations), and 5 (community behavior) have not been a topic of systematic exploration. This presents numerous challenges for testing the validity of efforts to use community-oriented policing to promote desirable community outcomes.
We begin with these caveats in order to emphasize the difficulty in drawing conclusions regarding the effects on community outcomes of community-oriented policing. We find this surprising in some sense, given the very strong focus of community-oriented policing on changing the relationship between police and the public (Kelling and Moore, 1988; National Research Council, 2004, pp. 85–90; Skogan, 2006c; Skolnick and Bayley, 1986). Despite this focus, the extant research literature makes it difficult to draw very strong conclusions about precisely those outcomes that community policing was meant to influence. The following sections outline what these studies show and what they are unable to show.3
Studies of the impact of community-oriented policing on community evaluations of specific aspects of police performance have focused on citizen perceptions of disorder (e.g., severity of drug problems, social disorder), citizen fear of crime, and citizen satisfaction with police performance). Gill and colleagues (2014) provided a detailed comparison of these effects
with 16 independent comparisons of perceived disorder, 18 comparisons of fear of crime, and 23 comparisons on citizen satisfaction, but fewer of these comparisons had sufficient information to calculate odds ratios (11 disorder, 10 fear, and 17 satisfaction outcomes). The meta-analysis produced only one statistically significant effect—citizen satisfaction increased—although all three outcomes showed small average effects in the expected positive direction. Satisfaction with police was characterized as a “moderate” effect (odds ratio of 1.37; Gill et al., 2014, p. 415). This effect qualifies as “small” according to some standard rules of thumb (see, e.g., Lipsey and Wilson, 2001), but many communities and their leaders might consider a 37 percent improvement in the odds of citizen satisfaction to be substantial. While perhaps insufficient to change very negative evaluations to very positive ones, it could arguably yield a discernible difference in a community.
Across individual studies in all three types of community-evaluation indicators, effect sizes were in the small range. Similarly, Skogan and Hartnett (1997, p. 210) concluded regarding Chicago’s community-policing efforts, “To be sure, the successes wrought by the program were not overly dramatic.” Returning to the full range of 17 evaluations of citizen satisfaction in the meta-analysis by Gill and colleagues (2014), very few (just two) yielded a small effect in a negative direction (odds ratios of 0.827 and 0.479), neither statistically significant. This pattern of infrequent backfire effects was repeated with the other community outcomes assessed in the meta-analysis.
The authors concluded that there was “robust evidence that community policing increases citizen satisfaction with the police” (Gill et al., 2014, p. 418), and “no evidence that community policing decreases citizens’ fear of crime” (p. 419). Of course, the potential synergy between program elements is not captured by this simple analysis, so the committee also considered a comparison of programs that had all three elements of community policing present in “strong” form: organizational decentralization (beat integrity), community engagement (regular community meetings, foot/bike patrol, or positive police–citizen contact), and problem solving. Six of the 17 comparisons had all three elements but showed only a small and not statistically significant differences from those that did not. Of course, the small number of cases for comparison makes this exercise tenuous, so the available evidence provides no guidance about how best to proceed with the particular policies and practices that will promote citizen satisfaction most effectively.
The Gill and colleagues (2014) meta-analysis included 10 independent comparisons of the effect of community-oriented policing on perceived legitimacy. The most frequent measure of perceived legitimacy was confidence in the police (six comparisons). Other indicators included perceived “trust in police,” “procedural justice,” “treating people politely,” and “police fairness.” On average, the odds that people living in areas where the local police had a community-oriented policing policy viewed those police as legitimate were about 1.28 times the odds for someone living in an area where local police had no such plan. This difference was marginally statistically significant (p = .077) (Gill et al., 2014, pp. 415–416).
A noteworthy feature of the sample of study comparisons in this meta-analysis is the large portion of comparisons that came from the same project. The evaluation by Tuffin, Morris, and Poole (2006) of the National Reassurance Policing Program (NRPP) in the United Kingdom accounted for 6 of the 10 comparisons. One advantage of this common origin is ease of comparability of design and measures across the six sites, which reduces the risk of variability in effects due to evaluation methodology differences in different studies. In this NRPP evaluation, there was some variability in effects across sites. Four showed stronger effect sizes (odds ratios of 1.66–3.34), and two showed much weaker changes (close to null effects). The evaluation report attributed differences in program performance to variation across sites in implementation, not to the socioeconomic characteristics of the sites (Tuffin, Morris, and Poole, 2006, pp. 88–90).
As with evaluations of police performance, the meta-analysis revealed only 2 of 10 studies showing a backfire effect on perceived police legitimacy. The most striking of these was a study of El Centro, California, which focused its intervention on a predominantly Mexican area of the city (Sabath and Carter, 2000).4 Although the treatment district showed statistically significant improvements in citizens’ familiarity with the police and perceptions of crime-control effectiveness, it showed no gains in trust toward the police, while the comparison district did show a statistically significant increase in trust.
4 The intervention included establishing a community center with a police substation to improve police-community relations, youth programming, permanent beat assignment of officers, and knock-and-talk visits using bilingual officers. The evaluation used a two-wave panel survey design with a matched comparison group and compared approximately 150 households in each of the treatment and comparison districts. The odds ratio calculated by the meta-analysis for the intervention’s effect was 0.440 and was statistically significant (Gill et al., 2014, p. 417).
Do citizens behave differently as a consequence of being exposed to a community-oriented policing intervention? Chapter 4 speaks to the impact of community-oriented policing on criminal and disorderly behavior. Our concern here is with two types of related behaviors: whether citizens are willing to cooperate with the police, and, as noted in Chapter 5, what Sampson, Raudenbush, and Earls (1997) termed “collective efficacy,” which refers to the degree to which people who live in communities trust their neighbors and are willing to intervene in community affairs. Both types of behavior speak to the ability of communities to enhance informal social controls either through alerting the police to community problems or working together directly to intervene in those problems.
Many expect that community-oriented policing should bring police and citizens closer together in common cause and should strengthen communication among various community groups as well as between police and the public. It should invest residents with the necessary skills, resources, and sense of empowerment to mobilize against neighborhood problems (Renauer, 2007; Sargeant, Wickes, and Mazerolle, 2013; Slocum et al., 2010; Velez, 2001). Much of the available research on policing precursors of collective efficacy focuses on the degree of police crime-control effectiveness or perceived legitimacy (reviewed below in the section on procedural justice policing). Research seeking to test the relationships, either associational or causal, between community-oriented policing and collective efficacy is limited (Sargeant, Wickes, and Mazerolle, 2013). Scott (2002) found in 77 Indianapolis neighborhoods that greater resident access to the police was associated with higher levels of social capital (not the same as collective efficacy, but sharing a concern for acting on behalf of community interests). However, other key measures of community policing failed to display a statistically significant association with social capital (e.g., frequency of police involvement in community events and activities). Renauer (2007) found evidence to support a backfire effect; increased police presence at community meetings was associated with less informal social control in 81 Portland, Oregon, neighborhoods. He speculated that low–socioeconomic status neighborhoods attracted more police attention. Sargeant, Wickes, and Mazerolle (2013), using qualitative interviews of key informants in two Brisbane suburbs, did not find the expected association between community-oriented policing and each community’s collective efficacy. In the suburb with low collective efficacy before the intervention and a high immigrant population, police efforts to reach out to residents did not yield the expected gains because those efforts were not perceived as legitimate (i.e., were not seen as fair or effective). Nor did residents possess
the knowledge and skills needed to act effectively to mobilize organizations on their behalf. In the wealthier suburb, which had high collective efficacy prior to the intervention, the relative absence of problems disinclined police to invest much community-oriented policing effort there, nor were residents particularly desirous of such police interventions.
The strongest evaluation of community-oriented policing’s impact on collective efficacy is the assessment of the NRPP in the United Kingdom (Tuffin, Morris, and Poole, 2006). This policing program had several elements: focusing policing activity on those “signal” crimes expected to have a disproportionate impact on public feelings of safety, community involvement in identifying priorities for targeting interventions and participation in the interventions, and making locally known authorities and police officers readily accessible to the community. This pre-post, matched comparison group design used a two-wave panel (1 year apart) to study program effects in one area for each of six different UK police forces.5 The study found evidence of desired changes attributable to the NRPP for many of the outcomes measured (decreases in crime and in perceptions of antisocial behavior, increases in feelings of safety and in confidence in the police), but virtually absent was a statistically significant change relative to comparison sites in measures of social cohesion, feeling trust in other members of the community, collective efficacy, or involvement in voluntary/community activity. Of the five outcome indicators used, only one (trust in the community) had a statistically significant (but modest) positive increase when pooled across all sites,6 but there were no statistically significant changes in measures of willingness of neighbors to intervene or of voluntary activity. At the individual site level, the difference in perceived legitimacy across treated and untreated groups was statistically significant in only 3 of the 30 tests. The evaluators speculated that the development of social capacity may take longer than changing community perceptions of conditions in their neighborhood and feelings about the police.
To summarize, most of this small number of studies on community-oriented policing’s record in promoting collective efficacy are cross sectional in nature. Given their designs, these studies can only establish whether there is the expected statistical relationship; they cannot distinguish how much of any association found is due to the effects of community policing on collective efficacy and how much is due to the effects of collective efficacy on community policing (the issue of potential reverse causality). Nor can they rule out the possibility of third common causes (confounders). On the
5 The number of respondents available from both waves varied between 170 and 205 for each community outcome assessed.
other hand, these studies can provide credible information about people’s feelings about their experiences, as well as suggestions about how those feelings are associated with their views about the police. The exception to this limitation is the evaluation of the NRPP, but that study found only small differences that were not statistically significant, using conventional measures of confidence.
There is a significant body of research on the correlates and predictors of citizens’ crime reporting behavior, but very little empirical research that explicitly examines the causal linkage between community-oriented policing and crime reporting (Schnebly, 2008).7 One study examined the effects of police department resource commitment to community-oriented policing on the willingness of victims and third parties to report victimizations to the police or other nonpolice third parties (apartment manager, school administrator) for 2,379 assault and robbery incidents recorded by the National Crime Victimization Survey from 1997 through 1999 (Schnebly, 2008).
Controlling for other factors known to influence reporting behavior (victim and city characteristics), Schnebly found that in cities with a larger percentage of the force working in full-time community-oriented-policing assignments, third parties were more likely to report victimizations to a police official. Further, victims in cities served by police agencies with higher portions of the force working as community-oriented policing officers were more inclined to notify nonpolice third parties than to make a report to the police. Additional analyses showed that the amount of training of police recruits and of residents in community-oriented policing showed no statistically significant relationship to victimization reporting behaviors. However, the proportion of current officers who had received community-oriented policing training showed a statistically significant positive relationship to the likelihood of residents reporting their victimization. The study also examined whether community-oriented policing’s relationships with the community were conditioned by either victim or event-related characteristics and found some associations of this sort. For instance, residential instability reduced the strength of the negative relationship of full-time community-oriented policing staffing to the likelihood of police notification. The study speculated about the apparent contradictions and complexities of the findings. However, it is difficult to draw conclusions from a single study, particularly one with a number of self-acknowledged limitations. The measures of community-oriented policing staffing did not distinguish between different approaches to community-oriented policing, nor did they consider the degree to which officers who were not community-oriented-
7 We exclude from consideration here the research that examines the effects of procedural justice policing and perceived legitimacy on crime-reporting behavior, which we cover in the section below on procedural justice policing.
policing specialists engaged in community-oriented policing activities. Further, variation in the degree of community-oriented policing effort within a given city may vary tremendously from neighborhood to neighborhood, but only city-level measurement was possible. And as is true with the literature on community-oriented policing and collective efficacy, this study was cross sectional, measuring all variables during the same time period.
In summary, the available literature on the relationship between community-oriented policing and community behavior consists predominantly of studies of collective efficacy and crime reporting. Three aspects of this literature are important: the number of studies is small, the findings across them are mixed, and there are many methodological limitations, particularly with interpreting study results as evidence for causal connections. These aspects make it inappropriate to draw conclusions about the effects of community-oriented policing on citizen cooperative behavior.
In addition to enhancing perceived police legitimacy, an important goal of community-oriented policing is to build, improve, or sustain communities. Such transformations rarely take place in the span of months or even a few years. Yet most studies of community-oriented policing’s effects (and associations with outcomes) use a time frame that is short term, generally a year or less. The sources of such temporal bias are many, but three are particularly powerful: (1) Research funding cycles tend to support short-term projects. (2) American police organizations experience a high rate of turnover at the top, which makes for greater program instability as new chiefs tend to be “new brooms,” sweeping out their predecessors’ innovations to make room for their own (Mastrofski, 2015). (3) It is difficult to sustain experimental and even quasi-experimental research protocols for extended time periods.
How long does it take for a policing innovation to register an effect and sustain it? One might expect that the longer an intervention has been operating, the greater its prospects for showing an effect. For example, it has been suggested that the changes to organizational structure that are part of community-oriented policing (e.g., decentralization and reduced hierarchy and specialization) may simply take many years to accomplish and to yield organizational transformation (Mastrofski and Willis, 2010, p. 71). Alternatively, some interventions may realize their successes early, and others may even decline over the long run because they are insufficiently flexible to respond to changing conditions.
One of the few exceptions to the bias toward short-term research is the decade-long evaluation of community-oriented policing in Chicago (Skogan,
2006b, Chapter 10).8 Between 1994 and 2003, fear of crime declined under this community-oriented policing intervention (at the greatest rate for Blacks and at the lowest rate for Latinos). During that same period, perceptions of disorder declined significantly for Blacks while increasing significantly for Latinos. And trends in evaluations of police “confidence” (demeanor, responsiveness, and performance) increased for all three racial groups. Interestingly, the generally increasing year-to-year level of these indicators (combined into a single quality-of-service index) plateaued for all ethnic groups after about 6 years, with the group scoring the lowest percentage of positive responses (Blacks) at 40 percent and the highest group (Whites) scoring more than 60 percent, with Latinos in the middle at nearly 50 percent (Skogan, 2006b, p. 280). Unfortunately, because community-oriented policing was implemented citywide for most of that time period, there were no comparison groups to help rule out the effects of other influences.9
Finally, it is worth noting that the study of long-term community effects calls for a consideration of the long-term history of police “treatments.” Neighborhoods with a long history of receiving one or more elements of community-oriented policing may respond differently from those with little or no such experience, and the response over time may vary with the duration of the treatment. Whether neighborhoods that have experienced several years of positive police outreach are more responsive to a new community-oriented policing program than those for whom there is no history of such outreach is an open question. Neighborhoods with a history of fraught relations with the police may take longer to respond positively than neighborhoods with a more positive history.
Because community-oriented policing requires interaction between the police and the community for it to achieve effective outcomes, the environment in which a community-oriented policing intervention is delivered is particularly important for its success. This means that one should approach generalizing about the effects of community-oriented policing with a healthy respect for the possibility that it will depend upon the character of the community where it is employed (Reiss, 1992; Klinger, 2004). At what sorts of
8 Another study that offered a slightly longer-term evaluation of a community-oriented policing program was a follow-up to the UK’s NRPP, which added a 2nd-year evaluation to the original 1-year study (Quinton and Morris, 2008). The follow-up found a continuation in the second year of the desirable impacts observed in the original evaluation by Tuffin, Morris, and Poole (2006).
9 It is difficult to determine whether the plateauing pattern was due to program features or how they were implemented, other features of the organization (e.g., the growth in Compstat’s potentially antagonistic influence), or a variety of external factors.
jurisdictions have community-oriented policing studies been conducted? It is instructive to consider the sample produced by the systematic review by Gill and colleagues (2014), the review used above for its outcome showing that the strongest outcome association with community-oriented policing interventions was citizen satisfaction. Of the 17 comparisons, 6 were made in UK areas of large size or served by large police departments, at least by American standards (e.g., Leicester, Surrey, Bexley, Thames Valley); 5 were conducted in Chicago, 2 in Australia, 1 in Newark, 1 in Houston, 1 in Madison, and 1 in a small California city. While in some respects this represents a diverse sample, it clearly ignores or grossly underrepresents rural, small town, and suburban agencies in the United States. The strong representation of the United Kingdom and Chicago in particular make it hazardous to formulate a basis for generalizing results broadly.
It is also appropriate to reiterate the point that studies of community-oriented policing mostly focus on effects at a level below the jurisdiction (police beat, neighborhood, or district/borough). The prospects of jurisdictionwide effects remain virtually unexamined.
Summary. The available empirical research on community-oriented policing’s community effects focuses on citizen perceptions of police performance (in terms of what they do and the consequences for community disorder), satisfaction with police, and perceptions of police legitimacy. There is considerable variability of findings within and between types of community outcome measures. Overall, community-oriented policing programs show a tendency to increase citizen satisfaction and have positive but weaker effects on perceptions of police legitimacy. Nonetheless, there are a number of limitations in the extant research that limit the committee’s capacity to draw firm conclusions about what this means.
As we noted in Chapter 2, the committee considers broken windows policing to be a strategy for a community-based approach. Our reasoning is that the mechanism that underlies the original formulation of the community-based approach is rooted in making changes in the community. Such changes are driven in part by changes in policing, but it remains the case that the long-term goal of broken windows policing is to enhance the ability of the community to exercise informal social controls presumed to play a central role in the nature and extent of community order and safety (Weisburd et al., 2015; Wilson and Kelling, 1982).
There are two specific outcomes relevant to our discussion that are predicted by the broken windows logic model. The first is that fear of crime is a key causal factor in increasing crime rates. A key purpose of broken
windows policing is to reduce fear of crime, which should lead in the long run to stronger informal social controls in urban communities. Wilson and Kelling (1982, p. 31) noted in discussing the Newark Fear of Crime Experiment:
First, outside observers should not assume that they know how much of the anxiety now endemic in many big-city neighborhoods stems from a fear of ‘‘real’’ crime and how much from a sense that the street is disorderly, a source of distasteful, worrisome encounters. The people of Newark, to judge from their behavior and their remarks to interviewers, apparently assign a high value to public order, and feel relieved and reassured when the police help them maintain that order.
The second outcome is similar to that which was discussed above in regard to community-oriented policing. Broken windows policing would be expected to increase the degree to which citizens are willing to intervene in doing something about community problems. For Wilson and Kelling (1982), social and physical disorder are key factors in the decline of communities. As discussed in Chapter 2, broken windows policing, with its focus on reducing disorder, is expected to reverse the decline of collective efficacy in communities, thereby preventing a breakdown in community social controls.
In assessing the impacts of broken windows policing, the committee drew heavily from a recent systematic review conducted by Weisburd and colleagues (2015). They examined studies that used either a control/comparison group design (experimental or quasi-experimental) or a before-after assessment of outcomes, and each study had to report impacts on fear of crime and/or informal social control. Overall, they identified just six studies that examined the impact of disorder policing on fear or collective efficacy/ informal social control. One of the studies was a randomized experiment. Four studies used quasi-experimental designs with comparison groups, and one study used a before-after design. All six examined impacts on fear, while only one examined impacts on informal social control (defined as collective efficacy). The committee’s review did not identify any additional studies.
The earliest studies that examined the impact of disorder policing on fear were a pair of Police Foundation studies by Pate and colleagues (1985b, 1985c). The first examined a police program in Newark, New Jersey, that aimed to reduce fear of crime by reducing the signs of crime (Pate et al., 1985a). Findings were mixed across different measures, but as a whole, the
authors concluded that the program was ineffective in reducing fear in the targeted area relative to the comparison area. The second Newark study involved an order-maintenance program as part of the police intervention, and it found that fear of property crime was significantly reduced, relative to the comparison area (Pate et al., 1985b).
Research by McGarrell, Giacomazzi, and Thurman (1999) examined the impact of a community policing program that involved elements of broken windows policing (improving physical conditions, targeting drug and social disorder problems) in the area surrounding a public housing facility. Fear of crime was significantly reduced relative to the comparison area, even though there were no statistically significant reductions in crime. On the other hand, a pre-post case study of a partnership policing program in two villages in Wales found no statistically significant impacts on fear (Rogers, 2002).
Finally, two more recent and related studies also produced mixed findings. (These findings are also reviewed in Chapter 5, as they both are also hot spots policing initiatives.) Using data from the Police Foundation Displacement and Diffusion study conducted in Jersey City, New Jersey, Weisburd and colleagues (2006b) and Hinkle and Weisburd (2008) found that aggressive police crackdowns on social and physical disorder appeared to increase fear of crime in the target areas relative to the surrounding catchment areas that did not receive any extra police attention. However, a randomized experimental evaluation of the impacts of broken windows policing in three cities in California, designed in part as a follow-up to the Police Foundation study, found that a 6-month police intervention that focused on reducing social and physical disorder but encouraged police use of discretion (see Kelling, 1999) had no impact on fear of crime or collective efficacy (Weisburd et al., 2011). An important point is that this study is the only one identified by the committee that evaluated the impact of broken windows policing on any measure of informal social control. The authors suggested that the differing findings across these two studies were due to the differing nature of the interventions. While both police programs were consistent with the broken windows strategy of targeting disorder, the Jersey City intervention involved a very aggressive crackdown on disorder that included sweeps, a violent offender removal program, and intensive enforcement aimed at street-level drug sales and use and at prostitution. The intervention in California used a less heavy-handed approach to broken windows policing. It emphasized rapid repair of physical disorder and a discretionary approach to handling social disorder through mediation and warnings.
In this regard, recall also the differing findings in the two studies by Pate and colleagues (1985b, 1985c) discussed above. The intervention that attempted to reduce fear by cleaning up disorder (reducing the signs of
crime) showed no impact on fear, while the policing program that had a disorder abatement component was found to reduce fear by a statistically significant amount. Thus, it may be that how the police design and deliver a disorder-focused program may affect the extent to which the mechanisms of broken windows policing are confirmed.
Weisburd and colleagues (2015) provided in their meta-analysis a quantitative summary of the evidence of these disorder policing programs on fear of crime. (They did not provide a quantitative summary regarding collective efficacy because only one study reported on these outcomes.) Using a random effects model because of the variability of treatments and outcomes, they found a slightly negative, albeit statistically not significant, impact. This suggests, if anything, a very slight backfire effect in the samples examined, but the authors concluded that the data do not, in general, support or refute any clear impact. We think their conclusion is reasonable, given the small number of studies available.
All in all, the committee simply does not have enough evidence to draw a solid conclusion regarding the impacts of broken windows policing.
Summary. The committee is not able to draw a conclusion regarding the impacts of broken windows policing on fear of crime or on collective efficacy. This is due in part to the surprisingly small number of studies that examine the community outcomes of broken windows policing and in part to the mixed effects observed. The committee notes how little attention has been paid to community processes in this area, given the emphasis on enhancing community social controls in the original logic model for this strategy as proposed by Wilson and Kelling (1982). The importance of informal social controls in their logic model would imply that collecting data on collective efficacy is critical. But we found only one study that attempted to assess collective efficacy. With regard to fear of crime as an outcome of interest to the model, there are more studies, but they differ considerably in the observed change in fear of crime, based on the policing tactics carried out in the intervention under study. Overall, it appears that softer approaches that focus on community engagement and utilization of police discretion are more effective in reducing fear. Such approaches are also more consistent with Kelling’s suggestions for how police should address disorder (Kelling, 1999; Kelling and Coles, 1996).
For a variety of reasons the question of perceived legitimacy has become more central to proactive policing in the United States over the past several years. Perceived legitimacy may be defined as the belief that the police are entitled to exercise authority within the community and that
as a consequence their directives ought to be accepted and receive deference. Recent events involving police shootings in different U.S. communities and subsequent public protests have led national police leaders to be concerned about the issue of public trust and to seek information about how to increase trust. An example of that effort is the recent report of the President’s Task Force on 21st Century Policing (2015), which made perceived legitimacy a core theme in its discussion of policing. Because of this concern about their legitimacy, police departments have increasingly developed proactive efforts to engage in policies and practices that promote and sustain their perceived legitimacy among the people in the communities they police. As we detail below, these efforts have typically focused on enhancing procedural justice in police-citizen encounters. Our main question is whether proactive policing programs based on a procedural justice model improve attitudes toward the police and cooperation with the police.
Perceptions of police legitimacy are subjective and must be studied by interviewing people and discerning their orientations toward the police. Hence, by definition, efforts to understand perceived legitimacy need to focus on people’s perceptions about the police and their subjective reactions to police actions. The model outlined in Figure 5-1 in Chapter 5 of this report presents a logic flow that incorporates these subjective responses to policing. It moves from police policies and practices to what is actually going on in the community (police behavior) to the subjective evaluations and orientations of the people within that community (police legality/perceived fairness; popular legitimacy). To the degree that this logic model is accepted as a causal model, it suggests that those perceptions, in turn, feed into law-related behaviors in the community (cooperation, engagement).
One key question is whether changes in police behavior do in fact change the law-related behavior of people in the community. A second question is why that change occurs, which is an issue of mediating mechanisms. The presumed mechanism in procedural justice models is that outlined in Figure 5-1 (perceived procedural justice shapes perceptions of police legitimacy). While some of the connections outlined in that model have been tested in prior studies that have been based upon the assumption that the logic model presented is a causal model, there has been no single study that tests this entire model. Nor have there been efforts to explore issues of bi-causality. In a similar case, Chapter 4 of this report outlines research that associates hot spots policing with crime rate changes. The presumed mediating (causal) mechanism in that case is deterrence. However, as is the case here, there are no studies that directly test whether hot spots policing changes the crime rate because it changes people’s perceptions about the risk of being caught. It could be the case that hot spots policing changes the popular legitimacy of the police. In other words, in both cases there is indirect evidence to support the presumed causal connection in the underly-
ing logic model, but in neither case has there been a direct test of that causal mechanism in a proactive policing intervention. In part, this lacuna reflects the inherent difficulties of testing mediating mechanisms.
One important aspect of this overall logic model is the linkage between evaluations, orientations, and behaviors—that is, the aspect of the model that begins with people’s subjective evaluations of the police and flows to their behaviors. This element in the logic model reflects the fact that perceived legitimacy of policing represents people’s evaluations and orientations, rather than objective realities. Therefore, it must be studied through interviews with members of the community.
Within the psychological literature on the antecedents of perceived legitimacy, a number of studies suggest that perceptions of the procedural justice of police actions are strongly related to perceived legitimacy.10 Procedural justice in policing refers to an interrelated cluster of evaluations of different aspects of the way police officers behave when dealing with the public. These non-experimental studies support a logic model that says that when people deal with authorities, their evaluations of the perceived fairness of the procedures through which authority is exercised influence their perceptions of police legitimacy more strongly than does the perceived outcome of the encounter (Tyler, 2006; Tyler, Fagan, and Geller, 2014; Tyler and Jackson, 2014). Similarly, when people are making overall assessments of the legitimacy of a criminal justice institution in their community, they appear to focus on how members of that institution generally deal with the public (Sunshine and Tyler, 2003; Tyler, 2006; Tyler, Fagan, and Geller, 2014; Tyler and Jackson, 2014).
The psychological literature on perceived procedural justice has identified four elements of experience that are linked to whether people evaluate institutions as being procedurally just. Those dimensions are not derived from prescriptive norms identified and defended by legal scholars and political philosophers. Rather, they have been drawn from research on the criteria that community members themselves use to rate their experiences (Tyler, 1988). Studies suggest that there is substantial agreement across race, gender, and income levels in the criteria that define a fair procedure
10Abuwala and Farole (2008); Bradford (2011); Elliott, Thomas, and Ogloff (2011); Farole (2007); Hasisi and Weisburd (2011); Hinds (2007); Hinds and Murphy (2007); Jonathan-Zamir and Weisburd (2013); Kitzmann and Emery (1993); Mazerolle et al. (2013b); Myhill and Bradford (2012); Tor, Gazal-Ayal, and Garcia (2010); Tyler (2006); Tyler, Casper, and Fisher (1989); Tyler and Fagan (2008); Wemmers (1996).
Two of the criteria shown in Box 6-1 are linked to how police officers are perceived to make decisions: (1) whether they provide opportunities for voice, allowing members of the public to state their perspective or tell their side of the story before decisions are made and (2) whether they make decisions in ways that people regard as neutral, rule-based, consistent, and
without bias. Two other criteria are linked to how the police are viewed as treating people: (1) whether they treat people with the dignity, courtesy, and respect that they deserve as human beings and as members of the community and (2) whether people believe that their motives are trustworthy and benevolent: that is, that the police are sincerely trying to do what is good for the people in the community. The model suggests that perceived trustworthiness is the key to community acceptance of discretionary decisions.
The key to understanding this model is that the criteria focus on how people experience policing, that is, whether they feel they have voice, whether they think the procedures are neutral, whether they feel respected, and whether they infer that the police are trustworthy. The underlying argument of procedural justice is that the way people perceive these features of police action shapes whether people do or do not judge the police to be legitimate.
Procedural justice as defined by these four criteria has been typically assessed in one or both of two ways. The first is to ask people how fairly “decisions were made” or how “they were treated.” The second is to ask about the four aspects of procedural justice that emerge from studies of the meaning of procedural justice (Tyler, 1988). When studies assess subjective voice, neutrality, respect, and trust, they typically find that these dimensions are highly correlated and that all four dimensions correlate strongly with evaluations of overall justice in decision making and treatment (Tyler, 1988; Tyler and Fagan, 2008; Worden and McLean, 2014).
These findings suggest that it is possible to view perceived procedural justice as an overall concept by asking people questions such as “were decisions made fairly” and/or “were you treated fairly”? It is equally possible to distinguish four component dimensions contributing to it. Empirical studies indicate that people distinguish more strongly among these four dimensions when they are evaluating their personal experiences than when they are making ratings of general police behavior in their community (Tyler, 2006).
In addition to perceptions of police treatment along the four dimensions that contribute to perceived legitimacy, researchers have also observed and coded officer conduct to determine how officer actions relate to those perceptions. That is, rather than relying upon a research participant’s personal perceptions and judgments about how the police treated her, researchers can construct a protocol for observing and classifying officer behavior that conforms to the definition of procedural justice, such as behavior showing those features listed in Box 6-1. Such a protocol requires sufficiently clear and detailed instructions to create reliable measures of officer conduct that trained third-party observers can replicate reliably (and in that sense, objectively) from situation to situation and across observers. Using this approach Worden and McLean (2014) coded officer conduct in the areas predicted to influence perceived procedural justice that fall into the category of “police practices” in the logic model portrayed in Figure 5-1.
Some type of coding of officer behaviors that are distinct from the subjective evaluations of either the people involved or the officers involved is essential for translating the concept of perceived procedural justice into terms that police officers can use to conform their behavior to the requirements of that concept. Interestingly, the relatively few studies that have explored objective measures of the components of perceived procedural justice have found that, unlike subjective measures (community members’ perceptions), the four elements portrayed in Box 6-1 are only modestly related, suggesting that they are best conceived as a formative index (Jonathan-Zamir, Mastrofski, and Moyal, 2015; Worden and McLean, 2014). Further, the only study (Worden and McLean, 2014) to have compared objective and subjective measures of officer conduct along these dimensions found that the two measures are themselves related but the magnitude of that connection varies across dimensions (see discussion below). An important emerging area of research uses the coding of police videos to establish the objective features of police behavior under different circumstances and the connection of that behavior to people’s experiences with the police (Voigt et al., 2017).
Given the relatively recent interest in the procedural justice model of proactive policing, there is, as we note below, a limited literature that examines whether perceived procedural justice is a key factor in explaining perceptions of legitimacy. At the same time, there is a large research literature that has been developing over the past century in social psychology, and more recently, in criminal justice outside policing. The committee thought it important to summarize this literature in drawing conclusions more generally about the relevance of the procedural justice model for policing.
What empirical evidence supports the procedural justice model? The theoretical underpinnings of perceived procedural justice are from social psychology (Leventhal, 1980; Thibaut and Walker, 1975), so initial evidence in this area comes from research in that field. The first research program in this area was that of John Thibaut and Laurens Walker (1975) and is summarized in their book Procedural Justice. Their research is reviewed in The Social Psychology of Procedural Justice (Lind and Tyler, 1988). The hallmark of these studies is that they are well-designed randomized controlled trials. Their context is variations in courtroom procedures, and they demonstrated that different procedures are rated differently in terms of perceived procedural justice. Procedural variations also shape a variety of types of evaluations of judicial procedures and/or authorities.
These procedural justice findings were replicated in a series of studies conducted within the Thibaut-Walker research group (Houlden et al., 1978; LaTour, 1978; Lind, Thibaut, and Walker, 1973; Lind et al., 1978; Thibaut, Walker, and Lind, 1972; Thibaut, Friedland, and Walker, 1974; Thibaut and Walker, 1975; Walker et al., 1974). The strength of these studies is their high internal validity, while their weaknesses include their laboratory context (Damaska, 1975; Hayden and Anderson, 1979), their lack of measurement of perceived legitimacy as an outcome of personal experiences, and—in the context of this report—their lack of focus on the police.
The theoretical elements in the psychological literature on procedural justice have been reviewed by Miller (2001) and MacCoun (2005). Miller identified two behavioral consequences of procedural injustice. The first is a marked disinclination to comply with authorities. The second is a diminished willingness to pursue group goals and concerns. He also noted the absence of any negative consequences of fair procedures and that a focus on using procedures for exercising police authority that are experienced by the public as fair valuably expands the universe of goals beyond compliance to include enhancing the viability of organizations.
When MacCoun (2005) conducted his review, the social psychology literature had more than 700 articles on the topic of procedural justice. MacCoun’s review suggests that, across the wide range of types of authority considered in this literature, experimental variations in actual procedural justice and differences in perceived procedural justice in different settings are both consistently found to shape compliance and cooperation with authorities. In particular, these effects were found with both experimental and correlational research designs. MacCoun (2005, p. 173) noted that “the sheer heterogeneity of tasks, domains, populations, designs, and analytic methods provides remarkable convergence and triangulation” in support of the core propositions of the procedural justice model.
The central arguments of procedural justice models have subsequently been tested in management settings, and a distinct literature on procedural justice has developed within the sub-disciplines of organizational psychology/organizational behavior. An early example is from Earley and Lind (1987), who reported on a study in which workers were randomly assigned to work under different procedures. These differences were found to influence the workers’ perceptions of fairness and performance on the job. The subsequent literature on procedural justice in work settings has expanded broadly to include variations in many aspects of work organizations and their association with a number of dependent variables, including but not limited to adherence to rules and work requirements. Some studies are conducted in ways that provide support for a causal connection between these variables, while others more appropriately support the demonstration of an association.
Cohen-Charash and Spector (2001) reviewed 190 studies (148 field studies and 42 laboratory studies) and found that variations in workplace characteristics reliably shaped perceived fairness. Procedural justice was reliably related to a number of workers’ evaluations, including satisfaction with one’s job, pay, supervisor, management, and performance appraisal procedures (Cohen-Charash and Spector, 2001, Table 7, p. 299). It was further associated with commitment to the job, normative commitment, trust in the organization, trust in one’s supervisor, and the employee’s intention to remain at or leave their job (Table 7, p. 300). Variations in the workplace characteristics associated with differences in perceived fairness were found to have an uneven relationship with required workplace behaviors. Studies found an association with workplace performance for field studies but not for lab studies. The studies consistently found an association with voluntary cooperation (organization citizenship behavior) and counterproductive work behavior (more perceived fairness leads to less shirking, sabotage, etc.). Many of these studies are experiments, and their results support the argument that these connections are not only associations but also reflect causal connections.
Colquitt and colleagues (2001) reviewed the organizational justice11 literature, and Colquitt and colleagues (2013) re-reviewed the original set of studies, as well as the subsequent literature. In the 2013 re-review, in which the authors identified 493 distinct studies, they found statistically significant overall influences of procedural justice on trust, organizational citizenship behavior, task performance, and (negatively) on counterproductive work behavior. The review found equally strong relationships for studies that focus upon particular events and those that make overall workplace evaluations. Perhaps most significantly, in terms of the model outlined, Colquitt and colleagues (2013) conducted a mediational analysis and found that the relationship between the organizational justice of the organization and relevant employee behaviors is partially mediated by “social exchange quality” (see Colquitt et al., 2013, Fig. 1, p. 217).12 Social exchange quality is quantified as an index that combines measures of trust, mutual respect, perceived management support, and commitment. In many respects, it is similar to the concept of perceived legitimacy in a management context. This type of mediating role has also been identified in more recent studies
11 Studies of procedural justice in organizational settings often use the term “organizational justice” to consider three interrelated aspects of what is here being called “procedural justice”: organizational justice, interactional justice, and informational justice.
12 The term partial mediation refers to a situation in which the direct relationship between two variables is significantly reduced when a mediator is introduced, but there is still a significant direct relationship.
of management settings (Ma, Liu, and Liu, 2014).13 Many of the studies reviewed are laboratory or field experiments that provide evidence not merely of statistically significant association but also of causal connection.
In the case of compliance, several studies illustrate the influence of the procedural justice of the climate of an organization as evaluated by employees and their compliance with rules and rulings, which is treated in this literature as an aspect of task performance. Greenberg (1994) manipulated the objective fairness of the enactment of smoking bans in a work setting and found compliance variations. Greenberg (1990) varied the objective fairness of pay changes and found an impact on employee theft. Lind and colleagues (1993) conducted a field study involving interviews with disputants and found that perceived fairness shaped the acceptance of arbitration awards. Dunford and Devine (1998) and Lind and colleagues (2000) interviewed employees and found that variations in the perceived fairness of termination procedures predicted whether terminated workers filed lawsuits. In a multinational setting, Kim and Mauborgne (1993) conducted a non-experimental survey-based study and found that rule following was linked to perceived management fairness.
In recent years there has been a series of studies of the association of procedural justice with the perceived legitimacy of the court system. Several studies deal with the courts. They find a significant association between trust and confidence in courts and their perceived procedural justice (Abuwala and Farole, 2008; Baker, 2016; Dillon and Emery, 1996; Farole, 2007; Kitzmann and Emery, 1994; Shute, Hood, and Seemungal, 2005; Tyler, 2001; Wemmers, Van der Leeden, and Steensma, 1995; Wemmers, 2013). A significant association was also found between perceived procedural justice and the willingness to accept court decisions (Baker, 2016; MacCoun et al., 1988; Tyler and Huo, 2002). Some of these studies are experiments, and their findings support an argument for the causal influence of procedural justice on these elements of perceived legitimacy in legal proceedings.
In summary, the logic model underlying the procedural justice policing strategy has been widely supported in studies varying in their focus and methodology. What is particularly striking is the convergence of these findings. Many studies, including those with experimental variations in pro-
13 This literature was also reviewed by Chang (2015), who concluded that there are statistically significant associations between organizational justice and task performance (Chang, 2015, Table 2) and between ratings of organizational justice and organizational citizenship behavior (Table 3). He suggested that both procedural justice (fair decision making) and interactional justice (fair interpersonal treatment of employees) are significantly associated with task performance and cooperative workplace behaviors (Chang, 2015, p. 34). Interestingly, this review found equally strong relationships irrespective of whether employee behavior was self-rated or assessed by independent third parties. Again, many of these studies are experiments.
cedures, suggest that it is possible to reliably create policies and practices that influence perceived procedural justice. Studies also suggest that such variations shape not only perceived procedural justice but also compliance, cooperation, and a variety of other types of organizationally relevant behaviors.
The large literature in social psychology establishes that it is possible to create settings that reliably influence perceived procedural justice (Lind and Tyler, 1988). The most replicable manipulations of procedural justice have involved variations in two procedural elements: voice (of those being acted upon) and neutrality (of those conducting the procedure). Voice manipulations typically vary whether or not people have input into legal decisions, while neutrality is manipulated through variations in whether or not the decision maker explains what facts or rules were used in making the decision.
The original Thibaut and Walker (1975) research varied court procedures between adversarial and inquisitorial, a variation which shapes whether people do or do not have (indirect) voice. Other studies varied whether or not the procedure produces decisions that are explained to participants. One element of procedural justice is whether or not authorities explain the basis for their decisions. In work-related studies conducted in experimental settings, there are often experimental variations introduced in terms of whether the supervisor does or does not explain how compensation was determined. Subsequent studies in this organizational justice literature have varied several aspects of work conditions in work organizations and then tested for any impact upon perceived justice. For example, variations of work conditions would include whether people are allowed to participate in a performance appraisal session at which their pay is determined or whether the reasons for job layoffs are explained to them. The study participants might participate in a performance task and receive or do not receive an explanation for the way their performance was rated when compensation was determined. The experimental variation might involve differences in how the basis for compensation was explained (or if it was explained at all) or, where appropriate, whether or not the participants had voice and could advocate for the quality of their work. These studies have found that a variety of types of human resource practice variations have a systematic impact (either positive or negative) with perceptions of procedural justice (Tremblay et al., 2010). Because these studies are experiments, they suggest evidence that variations in objective work conditions influence
perceptions of procedural justice. Similarly, elements of leader behavior are associated with procedural justice (Koivisto and Lipponen, 2015).
The court system is one type of organization in which organizational justice has been studied. An empirical literature evaluating the structure of the courts provides guidance concerning the features of courts that shape the nature of the interactions people have with authorities in courts. As an example, a substantial body of studies of restorative justice conferences have found that such conferences have a statistically significant association with later levels of recidivism, and are also experienced by participants as having more features of positive procedural justice than do the features of traditional case disposition (Hipple, Gruenewald, and McGarrell, 2014). Studies also have considered what happens in a courtroom. Greene and colleagues (2010) coded objective features of courtroom atmosphere and found that they were systematically related to litigants’ perceptions of justice.
The role of arbitrators is similar to the role of police officers in that they do not seek voluntary consent. However, both arbitrators and mediators (who do need the consent of the parties they deal with) want to craft solutions that will not be resisted and undermined by the two opposing parties, so they benefit from following the principles of procedural justice. There have been studies of the features of mediation and other alternative dispute resolution procedures that lead to their perceived fairness in the eyes of all of the parties in an interaction (Tyler, 1989). As with restorative justice conferences, those features can serve as the basis for procedural designs. Core features include giving both parties the ability to present their side of the story, having a neutral decision maker (the third party), believing that the third-party decision maker is listening to and considering each party’s arguments, and feeling that the third party is sincerely trying to reach a solution that is responsive to both opposing parties’ concerns.
Effective third parties in these informal proceedings know to treat the opposing parties with courtesy, to listen to and acknowledge their issues, and to account for those concerns when presenting proposed solutions (Tyler, 1987, 1988, 1989). They are aware that evidence of favoritism or bias undermines their authority. Because mediation focuses upon gaining voluntary acceptance, mediators involved in dispute resolutions learn from their experience to follow the principles of procedural justice.
Utility of employee training is another area in which the management literature helps in identifying impacts of procedural justice. To test the impact of training union officers in procedural justice, Skarlicki and Latham (1996) used a quasi-experimental design comparing union leaders who received procedural justice training with leaders who did not receive training. After 3 months of training, workers who were working under trained leaders reported greater procedural justice in their workplace and engaged
in more peer-assessed union citizenship behavior. These behavioral changes were found to be mediated by employee evaluations of procedural justice. Skarlicki and Latham (1997) replicated their first study and found similar outcomes, but they were only partially mediated by procedural justice. Cole and Latham (1997) replicated this training program and found that trained supervisors were rated by outside experts as solving problems more fairly. Another study conducted by Nakamura and colleagues (2016) randomly assigned managers to receive brief 90-minute training and found an impact 3 months later on the fairness of trained managers as perceived by lower-performing employees. Richter and colleagues (2016) designed a procedural justice training program for framing the delivery of bad news and found that trained managers were viewed as fairer and mitigated negative reactions associated with receiving bad news.
The workplace literature (see, e.g., DeCremer and Tyler, 2005) also identifies individual characteristics that are reliably associated with variations in perceived procedural justice. When people are more centrally focused upon their status and identity or when they draw more of their sense of themselves from membership and status in a group (e.g., because they strongly identify with it), they are more affected by their treatment. An explanation proposed to account for this association is that treatment communicates information about status and standing. Social scientists label such information relational because it communicates information relevant to social identity (Tyler and Lind, 1992).
The literature on social identity (Abrams and Hogg, 1988; Tyler and Blader, 2000) indicates that identification can be directly shaped by organizational structures and leader actions, suggesting another avenue for potential change management. In other words, these individual characteristics reflect variations in the nature of people’s connection to their community and to institutions in the community. Such connections are malleable and can be changed in a variety of ways.
When people receive feedback indicating either that their standing in a community is high or that the status of the community itself is high (or both), they are more likely to identify with that community. And as people identify more strongly with the community, they are more affected by whether or not they are treated justly, since such treatment communicates social identity–relevant information and their identities are more strongly intertwined with the community. Hence, a general approach to amplifying the role of procedural justice in the evaluation of community authorities is to strengthen the identification of residents with their community. This logic model also highlights the reciprocal influences of procedural justice and social identification upon one another. Procedural justice promotes identification of community members with both authorities and institutions (Tyler and Blader, 2000). Identification, in turn, leads to a greater emphasis
on procedural justice when reacting to authorities. Both of these processes evolve and interact over time.
Many of the ideas mentioned in the community-oriented policing literature reviewed above in this chapter are similar to ideas in the procedural justice research literature, in the sense that the focus is on the experiences of people in the community and on their behavior toward the police. Despite these similarities in conceptualization, studies of community-oriented policing have, as noted above, seldom directly assessed perceptions of procedural justice or injustice of different aspects of community-oriented policing programs. Hence, one clear limitation of the existing studies is the lack of examination of the connection between actual police policies and practices and measures of the different intervening psychological constructs outlined in the logic model. The committee therefore cannot draw upon the large community-oriented policing literature for guidance in this area.
On the other hand, in comparison to community policing studies that measure procedural justice, there is a larger policing literature that begins with perceived procedural justice and looks at its consequences (Donner et al., 2015). Although issues of causality and third (potentially confounding) variables remain open questions, a number of studies that measure associations among perceptions, either through a cross-sectional design or using panel designs involving interviews with members of the public, find statistically significant correlations between perceived procedural justice, perceptions of legitimacy, compliance, and cooperation.14
Several studies of policing suggest that procedural justice policing is
14 There have been a wide variety of approaches used to assess compliance, with most studies relying upon self-report of behavior. Cooperation has also been studied in a variety of ways. A typical approach has been to ask people if they would cooperate in an appropriate situation if one arose. For example, if called, would they serve on a jury? If they witnessed a crime, would they report it? See Bates, Allen, and Watson (2016); Bond and Gow (1996); Bradford (2011); Bradford et al. (2014, 2015); Casper, Tyler, and Fisher (1988); Dai, Frank, and Sun (2011); Elliott, Thomas, and Ogloff (2011); Fagan and Piquero (2007); Fagan and Tyler (2005); Gau and Brunson (2010, 2015); Goff, Epstein, and Reddy (2013); Hinds (2007, 2009); Hinds and Murphy (2007); Hasisi and Weisburd (2011); Jackson et al. (2012, 2013); Jonathan-Zamir and Weisburd (2013); Kane (2005); Mastrofski, Snipes, and Supina (1996); McCluskey (2003); Murphy (2005, 2013); Murphy, Hinds, and Fleming (2008); Myhill and Bradford (2012); Myhill and Quinton (2011); Norman (2009); Piquero, Gomez-Smith, and Langton (2004); Reisig and Lloyd (2009); Reisig, Tankebe, and Mesko (2014); Stott, Hoggett, and Pearson (2012); Sunshine and Tyler (2003); Tankebe (2013); Taylor and Lawton (2012); Tyler (1988, 2000, 2006, 2009, 2011); Tyler and Blader (2005); Tyler, Casper, and Fisher (1989); Tyler and Fagan (2008); Tyler, Fagan, and Geller (2014); Tyler and Huo (2002); Tyler and Jackson (2014); Tyler et al. (2007); Tyler, Schulhofer, and Huq (2010); Tyler and Wakslak (2004); Ward et al. (2011); Watson and Angell (2013); Wolfe et al. (2016).
strongly correlated with community members’ perceptions of legitimacy and their cooperation with police. For example, Donner and colleagues (2015) reviewed 28 studies and concluded that police interactions with the public that are informed by concepts of procedural justice are positively correlated with public views of police legitimacy and with trust in the police. This conclusion is supported by studies that use either subjective (i.e., citizen-experienced; see Mazerolle et al., 2013b; Wolfe et al., 2016) or objective (researcher-assessed) measures (Dai, Frank, and Sun, 2011; Mastrofski, Snipes, and Supina, 1996; Mazerolle et al., 2013a) of citizen cooperation. It also correlates positively with deference to police authority as reported in surveys (Tyler and Huo, 2002; Tyler and Fagan, 2008). At the same time, there is little evidence of correlation between objective procedural justice behaviors and citizen outcomes (Nagin and Telep, 2017). Indeed, only one study (Worden and McLean, 2014) compared objective versus subjective measures of procedural justice behaviors, and it found only a small, albeit statistically significant, correlation.15 That study also found that procedurally unjust behavior is more critical to evaluations than procedurally just behavior. These findings are consistent with Skogan’s (2006a) work suggesting that negative citizen/police encounters are far more consequential for citizen attitudes toward the police than positive encounters.
This is not to say that positive encounters cannot build trust; studies show that they can. Tyler and Fagan (2008) used a panel study design to demonstrate that fair contacts were found to be statistically significantly associated with increased trust among those with contact with the New York City Police Department, although negative contacts had a stronger influence. Tyler, Fagan, and Geller (2014) used a similar panel design but focused upon 18 - to 26-year-olds in New York City. They found that both fair and unfair contacts were associated with changes in perceived legitimacy, and both were equally influential.
Several recent experimental studies explore the impact of procedurally just treatment on citizen attitudes toward the police, as well as their cooperative behavior. These studies do not at this time provide a clear conclusion regarding whether procedural justice policing improves perceptions of police legitimacy and cooperation. Mazerolle and colleagues (2013b) conducted one such study focusing upon police stops in Australia. They found that a single-stop experience that the civilian viewed as reflecting procedural justice or injustice generalized to shape trust in the police in the community. This study, called the Queensland Community Engagement Trial, was a randomized controlled trial that delivered an experimental treatment to each
15 In this study, the categories used by observers were drawn from theories about procedural justice. Similarly, the dimensions of citizen perception assessed were drawn from those same theories.
stopped civilian in the form of a scripted set of officer statements during traffic checks for drunk driving. Randomly chosen officers were trained to follow a detailed protocol designed to maximize the procedural justice of the brief interactions occasioned by random breath testing (RBT). Reactions to those officers were compared with the reactions to officers not trained using this special script. Scripts were designed to incorporate the elements of procedural justice into officers’ statements during the stop. During 30 of 60 RBT operations, officers were directed to use the experimental script, and senior officers monitored their compliance with the statements listed in the protocol. These police-citizen encounters were quite brief: ordinarily (i.e., in the control condition), they were “very systematic and often devoid of anything but compulsory communication” (Mazerolle et al., 2013b, p. 40). The control-condition encounter was about 20 seconds in duration and did not have the procedural justice statements. The scripted, procedurally just encounters were longer, at 97 seconds on average, but still quite brief. Each driver who was stopped during these 60 RBT encounters was given a survey to complete later and return to the researchers. The procedural justice treatment had the hypothesized impact on civilians’ judgments. However, response rates, for both experimental and control drivers, were only about 13 percent. This low rate of return raised concerns about the strength and generalizability of the findings.
The design of the Queensland Community Engagement Trial, but not its results, has been replicated in other settings (MacQueen and Bradford, 2015; Sahin, 2014). MacQueen and Bradford (2015) used a block-randomized design with pre- and post-test measures built around a similar type of police-civilian experience. Their treatment was also a stop procedure that involved the presentation of key messages and subsequent distribution of a leaflet to motorists, through which they evaluated their experience. The study found no significant improvements in general trust in the police or in perceived police legitimacy.16 Similarly, a recent experiment using traffic stops in Turkey (Sahin et al., 2016) found that officer behavior during traffic stops shaped views about the particular police officers involved but did not generalize to overall perceptions about the traffic police as an organization. And Lowrey, Maguire, and Bennett (2016), who studied street stops by having observers view video clips of police and civilian actions and verbal statements during traffic stops, found an impact upon specific evaluations of the stop, including obligation to obey the particular officers and having trust and confidence in those officers, but not on generalizations to broader attitudes about the police as an organization.
16 The committee notes that the failure of the study may be due to implementation errors and does not necessarily suggest that the theory informing procedural justice is wrong (>MacQueen and Bradford, 2016).
These particular forms of police contact are all highly scripted and therefore do not vary in the ways that other forms of police contact do. They do reflect the highly scripted nature of traffic stops. Worden and McLean (2016, p. 34) commented: “Traffic checkpoints that involve very brief encounters between police and citizens are susceptible to such prescriptions, but police–citizen encounters in most domains of police work—and especially in those with the strong potential for contentious interactions—do not lend themselves to such experimental or administrative manipulation.” Studies of the police emphasize that the police normally deal with a wide variety of situations many of which are less scripted, and different officers have very different styles of addressing each type of situation (Muir, 1977). More specifically, Epp, Maynard-Moody, and Haider-Markel (2014) argued that it is investigatory street stops, not traffic stops, that are central to creating feelings of injustice among community residents, since traffic stops are routinized and linked to understandable violations of known laws, whereas citizens stopped on the street are often confused about what, if anything, they have done to justify the stop. Hence, traffic stops are much less likely to create variations in perceived unfairness in treatment on the part of civilians who have contact with police officers and hence are less likely to have an impact on perceived legitimacy.
In the case of assessing impact on cooperation, Mazerolle and colleagues (2013c) created a combined measure of self-reported behavioral ongoing compliance and future willingness to cooperate. They evaluated five experimental studies that provided eight outcome measures. In three of eight cases there is a statistically significant influence of police intervention upon compliance/cooperation. Mazerolle and colleagues (2013c, p. 261) concluded that the results suggest that the “interventions had [a] large, significant, positive association with a combined measure of compliance and cooperation.” Another study by Mazerolle and colleagues (2014) contains an extended meta-analysis on procedural justice effects. In reviewing community policing efforts with procedural justice elements, the authors found four studies exploring influence upon compliance/cooperation and reported three statistically significant relationships in the expected direction (Mazerolle et al., p. 28). Experiencing fairness promotes compliance and cooperation. For restorative justice conferencing, they found four studies that examined influence on compliance/cooperation and four statistically significant relationships (Mazerolle et al., p. 29). The authors concluded that procedural justice has positive effects upon perceived legitimacy and that procedural justice and perceived legitimacy jointly shape self-reported compliance/cooperation.17
17 Other studies also find an influence on cooperation (Hinds, 2009; McLean and Wolfe, 2016; Murphy, 2013; Sunshine and Tyler, 2003; Tyler and Fagan, 2008; Tyler, Goff, and MacCoun, 2015; Van Damme, Pauwels, and Svensson, 2015; White, Mulvey, and Dario, 2016).
This relatively optimistic conclusion was questioned by Nagin and Telep (2017), who reviewed the same and also more recent studies. In particular, as has been noted, several recent efforts have failed to replicate the Mazerolle study on traffic stops. In addition, the reviews by Mazerolle and colleagues took a more expansive view of studies that constitute tests of the effects of the perceived legitimacy of the police. They included any study that met other technical inclusion criterion (e.g., reported data required to measure effect sizes) and that either had as one purpose improving perceived police legitimacy or articulated an objective that was consistent with Tyler’s conception of procedurally just treatment.
In light of these issues affecting the evidence base, the committee agreed that a strong conclusion regarding the impacts of procedural justice policing on people’s evaluations of police legitimacy (i.e., on perceived legitimacy) or on people’s cooperation with the police could not be drawn from existing studies on the police.
Recent studies suggest that perceived procedural justice may impact identification with the community, social capital, and engagement in the community (Kochel, 2012; Tyler and Jackson, 2014). Kochel (2012) studied the police in Trinidad and Tobago through interviews with 2,969 people in 13 police districts and found that the nature of police-citizen interactions was associated with collective efficacy. Collective efficacy is particularly strongly and positively associated with judgments about the quality of police services, a combined measure that includes satisfaction with services and judgments about whether the police are competent, respectful, capable of maintaining order, and willing to help citizens with their problems.18Tyler and Jackson (2014) conducted a national survey and found that procedural justice and perceived legitimacy of policing are associated with identification with the community, collective efficacy, and behaviors, such as likelihood of shopping in the community and participating in local politics. These findings suggest that the perceived fairness of policing has an impact beyond the arena of crime and criminal justice—it more broadly affects communities and their well-being.
This literature has several problems. First, it generally begins with community perceptions and evaluations of what the police are doing, rather than using objective, researcher-assessed first-hand accounts of actual police actions. A small number of studies directly connect police actions to perceptions about the police (Worden and McLean, 2014). For example, Mazerolle and colleagues (2013a) conducted a meta-analysis that considers six experimental studies; they concluded that interventions are found to be associated with “large, significant increase in perceptions of procedural
18 Unfortunately, this study does not cleanly distinguish procedures from outcomes because it combines process and outcome measures.
justice” (Mazerolle et al., 2013a, p. 261). However, the specific police actions associated with this impact are often not clear. This is an important area for further research.
Another example is given by Jonathan-Zamir, Mastrofski, and Moyal (2015). This field observation of police–citizen interactions measured the relationship between researcher-established measures of the degree of police procedural justice behavior and the observable attitude of the citizen toward the police at the end of the encounter. Observers noted that in half of the 156 observed encounters, citizens manifested behaviors that signaled an attitudinal orientation to the police. They found a strong, statistically significant difference: “encounters in which the officer displayed higher levels of procedural justice were significantly likely to yield overall satisfaction with the police handling of the situation at the encounter’s conclusion” (Jonathan-Zamir, Mastrofksi, and Moyal, 2015, p. 862). Of course, displayed attitudes may not reflect how a citizen actually feels, and the study was unable to detect any attitudinal valence for half of the observed citizens.
Worden and McLean (2014) rectified this problem. They compared different aspects of overt police officer behavior, as identified by observers, to citizens’ self-reported perceptions of procedural justice. Using multiple regression analysis, they estimated that the objective ratings could account for only around 10 percent of the variance in subjective perceptions. Procedural injustice had a greater effect on subjective experience. This asymmetry is found to stem not from the relatively strong effects of negative experiences but rather from people’s tendency to overestimate the procedural justice with which the police are acting, as compared to researchers’ objective judgments of how the police are acting. People who deal with the police are generally positive in their ratings of police performance, even when the degree of procedural justice, as rated by observers, is low. The authors suggested that reactions to a specific experience reflect both what happens in that experience and the general attitudes toward the police that people bring into the situation.
Interestingly, the Worden and McLean (2014) findings also indicate that people’s judgments about the propriety of police action are correlated more strongly to perceptions of the procedural justice of police actions than to the actual legality of officers’ behavior. This echoes the results of a recent experimental study that presented people with videos of police-citizen interactions varying in procedural justice (Meares, Tyler, and Gardener, 2016). That study provided contextual information indicating that the officers acted legally or illegally. Also, respondents were presented with scenarios varying in the actual legality of police conduct. These variations had little impact upon judgments about the appropriateness of police actions. Instead, the results indicate that these citizen judgments of police propriety
were primarily driven by the procedural justice of police actions, not by their actual legality.
Finally, Worden and McLean (2014) speculated that the relationship between the police and the public is a reciprocal one. They postulated that if the citizen is disrespectful or resistant, then that can lead the police to use physical force, and when the police use physical force they are then evaluated as less procedurally just. This suggests the potential limitation of studies that do not consider reciprocal influences—a possible limitation in any non-experimental study. The type of contact people have (traffic stop, investigatory stop, call for help, etc.) also shapes ratings of the police. Searches are associated with low ratings of procedural justice.
As this review has noted, there have been very few studies in the area of policing that connect police policies and practices and/or the actions of police officers to the perceptions of people in the community about the police. Despite the current lack of direct evidence in the policing arena, evidence exists in other literatures that suggests that developing procedural justice approaches may be possible in the arena of policing. One such area is a substantial body of research consistent with, but by no means conclusive proof of, the hypothesis that procedural justice training may change police behavior in the field. For example, there is research consistent with the idea that officers trained in the principles of procedural justice express more support for using procedural justice when dealing with people in the community than do officers without this training. The trained officers also express stronger commitment to the goals and standards of the organization they work for. Some of this evidence is the result of experimental evaluations of training programs, which can be interpreted as causal evidence (e.g., Schaefer and Hughes, 2016; Skogan, Van Craen, and Hennessy, 2015). However, the majority of this research is based on correlational analyses of the results of officer surveys, sometimes augmented with objective or third-party performance evaluations (Bradford et al., 2014; DeAngelis and Kupchik, 2007, 2009; Farmer, Beehr, and Love, 2003; Taxman and Gordon, 200919; Trinkner, Tyler, and Goff, 2016; Tyler, Callahan, and Frost, 2007; Wolfe and Piquero, 2011). One should therefore be careful to not attribute a causal interpretation to these findings.
There are also a handful of studies that suggest that officers trained in procedural justice concepts may be more successful at incident de-escalation
19Taxman and Gordon (2009) survey correctional officers, and we include this study because of the strong relationship between the oversight and enforcement aspect of police and correctional officer’s professional tasks.
in the field (Wheller et al., 2013; Owens et al., 2016). One approach to changing police officer behavior is through training officers to use procedural justice in their policing activities. A second approach is to make internal department dynamics more consistent with procedural justice, on the assumption that, as a consequence, officers will adopt these fairer approaches as a general aspect of how they police, without the need for explicit training programs. There is evidence consistent with the suggestion that changes in the internal dynamics of police departments lead to changes in police behavior. When officers experience their superiors in their own departments as being procedurally fair, they are perceived to be fairer in their actions when dealing with the public, they express more support for using procedural justice when dealing with people in the community, and they are less likely to engage in actions such as the use of force (Bradford et al., 2014; DeAngelis and Kupchik, 2007, 2009; Farmer, Beehr, and Love, 2003; Harris and Worden, 2014; Taxman and Gordon, 2009; Trinkner, Tyler, and Goff, 2016; Tyler, Callahan, and Frost, 2007; Wolfe and Piquero, 2011).
The research literature on interventions that take a community-based approach concentrates on three main strategies for proactive policing: community-oriented policing, broken windows policing, and procedural justice policing. The committee reviewed each of these strategies in terms of the evidence for associations with and causal impacts on community outcomes. Given the focus in the logic model for each of these strategies on altering community perceptions and behavior, there is a surprisingly limited research literature on community outcomes.
Of these three strategies, community-oriented policing has had the most extensive examination of the association of police practices with community outcomes. Nonetheless, as we noted in the beginning of the chapter, it is difficult to draw very strong conclusions from this literature. The nature of the benefits of community views of police and policing is ambiguous because there is inconsistency across studies in the conceptualization and measurement of different community outcomes. For example, measures that are presented as indicators of citizen satisfaction with police practices in one study are considered indicators of perceived legitimacy in another. This ambiguity makes the synthesis of findings across studies challenging because researchers do not apply a consistent or standardized set of measures for a given outcome.
A fundamental challenge for understanding the implications of evaluations of community-oriented policing is the great variation exhibited in the content of community-oriented policing elements (or tactics) that comprise the actual intervention evaluated. The range of elements, how they are spe-
cifically accomplished, and the intensity with which they are implemented vary tremendously from study site to study site. Many evaluations give short shrift to the implementation issue, yet those that have examined it in depth have found such challenges to be profound, implicating this as a source of the heterogeneity of effects that have been observed. The absence of a standardized framework for developing a meaningful taxonomy of community-oriented policing practices employed in actual interventions prevents the committee from identifying with confidence specific features, much less combinations of features, that contribute to stronger positive community impacts. Moreover, very few studies of community-oriented policing have traced its long-term effects (beyond a year) on community outcomes or its jurisdictionwide consequences. Therefore, it is difficult to say with confidence what long-term exposure to community-oriented policing produces in community reactions across the full jurisdiction. Understanding and explaining long-term trajectories of community impacts requires monitoring program implementation fidelity over time, as well as monitoring an array of forces and events that originate outside the program and the police organization.
With these limitations in mind, the committee drew the following conclusions from its review of the community-oriented policing research literature.
CONCLUSION 6-1 Community-oriented policing leads to modest improvements in the public’s view of policing and the police in the short term. (Very few studies of community-oriented policing have traced its long-term effects on community outcomes or its jurisdictionwide consequences.) These improvements occur with greatest consistency for measures of community satisfaction and less so for measures of perceived disorder, fear of crime, and police legitimacy. Evaluations of community-oriented policing rarely find “backfire” effects on community attitudes. Hence, the deployment of community-oriented policing as a proactive strategy seems to offer prospects for modest gains at little risk of negative consequences.
CONCLUSION 6-2 Due to the small number of studies, mixed findings, and methodological limitations, no conclusion can be drawn about the impact of community-oriented policing on collective efficacy and citizen cooperative behavior.
Broken windows policing is often evaluated directly in terms of its short-term crime-control impacts. We have emphasized in this report that the broken windows policing model seeks to alter the community’s levels of fear and collective efficacy as a method of enhancing community social
controls and reducing crime in the long run. While this is a key element of the broken windows policing model, the committee’s review showed that these outcomes are seldom examined. In the case of collective efficacy, only one study reported an outcome on this issue, and the committee did not believe that this evidence was persuasive enough to draw a conclusion. In the case of fear of crime, a larger number of studies were available.
CONCLUSION 6-3 The committee is not able to draw a conclusion regarding the impacts of broken windows policing on fear of crime or collective efficacy. This is due in part to the surprisingly small number of studies that examine the community outcomes of broken windows policing and in part to the mixed effects observed.
Procedural justice policing relies on a logic model that posits that perceptions of police legitimacy are primarily responsive to community members’ evaluations of the procedural justice that people experience when dealing with authorities. Procedural justice involves judgments about how fairly: (1) decisions are made and (2) people are treated. The procedural justice model of perceived legitimacy has received empirical support in psychological studies conducted in laboratory settings. The procedural justice model has also received empirical support from studies conducted by organizational psychologists in work settings. The key question for the committee is whether the relationships found in these domains can be extended to the domain of proactive policing practices in real-world communities. While there is a rapidly growing body of research on the community impacts of procedural justice policing, it is difficult to draw causal inferences from these studies because most existing studies rely on cross-sectional or correlational designs, and there are very few field experiments to clarify the causation underlying observed statistical associations. The committee therefore reached the following general conclusions regarding this question:
CONCLUSION 6-4 In general, studies show that perceptions of procedurally just treatment are strongly and positively associated with subjective evaluations of police legitimacy and cooperation with the police. However, the research base is currently insufficient to draw conclusions about whether procedurally just policing causally influences either perceived legitimacy or cooperation.
CONCLUSION 6-5 Although the application of procedural justice concepts to policing is relatively new, there are more extensive literatures on procedural justice in social psychology, in management, and with other legal authorities such as the courts. Those studies are often designed in ways that make causal inferences more compelling, and
results in those areas suggest that the application of procedural justice concepts to policing has promise and that further studies are needed to examine the degree to which the success of such strategies in those other domains can be replicated in the domain of policing.
While Conclusion 6-4 may appear to be at odds with a growing movement to encourage procedurally just behavior among the police (see, e.g., President’s Task Force on 21st Century Policing, 2015), the committee stresses that a finding that we did not have evidence to support the expected outcomes of procedural justice policing is different from finding that such outcomes do not exist. The extant literature in this area is sparse and has only begun to develop in recent years, and the evidence from this small group of existing studies is simply not consistent enough for the committee to draw a stronger conclusion. At the same time, the principles of procedural justice are likely to be consistent with many of the goals of policing in democratic societies, a subject discussed further in Chapter 3 of this report. What is missing to date is information on the extent to which these principles will affect community attitudes toward the police as well as individuals’ cooperation with the police. On the other hand, studies generally do not find negative effects of pursuing procedural justice strategies, suggesting that there is little likelihood of undermining existing trust in the police or otherwise undermining policing through implementing these approaches (although, as we suggested in Chapter 3, they may raise other concerns about legality and transparency not yet explored in the empirical literature).
This page intentionally left blank.