5

Community Reaction to Proactive Policing: The Impact of Place-Based, Problem-Solving, and Person-Focused Approaches

The purpose of this chapter and the one that follows is to describe what is known about community reactions to various forms of proactive policing. We treat these outcomes as a distinct category, separate from the outcome of efforts by the police to manage proactively the rate and type of crime. There is broad recognition that a positive relationship between the police and the community has value in its own right, irrespective of any influence it may have on crime, disorder, or public safety (National Research Council, 2004, p. 291; Lum and Nagin, 2017). This view has gained traction in the recent public discussion of policing. As an example, the President’s Task Force on 21st Century Policing (2015) labeled popular legitimacy (i.e., public trust in the police) the “first pillar” of policing. This perspective was also echoed in the discussions that the committee had with representatives from various community organizations, as well as police practitioners (see Appendix A). Police leaders consistently emphasized that community perceptions and feelings about their police and the practices of those police were essential criteria for selecting policing strategies and judging police performance. Representatives of community organizations observed that members of a community give high priority to a broad range of performance issues extending well beyond the relatively narrow confines of crime and disorder control. They argued that those members of the public most alienated from and resistant to the police are profoundly motivated by perceptions of long-term police disrespect and inattention to the broader welfare of communities.

Such judgments derive from philosophical valuations of what is prized in a democratic society and the role of police in pursuit of those goals.

Democratic theories assert that the police, as an arm of government, are here to serve the community and should be accountable to it in ways that elicit public approval and consent. Specific notions of precisely what constitutes democratic policing vary, but most are built on a foundation of “trust, equality, and legitimacy” (Manning, 2010, p. 3; see also Sparrow, 2016), with restraint and the minimization of harm, responsiveness to what people want, accountability to legal institutions, and the reduction of inequality as frequent themes of what is essential to the creation and preservation of democratic policing (Manning, 2010, Chs. 1 and 11). It is easy to see why a democratic society wants authorities who strive to meet these expectations, and it has been claimed that public feelings about the trustworthiness, equality, and perceived legitimacy of policing have played a key role in fueling the intense public dissatisfaction and scrutiny experienced by American police organizations in recent years (President’s Task Force on 21st Century Policing, 2015).

The proactive policing strategies reviewed in this report have as one of their primary goals the reduction of crime and disorder. The initial motivating force behind some of these approaches was this goal of reducing crime and disorder, with the potential for negative community outcomes constituting collateral concerns (Rosenbaum, 2006). Place-based, problem-solving, and person-focused approaches fall into this category. However, strategies falling into the broad category of a community-based approach were launched first and foremost as a corrective to community alienation from the police, with subsequent interest growing in their capacity for crime and disorder control as well (Skogan, 2006b). Hence, this chapter focuses on the community impacts of the three approaches that give primacy to crime and disorder control. Chapter 6 considers the community effects for strategies that were launched with improving community effects as the initial rationale.

This chapter is organized as follows. First it discusses the key types of community impacts on which it focuses. It then provides a preliminary model that links the key elements of community effects: a model that underlies much of the research that is relevant to tracing the impacts of proactive policing on the community. We then organize our discussion of findings into three separate sections, one each for the three broad proactive policing approaches that give primacy to controlling crime and disorder: the place-based, problem-solving, and person-focused approaches as defined in Chapter 2. Each section includes a description of the presumed mechanisms by which the intervention affects community outcomes and a discussion of limitations of the extant research. In addition to these more-or-less proximal community reactions to proactive strategies, we discuss the small and diffuse literature on the indirect, or “collateral,” effects of proactive strategies on societal conditions such as public health and civic engagement. In

the chapter’s final section, we present and briefly discuss the conclusions we have drawn.

WHAT DO WE MEAN BY COMMUNITY IMPACTS?

The committee considers three types of community reactions or outcomes: evaluations, orientations, and behaviors. First, how do proactive strategies affect the way people evaluate their experiences with and impressions about what police do? Do they judge that police behave effectively (e.g., in reducing crime and disorder; in responding to calls for help)? Are the police fair and considerate toward the public? Do the police apply their authority and distribute their services equitably? Are the members of the community content with the nature of police service?

Second, how do proactive strategies affect the way people orient toward the police as an institution? Do people have trust and confidence in the police—that is, view them as legitimate? Much of the recent public discussion of policing has focused on public trust, which is one aspect of what is more generally called perceived legitimacy. The other aspects of perceived legitimacy are the perceived obligation to defer to the police, which motivates a willing acceptance of police authority, and normative alignment, the belief that the police and community share common values (Tyler and Jackson, 2014).

Third, how do proactive strategies affect the ways that people behave toward the police, the law, and their communities? Do people become more cooperative with police and other legal institutions? The legal system relies upon members of the community to report crimes, identify criminals, act as witnesses in court, and serve on juries. More broadly, do the police behave in ways that strengthen the community’s collective efficacy¹ and thereby facilitate the creation of social capital among members of the community?

A MODEL OF THE EFFECTS OF PROACTIVE POLICING ON COMMUNITY OUTCOMES

For assessing community outcomes in this and the following chapter, the committee relies on a logic model that has framed much of the research on community effects, one that links community evaluative judgments to community orientations and ultimately to behaviors. The model begins with formal police policies, which are presumed to shape police officer actions on the street that are relevant to the community. The policies are

___________________

¹Sampson, Raudenbush, and Earls (1997) coined the term “collective efficacy” to refer to the degree to which people who live in communities trust their neighbors and are willing to intervene in community affairs.

also assumed to affect community practices where community involvement with police is part of the intervention (e.g., community participation in collaborative efforts with the police). Police and community actions, in turn, are hypothesized to shape the sort of evaluative judgments community members make about police performance (effective, fair, lawful). And these evaluations are seen to shape the general orientation toward the police (perceived police legitimacy). Perceived legitimacy in turn is hypothesized to shape the behavior of community members in terms of law abidingness, cooperation with authorities, and engagement in the community. Figure 5-1 depicts this linkage.

We label this a logic model to make clear that it depicts a theoretically postulated flow of effects. The validity of this flow as a causal description is something that must be separately evaluated and will be discussed in our review of the evidence. In addition, it is important to recognize that although this logic model proposes a linear progression through stages 1 to 5, it is possible that there are reciprocal influences, an expectation recognized by arrows pointing in the reverse direction in Figure 5-1. Also, there could be other factors (“third variables”) that are a part of this model. These issues need to be considered when determining whether this logic model is empirically supportable as a causal model for any particular policing strategy or fielded intervention.

The stage numbers in Figure 5-1 are intended to convey a temporal sequence. Policies are the purposive, official acts of public figures with responsibility for directing the practices of the police. Policies will vary in the nature and extent to which they promote or emphasize a given proactive strategy. Indicators of police practice reveal the fidelity of actual police actions to the ideal established by policy—or the extent or dosage of the implementation. Community evaluations reflect how members of the community rate the performance of the police on relevant criteria. A variety of criteria for evaluating police performance have been proposed, including crime-control effectiveness, equity in the distribution of services, palatability of the experience of contact with the police, the perceived procedural justice of police actions, and satisfaction with police services. These are judgments about what people believe that police officers actually do or accomplish while on the job, particularly within their own communities.

Community evaluations are hypothesized to develop from people’s personal experiences, the experiences of their family and friends, and what they see occurring within their community.² Community orientations indicate how community members feel about the police: their trust, confidence, or deference to police authority. And, finally, community behavior refers to actions community members take that are relevant to levels of crime, disorder, and manifestations of civic virtue or societal economic contributions. In later sections, we will consider some limitations to this model as a representation of the causal process.

Before beginning the review of findings in this and the following chapter, we reiterate an earlier point about the geographic level of impacts of the approaches examined. As with studies of crime effects, units of analysis for community effects in the literature we reviewed were usually areas much smaller than the entire jurisdiction: typically neighborhoods, police beats, districts (multiple beats), census blocks, or hot spots. We know surprisingly little about whether and to what degree proactive policing strategies influence community outcomes in the larger urban areas within which such strategies are implemented. Without estimates of the areawide impacts of proactive policing strategies, it is difficult to assess whether these strategies, applied broadly in jurisdictions, would have meaningful effects across entire jurisdictions.

PLACE-BASED INTERVENTIONS

Place-based interventions concentrate police efforts at the microgeographic spaces where crime or disorder concentrates. Hot spots policing is the most common strategy for this proactive policing approach, but that speaks only to the concentration of police resources according to the concentration of crime or disorder. The content of such interventions can vary widely, drawing on tactics also used in one or more of the other strategies considered in Chapters 2 and 4, such as police patrol, crackdowns, or practices typical of problem-oriented policing. In this section, we consider the full variety of such hot spots interventions as place based.

One of the earliest tests of hot spots interventions on community perceptions is provided by Shaw (1995), who conducted a matched comparison group quasi-experiment in Kansas City, Missouri, comparing residents’ reactions to gun-detection patrols in a target area to a comparison area. The two-phased, person-focused hot spot intervention involved a door-to-door consultation with the community preceding the proactive patrols. The precise nature of the proactive tactics, all involving officer-initiated contacts

___________________

² Of course, people may observe these events directly, but their impressions may also be shaped via news and social media.

with the public, was left to officer discretion. Car checks, frequently involving a traffic violation, were the most frequent occasions to look for illicit weapons. Field observations of unknown reliability³ and the absence of complaints and lawsuits suggested a “general absence of excess in police encounters” (Shaw, 1995, p. 700). The study found that there was no appreciable difference between the target beats and control beats in terms of support for proactive police interventions (high portions of both saying it was “good for the neighborhood”) (p. 704).⁴ It also found that target-area residents observed a higher quality of neighborhood life following the intervention (for both social disorder and fear of crime), but both areas experienced similar reductions in crime. The panel design of this study suffered from a relatively low sample size, so that, with attrition, the time 2 samples were relatively small for both the treatment and comparison groups (64–71 respondents per group).

A subsequent randomized controlled trial (RCT) used an interesting approach to measuring community reactions to place-based problem-oriented policing in a hot spots framework by interviewing 52 “key community residents” who shaped the way a public space is used at some point during the day in treatment and control areas (Braga and Bond, 2009; see reference to this study in the section below on “Problem-Solving Interventions”). The specific problem-solving practices (in Lowell, MA) included aggressive enforcement, but many also used social service tactics to disrupt underlying conditions. The outcome analysis revealed that the key residents in the treatment areas observed heightened police presence⁵ and a decline in perceived disorder. But they did not note changes in policing strategy, inclination of the police to work with residents, police demeanor toward the public, or fear of crime. Hence, there were some positive community reactions and no significant “backfire” collateral effects of the crime-prevention strategies.

Two other studies examined the impact of police crackdowns, one with a disorder reduction approach. A quasi-experiment reported by Hinkle and Weisburd (2008) found that in Jersey City, New Jersey, intensive police crackdowns meant to reduce crime and disorder at a drug market increased

___________________

³ The study was written and published posthumously by the author’s colleagues, who were unable to access some of the field observation material.

⁴ Eighty-eight percent of treatment-area residents rated this type of enforcement as good for the neighborhood; 82 percent of the control-area residents gave the same rating. The difference was not statistically significant.

⁵ It is difficult to know how to interpret this indicator, inasmuch as it covered the complete array of possible reasons to have contact with the police.

fear of crime, and the effect size was substantial (odds ratio = 3.12).⁶ Residents exposed to a crackdown strategy had more than three times the odds of developing fear of crime compared to people not exposed to a crackdown. The researchers speculated that the greatly heightened police presence and visibility created by the enforcement crackdowns may have increased residents’ sense that their neighborhoods were not safe. However, a later block randomized experiment by Weisburd and colleagues (2011) looked at typical broken windows practices for disorder reduction applied to hot spots policing in three medium-sized California cities. They found no statistically significant effects across a broad range of community indicators, including fear of crime, perceived police legitimacy, collective efficacy, and perceptions of crime. Higher levels of perceived physical disorder, which were marginally statistically significant, were found in hot spot treatment areas compared to control areas, but these perceptions had not manifested themselves in more fearful or unhappy residents. So overall, the results appeared not to confirm concerns some have expressed about potentially negative consequences of hot spots policing on community outcomes (Kochel, 2011; Rosenbaum, 2006).

Weighing the differences between these two studies of the impact of place-based use of broken windows tactics is important for drawing conclusions. The intervention in Jersey City was not designed to undertake a full broken windows strategy; rather, its tactics included an intensive crackdown on drugs, prostitution, and social/physical disorder and the removal of violent offenders, tactics shared with the broken windows strategy. Evaluators of hot spots policing in the three California cities examined a program specifically designed as a broken windows strategy but incorporating more measured police interventions than sometimes used in that strategy (e.g., warnings and explanations for first offenders, citations and arrests for repeat offenders). The Jersey City intervention took place in high-crime, high-violence neighborhoods; the California city intervention took place in three smaller cities with lower levels of serious crime and disorder. And the researchers noted that both the duration/dosage of the treatment and the short-term measurement of outcomes may have been inadequate to engender and measure a range of community effects.

Using an RCT, Weisburd, Morris, and Ready (2008) reported the effects of a different place-based intervention: a community policing/problem-solving combination that targeted risk and protective delinquency factors.

___________________

⁶ See Lipsey and Wilson (2001) and Wilson (n.d.) for guidelines on interpreting odds ratio effect sizes. Guidelines based on Cohen’s “Rules-of-Thumb” thresholds are small (1.50), medium (2.50), and large (4.30). They note that these guidelines do not consider the intervention’s context. For example, a small effect could be impressive if it requires few resources or other costs. Smaller effects could also be interpreted as substantial where the problem at issue is severe or impervious to change.

The principal policing tactic was increasing positive contact between police and the juveniles living in selected block groups. Notably, this intervention used a much larger geographic unit (a census block) than would qualify as a microgeographic space (e.g., an address, street segment, or small cluster of street segments). The study found no appreciable, statistically significant difference between treatment and control groups in the students’ perceptions of police legitimacy (a composite of ratings of police respectfulness, trust, fairness, and honesty).

Another study examined the effects of hot spots policing in Philadelphia under three experimental conditions: foot patrol, person-focused policing (repeat offenders), and problem-oriented policing (Ratcliffe et al., 2015).⁷ Control hot spots maintained the usual random patrol between calls for service. (Only the person-focused condition yielded statistically significant crime reductions compared to the control condition.) Survey data of community outcomes were analyzed using a quasi-experimental design. None of the three experimental interventions showed statistically significant effects across seven community outcome indicators: perception of violent crime, satisfaction with police services, perceptions of property crime, perceptions of physical disorder, perceptions of social disorder, perceptions of safety, and perceptions of procedural justice. The mailed citizen survey had a low (9%) response rate that could be attributed in part to underestimates of unoccupied addresses; area weighting of census data was used to adjust for over- and under-representation of different demographic groups in the sample (Ratcliffe et al., 2015, p. 402).⁸ This study is noteworthy in part because the city studied had higher treatment duration levels than other studies and took place in a higher-crime urban area.

The most recent of this group of studies was an RCT reported by Kochel and Weisburd (2017). This study assessed the effects of hot spots policing on police legitimacy and collective efficacy in St. Louis County, Missouri. Two types of hot spots interventions were evaluated—doubling time spent in hot spots (directed patrol) and problem solving in hot spots—with both treatment conditions compared to standard police practice. A diverse array of community outcomes was measured in three waves of resident surveys, the last wave occurring 9 months after treatment ended. Outcomes included assessment of police competence and satisfaction with

___________________

⁷ This is a truly hybrid approach that implemented a place-based strategy with tactics typical of both person-focused and community-based strategies.

⁸ Research indicates that nonresponse by itself is a weak, and sometimes even negative, predictor of nonresponse bias. (See Pickett, 2017, for a recent overview of this literature as it pertains to criminal justice research.) The comparison of survey respondents to 2010 census data showed that White and older female respondents, as well as those with more education, were a little overrepresented, compared to their presence in the general population. The researchers concluded that the sample of respondents closely approximated that of the actual population.

police, police legitimacy, procedural justice and trust, perceptions of police misconduct, feelings of personal safety, cooperation with police, and collective efficacy.

Community impacts were diverse. Directed patrol and problem solving did not affect perceptions of police competence or resident satisfaction with police. There were initial declines in perceived legitimacy for directed patrol and problem solving, but by 9 months there were no differences (there were statistically significant increases in perceived legitimacy between the 6th and 9th month). Directed patrol dampened increases in procedural justice/trust from wave 1 to 2, but the effect evaporated by the last wave. Directed patrol did not generate resident concerns about aggressive policing. Residents in problem-solving areas experienced negative results regarding feelings of personal safety, but these dissipated over time. Both directed patrol and problem solving generated long-term improvements in residents’ willingness to cooperate, and there were long-term benefits for collective efficacy delivered by directed patrol.

Overall, then, this study showed no “long-term” (9 month) effects for most community indicators and improvements in a couple of them (cooperation with the police and collective efficacy). Given that the analysis also showed crime reduction for both treatment groups, one might interpret these results as encouraging; there appear to be no tradeoffs in the longer term. But an important lesson taken from this study is that place-based directed patrol and problem solving did generate some initial community negativity on some indicators, but over a relatively short time these effects were either nullified or reversed. This temporal effect reinforces the notion that outcomes are dynamic and that it matters how far out from the intervention they are measured. This RCT measured those dynamics over the course of a relatively short time period. It would be useful to know temporal patterns over a time period of several years.

These studies of place-based strategies have centered on interventions in hot spots, but the diversity of police tactics employed is remarkable: gun detection patrols, broken windows enforcement, focusing on repeat offenders, directed patrol, and problem-oriented policing. Despite this diversity, there has been relatively little variation in findings about the community consequences of the interventions: for the most part, researchers do not find statistically significant effects. One evaluation of a Jersey City crackdown did yield a fairly substantial increase in fear of crime, one that was not replicated in a later RCT in three small California departments, an intervention that may not have been as intense in the sorts of enforcement activities that are visible to the public. Statistically significant beneficial effects were also relatively rare: a Kansas City, Missouri, gun detection patrol project and a problem-oriented hot spots approach in Lowell, Massachusetts, both registered some statistically significant reductions in fear of crime or perceptions

of disorder. And directed patrol and problem solving in St. Louis County, Missouri, yielded statistically significant improvements in collective efficacy and cooperation with police (though no statistically significant outcomes across the many other community impacts measured). Tellingly, none of the five tests of outcomes that could be classified as citizen satisfaction with the police or perceived police legitimacy yielded statistically significant effects.

The committee concluded that the extant research suggests that a place-based policing strategy rarely leads to negative community outcomes among those measured. Caution is warranted, however. First, the available evaluations concentrate on relatively short-term effects, leaving unexamined the possibility of multiyear accretions of community effects. For example, it may take much longer for the informal community networks of a geographic area to embed a cumulative and incrementally created positive or negative perspective that could exert a powerful indirect influence over how residents evaluate their recent experiences with the police (Gau and Brunson, 2010; Rosenbaum et al., 2005; Weitzer and Tuch, 2002). Also unexamined are several important forms of community reaction going beyond attitudes toward the police, such as legal cynicism (Desmond and Valdez, 2013) and crime reporting (Desmond, Papachristos, and Kirk, 2016). Also, possible jurisdictionwide effects are rarely examined.

One concern about the rarity of observed negative community outcomes from place-based proactive strategies is that they could be concentrated in places where the police–community climate is already so negative at the outset that the strategies have little margin to make matters worse (i.e., to have a backfire effect). The committee was able to review the pretreatment outcome levels for four of the six studies showing null effects. In all of these studies, the pre-treatment outcome levels fell in the middle to positive side of the outcome scales, allaying concerns that in these communities the state of police–community affairs was so bad they could not be made worse by a place-based proactive strategy.

The committee notes that the evaluations report little about what police officers in these programs actually did. There is a general absence of a detailed, systematic monitoring of the interventions that are most likely to affect community reactions, especially those pertaining to citizen satisfaction and perceived police legitimacy. For example, most evaluations only describe the training protocol and report the amount of time spent in the hot spot or a count of incidents handled. But citizens likely react to more than mere police presence. They care about the risk of being stopped, questioned, and searched, and they care how those activities are executed. If community impacts are a concern, then evaluators need to include assessments of what police organizations did to control police discretion, to limit abuses, and to promote quality service. In effect, the street-level police practices generated by place-based programs (stage 2 in the logic model

depicted in Figure 5-1) are black-boxed (i.e., not examined in the research as reported) so that the study report fails to provide readers with a good grasp of the character of the intervention as the community experienced it.

The logic models proposed for how place-based proactive practices are expected to affect community outcomes are diverse, which complicates interpretation of the pattern of results across studies. Some research seems animated primarily by concerns about the collateral damage that place-based strategies could produce by more effectively concentrating law enforcement efforts in a small geographic space than other approaches, such as the reactive “standard model” of patrol, which shows a weaker link between police resource deployment and where crime and disorder are distributed (Weisburd et al., 2011). The collateral damage approach simply examines whether the public is troubled by the place-based intervention, regardless of the possible causal linkages among different outcomes, such as those displayed in Figure 5-1.

Other research begins with hypotheses of more positive community outcomes, and some of this research does explore causal linkages across various community outcomes. A brief exposition of this rationale is given in Kochel and Weisburd (2017). Place-based directed patrol and problem solving are expected to increase police visibility, increase police-public interactions, increase both negative and positive experiences with police, and increase particular kinds of police activities (more enforcement for directed patrol and more efforts to change routine activities in places targeted for problem solving). The first three of these first-order effects (visibility, police-public interaction, and positive/negative interactions) are not reflected in the logic model displayed in Figure 5-1; they fall between stages 2 and 3 of that model and might be termed “direct experiences of policing.” They in turn are expected to affect public perceptions about police service and conduct (community evaluations), which in turn affect third-order effects of perceived police legitimacy, perceptions of safety, and collective efficacy. The enforcement and problem-solving first-order effects are expected to affect public perceptions about police service and conduct indirectly by the intermediating effects on crime at places of concern. Unfortunately, Kochel and Weisburd (2017) presented all community outcomes as direct effects and did not offer estimations of the strength of intervening process pathways. While this is consistent with the experimental design of their project, estimations of these pathway effects would promote a better understanding of assumptions about place-based effects. However, it is easily conceivable that a reordering of effects could be justified, which is of particular concern for non-experimental studies. For example, engaging in acts of collective efficacy could easily be viewed as a cause of reduced crime levels (Sampson, 2002), as well as a downstream consequence.

A clearer exposition of the causal process by which place-based in-

terventions affect community outcomes—and a focused empirical testing of that process—would be especially helpful in trying to explain why the evaluations of place-based strategies have shown few community effects of any sort. A model for such an exposition is given by Weisburd and colleagues (2015), who develop and evaluate the mechanisms by which broken windows policing is presumed to reduce crime. The authors outline an underlying causal sequence tracing effects from police reducing social and physical disorder to reduction of residents’ fear of crime to increases in community social control to crime reduction. They offer a narrative review and meta-analysis to assess what empirical research shows about this process. They note significant variation in the impact of broken windows policing tactics (some place based and some not) on fear of crime. Of six experimental/quasi-experimental studies, three found no change, two found a significant reduction, and one showed a backfire effect. Only one tested for impact on informal social control, finding no effect. A meta-analysis reinforced the sense that fear reduction and collective efficacy were not attributable to the broken windows tactics, yet the authors cautioned that there were various limitations in the research: for example, the confounding of a broken windows strategy with many other proactive strategies, the failure to measure and model informal social control, and the specification of a theoretically reasonable follow-up time period to assess program impacts (years longer than most available studies).

Summary. There is only a small, emerging body of research evaluating the impact of place-based strategies on community attitudes, including both quasi-experimental and experimental studies. Place-based policing tactics were often co-implemented and integrated with tactics typical of other approaches, such as problem solving, community based, and person focused, making it difficult to know how much of an intervention’s effects were due to its place-based character. The available research is also limited in its focus on outcomes measured as attitudes toward police and on short-term and less-than-jurisdictionwide effects. However, the consistency of the findings of the available studies leads the committee to conclude that place-based policing strategies rarely have negative impacts on short-term, police-focused community outcomes; at the same time, such strategies rarely improve community perceptions of the police or other community outcome measures. Caution about the broad generalizability of this finding is therefore warranted.

PROBLEM-SOLVING INTERVENTIONS

For this report’s purposes, problem-solving interventions have been defined as strategies that try to identify causes of problems, select innovative

solutions (backed by scientific evidence wherever possible), assess the effects of the intervention, and adjust future interventions accordingly. The most prevalent strategy for this approach is problem-oriented policing, but also included is third party policing.⁹ Our analysis focuses on 18 reports that offered some evidence on problem-solving strategies using experimental or quasi-experimental designs (sample compiled mostly from the Evidence-Based Policing Matrix [Lum, Koper, and Telep, 2011]¹⁰ and from Gill et al. [2014]).¹¹ These reports have generated 26 independent tests that assess one or more indicators of community reactions to a problem-solving strategy.

The method for problem solving among the projects studied has been remarkably similar, while the nature of particular problems targeted has been diverse. Most projects designated one or more geographic areas (neighborhoods, beats, precincts, public housing area, microplaces) and leave it to the assigned police and residents of the areas to identify the problem to solve, drawing on some version of the scanning, analysis, response, and assessment (SARA) process. The range of problems targeted for intervention has been wide, although addressing neighborhood social and physical disorder in its various manifestations has been a popular choice. A few studies were launched with a much narrower mandate, such as targeting juveniles with a high risk for delinquency (Weisburd, Morris, and Ready, 2008), architectural design to reduce crime (Armitage and Monchuk, 2011), drug crime in public housing (Giacomazzi, McGarrell, and Thurman, 1998), or juvenile crime in a park (Baker and Wolfer, 2003). We do not know how representative our sample is of the population of problems that are targeted by problem-solving practices in American police agencies in general.

Evaluations of the community reactions to problem solving have concentrated on four types of outcome measures: perceived disorder or quality of life of the respondent, fear of crime or perception of crime risk, satisfaction with the police, and the perceived legitimacy of the police. Simply looking at the statistical significance of study results, respondent satisfaction is the only indicator that shows a positive impact with strong consis-

___________________

⁹ To a lesser extent, proactive partnerships with other organizations (such as code or liquor enforcement agencies, schools, probation, and private businesses) may also be considered as a problem-solving intervention.

¹⁰ The Evidence-Based Policing Matrix focuses on interventions that are “primarily” police interventions; scored a 3 or higher on the Scientific Methods Scale (Sherman et al., 2002) and included at least a well-matched comparison group, multivariate controls, or rigorous time series analysis.

¹¹Armitage and Monchuk (2011); Baker and Wolfer (2003); Bond and Gow (1995); Braga and Bond (2009); Breen (1997); Collins et al. (1999); Giacomazzi, McGarrell, and Thurman (1998); Graziano, Rosenbaum, and Schuck (2014); Jesilow et al. (1998); Kochel and Weisburd (2017); Pate et al. (1986); Segrave and Collins (2005); Chicago Community Policing Evaluation Consortium (1995); Skogan and Hartnett (1997); Tuffin, Morris, and Poole (2006); Ratcliffe et al. (2015); Weisburd, Morris, and Ready (2008); Wycoff and Skogan (1993).

tency across evaluations (14 significantly positive, 4 no significant effect). Virtually all of the others show mixed results with respect to direction of statistically significant effect or any significant effect: perceived disorder/ quality of life (7 significantly positive, 1 significant backfire,¹² and 5 no significant effect), fear of crime (6 positive, 9 no significant effect), and legitimacy (6 significantly positive, 6 no significant effect). Notably only 1 of the 26 evaluations produced a statistically significant backfire effect, and that was only for a single community outcome. The size of intervention effects in these studies tends to be modest or moderate. For example, Skogan and Hartnett (1997, p. 210) matched comparison-group evaluation of five Chicago police districts employing problem-oriented policing in a community-based policing framework yielded an assessment of mostly consistent positive effects that were “not overly dramatic” on citizen satisfaction (a combined index of police responsiveness, demeanor, and effectiveness in dealing with crime).¹³ Similarly, small-to-moderate effect sizes were recorded for reductions of citizen perceptions of gun/drug problems in these districts (Gill et al., 2014, p. 415).¹⁴ A comparable pattern emerged in the six-site matched comparison-group evaluation of “reassurance” policing in the United Kingdom, an intervention that also embedded problem-oriented policing in a community-based approach (Tuffin, Morris, and Poole, 2006; Gill et al., 2014, pp. 416-417).

Only a handful of the evaluations relied on randomized experimental designs. Two (Kochel and Weisburd, 2017; Weisburd, Morris, and Ready, 2008) showed no significant effects on the four indicators commonly studied in quasi-experimental evaluations: perceived disorder, fear of crime, citizen satisfaction, and perceived police legitimacy. One evaluation (Graziano, Rosenbaum, and Schuck, 2014) showed significant positive effects on citizen satisfaction and on police legitimacy with the community as perceived by officers.

One RCT (Braga and Bond, 2009) of a problem-oriented practice, embedded in a place-based policing strategy, showed consistent pretest-posttest improvements across a range of community outcomes (see detailed discussion of this study above, in the section on place-based interventions). However, statistically significant positive effects were observed only on perceived social and physical disorder and on frequency of contact with

___________________

¹² This was a comparison of a single treatment in a single neighborhood and control area in a small Connecticut city, which generated a large, significant backfire finding, a distinct outlier in a body of studies that showed much smaller effects. Given the small number of observations in this study, this finding should be interpreted with caution.

¹³ The odds ratios calculated for these interventions were in the small to very-small range (Gill et al., 2014, p. 416).

¹⁴ All of the five sites recorded odds ratios in the “small” range (below 2.5). See Lipsey and Wilson (2001) on interpreting effect sizes.

police; no statistically significant changes were found in perceptions of policing styles and strategies, demeanor toward citizens, police willingness to collaborate with the public on crime and disorder control, and fear of crime. Effect sizes fell in the small range (odds ratios below 2.5), except for the number of contacts with police, which showed a large effect (odds ratio well above 4.3). This study was especially noteworthy for its employment of surveys of “key community residents” who were in a good position to know and shape what was going on in the studied hot spot in which they resided or worked (e.g., an apartment complex manager).¹⁵

One RCT found an increase in residents’ sense of efficacy in problem solving (Graziano, Rosenbaum, and Schuck, 2014), and the other found a positive trend in collective efficacy over time, but it was not statistically distinguishable from the control condition (Kochel and Weisburd, 2017). Finally, only one RCT assessed the impact of problem-solving efforts on the community’s inclination to cooperate with police, finding a small, but statistically significant increase over the course of 9 months (Kochel and Weisburd, 2017, p. 162).¹⁶ Perhaps the most extensive exploration of problem solving’s effects on collective efficacy and other forms of citizen self-help and supportive behavior toward police is the evaluation of reassurance policing in the United Kingdom, a program that incorporated problem solving as a key element of a community-based policing approach in six jurisdictions (Tuffin, Morris, and Poole, 2006). This matched comparison-group evaluation found little evidence to support an effect for this program on these indicators.¹⁷

It is worth considering why community satisfaction should emerge as a fairly reliable consequence of practices typical of a problem-solving approach but not other outcomes such as perceived disorder, fear of crime, or perceived police legitimacy. One possibility is that problem solving may be perceived to reduce crime sufficiently to satisfy community members with the police effort but still insufficient to reduce fear of crime and perceptions of disorder. Another possibility is that the effect is due not so much to the problem-solving aspect (especially the problem analysis and customizing of the solution to the problem diagnosis) of the intervention as it is to the community outreach aspect that so often is also a feature of this policing approach (see the community-oriented policing section of the next chapter).¹⁸ Nearly all of the interventions evaluated for the problem-solving approach

___________________

¹⁵ The number of cases in treatment and control groups was small (26 each), but the evaluators argued that the careful selection of this small sample of informants increased the power of the statistical analysis to detect effects.

¹⁶ Over the “long-term” (9-month) period, problem solving showed a 6 percent increase in residents’ willingness to cooperate, compared to a 2 percent decline for “standard practice.”

¹⁷ See the “Impacts on Community Behavior” section of Chapter 6 for details.

¹⁸ See Kochel and Weisburd (2017, pp. 165–166) for an exposition of this argument.

in this chapter included one or more elements of community-oriented policing (heightened police–community engagement). It is possible that variation in the execution of the community engagement element of these interventions accounts for the pattern of variation in citizen satisfaction. One might anticipate that the more substantial the community engagement in the problem-solving process, the more likely that police effort and performance will concentrate on community priorities, the better “advertised” the results, and the more positive the community spin on those results among community members when they (or some community representatives) had a hand in the process. This was in fact a key feature of the focused deterrence (“pulling levers”) strategy first introduced in Boston as part of a problem-solving tactic (applied within a person-focused policing approach) to deal with gang violence in the 1990s (Braga, 2001; Kennedy, 1997). Yet researchers have also observed that the norm for community engagement in problem solving tends to be the identification of problems and the assessment of outcomes, not the analysis of those problems or “coproductive” involvement with the community in the intervention strategy itself (Braga and Bond, 2009). Nonetheless, it is possible that community satisfaction from problem-solving experiences is due not to technical success in reducing problems but to the public’s observation of and even limited participation in the process itself, afforded by activities that incorporate tactics from community-oriented policing. In the research to date, these two possible mechanisms are confounded and cannot be isolated for analysis.

Regardless of the resolution of the role of community-oriented policing practices in problem solving’s apparent capacity to satisfy the public, one might still wonder why the other community impact measures did not reflect a similar positive effect with much consistency. The possibilities are numerous. The most straightforward explanation is the absence of evaluations that assess problem-solving interventions that have the reduction of community alienation as the targeted problem, instead of reduction of crime or disorder. In this regard, the true capacity of focusing the SARA process of problem-oriented policing to improve most community outcomes remains untested.

Another explanation is that some effects take longer than others and that they depend on demonstrating the success of certain indicators in a causal chain. Given that the vast majority of these studies relied upon evaluations of effects within a year or two after the intervention’s onset, there may simply have been insufficient time to register effects. One underlying logic model that fits the observed pattern is that community satisfaction with the police is a necessary precursor to community cooperation with successful problem-solving interventions, which will later yield reductions in the targeted problems, and those in turn will reduce fear, increase perceptions of disorder, and enhance perceived legitimacy (Gill et al., 2014). This logic model may be contrasted with one that presupposes that the

community’s judgments of the police are driven by their perceptions of the police’s success in problem solving, which Skogan (2009) termed an “accountability” model and which is consistent with the sequence displayed in Figure 5-1. A third possibility—and the one the committee considers most likely—is that both causal processes are at work simultaneously.

Testing the process of community outcomes to establish causation would require long-term, multiyear studies with extensive longitudinal measurement of community members’ perceptions, attitudes, and behaviors. An example of short-term longitudinal measurement is given in a three-wave RCT evaluation of a St. Louis County, Missouri, project that included a problem-solving component among the interventions compared (Kochel and Weisburd, 2017; see also the description earlier in the “Place-Based Interventions” section of this chapter). This evaluation used a large assortment of community and crime/disorder outcome indicators. Extending the longitudinal impact analysis across waves of much longer duration than 3 months would allow an empirical assessment of the possible underlying causal processes.

Variation in effects across evaluations of problem solving may also reflect variations in the fidelity or intensity of the problem-solving component of the intervention. The challenges of problem-solving implementation are widely acknowledged by researchers and police leaders (Braga and Weisburd, 2006; Braga, 2010; Weisburd et al., 2010) but are typically discussed in the context of evaluating crime and disorder outcomes, not community outcomes. Achieving insufficient rigor in the SARA process is frequently mentioned as a limitation, which yields at best shallow problem solving (i.e., weak problem analysis and constrained or uncreative responses) (Braga, 2010; Braga and Bond, 2009; Braga and Weisburd, 2006). Perhaps shallow problem solving should not be a great surprise when considering the typical low intensity of the organizational efforts to enable and promote these activities. For example, training in problem-solving techniques typically lasts only a few days.

Most studies do not report implementation with enough detail to make comparisons across studies. One that does go into considerable depth suggests the complexity of measuring problem-solving implementation, which could involve the following aspects of the program: problem identification, development and implementation of solutions, community organization involvement, involvement of other government and private organizations, and police involvement (Skogan and Hartnett, 1997, pp.184–193). As we note in the community-based policing discussion in Chapter 6, the onsite evaluators gave the Chicago Police Department’s problem solving a grade of C, illustrating how difficult it can be to align all of the strategy’s elements on a citywide scale (Skogan and Steiner, 2004, pp. viii–x). Another onsite process analysis of the National Reassurance project in the United Kingdom

went a step further and found a correlation across six sites between the extent of problem-solving implementation (community involvement, specification and delineation of the problem’s nature) and one of the community outcomes: public perceptions of juvenile-caused nuisances (Tuffin, Morris, and Poole, 2006, pp. 80–82). It would be especially helpful for comparing the effects of problem-solving efforts to develop a comprehensive rating system for determining the extent of implementation, notwithstanding the challenges of rating so diverse a set of interventions (Eck, 2006).

While accounting for the implementation of interventions in a “mediation” analysis is an important consideration in evaluating the technical crime-control efficacy of a problem-solving approach (Braga, 2010, p. 176), there is a second aspect that is probably far more relevant to assessing community reactions. This aspect concerns the tactics and strategies actually employed (Braga and Bond, 2009; Braga, Hureau, and Papachristos, 2014). There is a world of difference in how a group of juveniles and their family/friends will likely react to stop, question, and frisk (SQF) compared to midnight basketball. Hence, researchers’ capacity to predict community outcomes will be heightened to the extent that evaluators take into account differences in the tactics selected for an intervention and differences in the efforts by the police to achieve community acceptance of those tactics. No such analysis is currently available for community effects evaluations.

Evaluations of problem solving are concentrated in large urban communities. About two-thirds are in American communities, and the remainder are in the United Kingdom or Australia. Chicago, the third most populous American city, accounts for 23 percent of the evaluations, although 27 percent of the evaluations were conducted in communities of under 115,000 population. No clear differences in community reactions to problem solving have been reported across these geographic and demographic ranges.

Summary. The available evidence on the short-term community outcomes of interventions using a problem-solving approach shows an intriguing and somewhat encouraging pattern. (There is little evidence available on the long-term impacts of problem-solving strategies on community outcomes or on jurisdictionwide impacts.) Most of the quasi-experimental or experimental evaluations of community satisfaction register small or moderate, positive short-term effects, while other community outcome measures show at best mixed findings. There is no obvious single factor to account for this variation. The virtual absence of backfire effects should reassure practitioners that problem-solving tactics have not obviously undermined

police-community relations.¹⁹ The principal challenge here is knowing what to do with these findings, since there are a number of possible explanations, one of them being that positive community effects derive primarily from the processes of community engagement, which are virtually always a part of the interventions that have been evaluated, and not from the reduction of the targeted problems.

PERSON-FOCUSED INTERVENTIONS

Person-focused strategies attempt to capitalize on the strong concentration of crime among a small portion of the criminal population. Two types of community outcomes seem relevant to person-focused interventions (Shaw, 1995, p. 708). First, there is interest in how the targeted offenders or offender groups (e.g., gang members in a focused deterrence intervention) react to this strategy, not only in terms of the degree to which they are ultimately deterred from crime but also in how they evaluate their experience with the police and its consequences, especially the procedural justice–perceived police legitimacy linkage. These are involuntary “clients” of the police. Knowledge of their reactions would help researchers establish the extent to which alternative mechanisms to deterrence, such as procedural justice, play a role in mitigating negative outcomes and promoting positive ones. Second, there is interest in the effects on the broader community in which any person-focused intervention is implemented. Here the interest is in the community as “citizenry” with a stake in how their society is policed, with “community” usually defined operationally as the residents of a study area. A focused deterrence strategy in particular attempts to secure broad community support for the interventions, involve members of the community in the intervention, and thereby secure acceptance of the fairness and ultimate perceived legitimacy of a process that includes a highly targeted punishment element. How does the broader community feel about the focused deterrence process? How have their feelings about the police been affected? What were their views about the quality of life in the community as a consequence of the focused deterrence intervention?

Evaluations need to be conducted from the perspectives of both the offenders being targeted and the larger community. It is important to distinguish them because the experiences and perspectives of the two groups may be strikingly different (Braga, Hureau, and Papachristos, 2014). One would expect that if the person-focused program is successful in concentrating police enforcement interventions on a particular targeted group,

___________________

¹⁹ This must be qualified by the notation that evaluations of problem solving have virtually ignored certain collateral effects measures, such as physical and mental health, employment, and legal cynicism. See the “Collateral Consequences” section of this chapter below.

then the people experiencing that intervention first-hand will show greater negativity about the police than those toward whom the police enforcement practices are not concentrated. On the other hand, the general (e.g., residential) population in areas experiencing person-focused policing, who are the presumed primary beneficiaries of that program, may be more inclined to evaluate the results positively—if they are aware of the intervention.

Unfortunately, empirical research on person-focused interventions has concentrated heavily on crime control outcomes and has largely left community outcomes unexamined. The absence of experimental evaluations of community outcomes of person-focused strategies is also noteworthy. And notably, the available empirical research looks at SQF, traffic stops, and repeat offender practices, but the committee could locate no research assessing the impact of the focused deterrence strategy on community outcomes.

Several studies relying on correlational analysis of cross-sectional data and qualitative field observation show, with consistency, the expected negative correlation formed by citizens who experience SQF and aggressive traffic enforcement. The Police–Public Contact Survey of 2011 provides broad insights into the scope of police actions nationwide and their relationship to perceptions of citizens who experience those actions (Langton and Durose, 2013, p. 3). Less than 1 percent of 241 million U.S. residents ages 16 and older reported experiencing a street stop (not in a moving motor vehicle) as their most recent contact with police in 2011. Ten percent of 212 million drivers ages 16 and older reported being stopped while operating a motor vehicle during that period as their most recent contact with police. Twenty-nine percent of respondents subjected to street stops felt that police had not behaved properly, while 12 percent of stopped drivers made that assessment of their experience. Thus, although large majorities of those stopped did not find police actions improper, substantial numbers of citizens across the nation had formed negative judgments about the propriety of police actions. Only 3.5 percent of drivers stopped by police received a personal or vehicular search, but 39 percent of those searched felt that police had not behaved properly, while only 11 percent of those stopped but not searched felt that way (Langton and Durose, 2013, p. 9). The survey found that the likelihood of Black drivers being stopped was significantly higher than for Whites and Hispanics and that there were no statistically significant differences by race/Hispanic origin for street stops. Other studies have offered more in-depth analyses of high-risk populations defined by race, gender, and age.

A study of 45 young Black males (13–19 years old) living in disadvantaged St. Louis, Missouri, neighborhoods used a survey and in-depth interviews to learn their impressions and reactions to the policing they and others received in their neighborhoods (Gau and Brunson, 2010). Nearly 8 in 10 respondents had been stopped at least once during their lifetime,

the average number of stops being nearly 16. This study and other analyses of the same data (Brunson, 2007) paint a picture of Black youths who perceived police order-maintenance practices in their predominantly Black neighborhoods as frequently experiencing police stops as harassment (about 8 in 10) and knowing someone who was harassed or mistreated (about 9 in 10), the most common complaints being harsh, illegal, and disrespectful police treatment (Brunson, 2007, pp. 78–95). Two-thirds of respondents indicated that police were not easy to talk to, and a frequent theme was that the police gave their neighborhoods low-quality service (slow response times and ineffective crime prevention and case solving). While acknowledging the need for police to deal with crime and disorder, respondents felt that officers were too narrowly focused on drugs and gangs, with insufficient attention to other problems, especially the needs of crime victims.

Two types of perceived police misconduct strongly shaped respondents’ negative views toward police: being stopped with insufficient evidence and police violence or threat of violence in excess of what circumstances required (experienced directly and vicariously through second-hand accounts of family, friends, and neighbors). For both types, respondents were turned off by the failure of police to conform their practices to the requirements of law. And in the first type of misconduct, respondents were especially frustrated by the irrelevance of their own adherence to the law to inoculate them from unwarranted police attention. Finally, an especially disliked practice was when officers who were frustrated by failing to find evidence to support an arrest drove the respondents to a hostile or unfamiliar neighborhood and released them to get home on their own, knowing that this put the youths’ safety at great risk. Many respondents attributed the concentration of these policing practices in their neighborhood to their being predominantly Black and disadvantaged.

A large, cross-sectional survey of young persons in New York City found similarly negative associations of respondents’ perceptions of SQF experiences. Tyler, Fagan, and Geller (2014) found a strong inverse correlation between the number of stops experienced or observed by young people in New York City and the legitimacy they accord the police.²⁰ In their analysis it is neither the frequency nor amount of intrusiveness of the stops that strongly affects these feelings but rather the lawfulness and fairness they perceived to have occurred during those stops, similar to the outrage expressed by adolescent Blacks in distressed St. Louis, Missouri, neighborhoods. Though this pattern suggests there might be a “right” way to conduct stops that minimizes the risk of negatively affecting perceived

___________________

²⁰ See Fratello et al. (2013) for a description of similar findings from a randomly drawn street sample of 474 young people (ages 18–25) at risk for SQF experiences in several hot spots of SQF policing in New York City in 2011.

legitimacy, the researchers also found that repeated-stop experiences of the same person were associated with declines in perceived legitimacy over time, irrespective of how people were treated. In New York City during the time of this study, the overwhelming majority of stops were of young people who were not engaged in criminal activity at the time they were stopped. Hence, it is easy to see how a person repeatedly stopped while innocent would over time come to view the police as acting unfairly and inefficiently.

A qualitative study of involuntary encounters with the police in the Kansas City, Missouri, metropolitan area focused on traffic stops and found different results depending upon the reason for the stop—as perceived by the driver (Epp, Maynard-Moody, and Haider-Markel, 2014). Those perceived as traffic safety stops (e.g., speeding, traffic light, driving under the influence) were distinguished from investigatory stops (where the officer was looking to acquire evidence of a more serious criminal offense). Motorists inferred from the officer’s stated reason for the stop which sort of situation they were encountering. Safety stops were inferred when officers stated a safety offense; investigatory stops were inferred when the officer gave a reason that was a minor violation (license plate light out, turning too wide, driving too slow) or offered no reason at all.

Black motorists had a higher probability of being subject to presumed investigatory stop than White drivers, whereas there was a general absence of race effects for the presumed traffic safety stops. Black drivers and White drivers indicated that they experienced similar levels of impolite demeanor during traffic stops, but Black drivers were much more likely than Whites to report impolite police behavior during investigatory stops, and they were less likely to accept as legitimate the officer’s decision to pull them over. The researchers noted that it was not only the difference in treatment shown by the police during the stop that mattered here but also (as with previously reviewed studies) the feeling that there was no justifiable reason for the investigatory stop.

Some studies conducted in the United Kingdom point to the negative effect on ratings of the police when someone has experienced a police search (see Miller and D’Souza, 2016, for a review). Searches in general (Miller, Bland, and Quinton, 2000; Skogan, 1994), and pedestrian searches in particular (Clancy et al., 2001), are associated with lower levels of satisfaction with and confidence in the police. While these studies do not measure the effect of proactive policing as the product of a strategy, they offer a broader empirical base to generalize beyond studies of policing in the United States.

The above studies point to a consistently pronounced negative association of citizens’ experiences with SQF and traffic (investigatory) stops with assessments of the police. However, because they are just correlational and qualitative studies, they have limited capacity to support causal inferences about the contribution of these person-focused practices to the views of

those who experience them. For example, research suggests that evaluations of the police in specific situations are strongly influenced by the broader orientations that citizens bring to those encounters; in fact, much more so than the influence of an individual encounter on the citizen’s general orientation to the police (Brandl et al., 1994; Worden and McLean, 2017b). Furthermore, these broader orientations are not merely a summation of the consequences of the citizen’s past personal experiences with the police but are shaped by a variety of socialization sources among friends, family, coworkers, etc. Not taking these “global” orientations toward police into account risks overstating the contribution to individuals’ perception of a given experience with a person-focused encounter.

A few studies examine the effects of person-focused interventions on the residential population at large, and they do not show negative effects. A recent quasi-experiment explored the impact of a person-focused (repeat offenders) practice imbedded in a hot spots policing intervention in Philadelphia (Ratcliffe et al., 2015). The police developed a list of active repeat offenders living or operating in the treatment areas. Officers had discretion in selecting interventions, which could include merely talking to the offenders, performing field interrogations, or serving criminal warrants. Researchers reported that officers actively pursued this focused offender tactic over an 8-month period. The study found that residents living in areas exposed to that strategy had no statistically different ratings of seven community outcomes than did those in the control areas (see above discussion of this study in the place-based interventions section).

Another study examined the impact on public confidence in the police from both direct personal exposure to SQF and from indirect sources (e.g., coming from second-hand accounts and general impressions or transmitted by word of mouth) (Miller and D’Souza, 2016). The researchers asked whether SQF experiences in a given geographic area affected the attitudes of residents besides those immediately involved in these events, a useful question because most people rarely have contact of this sort with the police. The study employed a multilevel longitudinal multiple regression analysis of the survey responses of nearly 108,000 London residents in 32 boroughs between 2006 and 2013. The first analysis refers to the direct impact of first-hand SQF experience compared to those who have not had such an experience in the last year. The second analysis refers to the impact of the SQF rate in the entire area (borough) in which the respondent resided. Consistent with prior research, respondents who had been stopped in the last year or searched or arrested in the last year were less likely to rate the police as fair/responsive or effective (difference was statistically significant). These effects were larger than those of various personal characteristics (e.g., age, socioeconomic status, ethnicity) but not as large as the visibility of foot or bike police patrol to respondents, which showed the expected positive

relationship to confidence in the police. Controlling for respondents’ personal SQF experiences, personal characteristics, and crime/disorder rates in the borough, the analysis also examined the impact of borough-level exposure to two sorts of SQF search rates determined from police records (n = 224 borough-months). One type of search was covered by the Police and Criminal Evidence Act of 1984 (PACE), governing searches that require “reasonable grounds” for suspicion. The other type of search was covered by Section 60 of the 1994 act: searches not requiring such grounds for suspicion (e.g., weapons searches at sporting events). Using both lagged and unlagged estimates, there was only one statistically significant effect: for Section 60 searches on perceptions of police effectiveness (a positive correlation statistically significant for lagged effects only).²¹ There was some indication of variable effects across subgroups of the respondents. Blacks and persons with low socioeconomic status perceived lower levels of fairness and responsiveness where Section 60 search rates were higher and at the same time were more positive in their perceptions of police effectiveness under those conditions (both differences were statistically significant). However, the meaning of this pattern is not entirely clear, since the statistically significant effects were lagged for the effectiveness dependent variable (hence taking longer to show an effect) and unlagged for the fairness and responsiveness dependent variable. The authors accounted for this pattern by speculating that “negative stories about searches move quicker through social networks than positive stories or that the interpretation becomes more positive with the lapse of time” (Miller and D’Souza, 2016, p. 472).

Two features of this analysis of Black and low-socioeconomic status subgroups are particularly noteworthy. When one outlier borough that had very high and variable search rates was excluded, the effects were no longer statistically significant. Also, the size of these effects was quite small. The average effect of Section 60 searches on Black respondents’ perceptions of police fairness and responsiveness was only 0.024 points on a 4-point scale, with a standard deviation of 0.6. Effects were even smaller for respondents of low socioeconomic status. Hence, the evidence from this study of wider effects of SQF, excluding the study’s results from direct exposure to SQF, does not support a finding that this SQF proactive strategy generated a strong reaction in the general public. Finally, the authors acknowledged that boroughs may be too large a unit of analysis to capture variation in

___________________

²¹ The researchers speculated on reasons for the absence of effects for PACE searches (Miller and D’Souza, 2016, p. 472). One possibility is that because Section 60 searches are executed in specific locations, they may yield a greater sense of safety for some residents but not others. Another explanation is that PACE searches, which require reasonable grounds, may generate weaker positive or negative effects than Section 60 searches, which require no grounds. Finally, the PACE searches displayed much less annual variation than Section 60 searches, which may make it more difficult for residents to perceive changes in PACE search levels.

SQF rates that is meaningful to residents and that the high daily mobility of urban dwellers, coupled with heightened use of social media, may simply render geographic-based exposure to SQF less relevant.

The pattern among this handful of studies is suggestive but insufficient to state with confidence what the effects of person-focused strategies are on community views. All of the four studies looking at the impact of person-focused strategies on those directly targeted (one focusing on SQF of Black youths living in high-crime St. Louis neighborhoods, one of SQF in New York City, one of traffic stops in Kansas City, Missouri, and one of SQF searches in London) illuminate consistent and fairly strong negative reactions. Only two studies examined the larger residential (stakeholder) community’s response to these practices. A quasi-experiment in Philadelphia found no community effects. The London study found that the general community response depended upon the citizen’s race, but this association disappeared when a single outlier geographic area was dropped from the analysis.

One might infer from these results that communities at large have no predictable reaction to person-focused interventions. But two studies are a precarious basis for drawing such a conclusion at this point. Especially important for future research is to compare the profile of personal characteristics of the targeted offenders to that of the larger community of stakeholders. In that context, much more needs to be known about the public’s tolerance/enthusiasm for such interventions. Here are propositions worth testing from this small body of research. The more indiscriminately and intensively person-focused proactive policing is practiced in a given community area, one presumes the smaller the population of those residents who embrace or tolerate such practices. Regardless of their own personal experiences, the more that residents of a given community group feel at risk for person-focused proactivity by virtue of their personal characteristics (e.g., race), the less likely they are to accept those practices and the police who practice them. However, the negative impact of this practice may be mitigated by police taking care to clearly articulate the legitimacy of each stop (e.g., reasonable suspicion; see the discussion of procedural justice in Chapter 6).

As with place-based and problem-solving strategies, empirical research on person-focused strategies would benefit from illumination of the community elements that may accompany those person-focused strategies. For example, while much has been made of the importance of community leaders’ involvement and support of focused deterrence strategies (Braga, 2001; Kennedy, 1997), it is not known how important that is in promoting broad community acceptance of these strategies, especially for those community members most proximate or similar to those who are actually targeted for proactive intervention. Further, there is considerable variance in the sorts of law enforcement and community activities employed. How much do

those differences affect community reactions, and does that in turn depend upon which segments of the community are at greatest risk of being the targets of the person-focused tactics? And, as noted previously for place-based and problem-solving strategies, the absence of measures of long-term (multiyear) effects of person-focused strategies means that the evidence base lacks results from tests of the potential cumulative impact of such methods on community outcomes. For example, the cumulative consequences of individuals’ exposure to SQF may couple the intensification of broader community hostility toward the police resulting from diffusion of negative views through informal community networks. The absence of empirical research on these issues precludes offering evidence-based answers to such questions. However, given the positive findings regarding the impact of focused deterrence strategies on crime (see Chapter 4), tracking community effects seems an especially worthwhile endeavor for future study.

Summary. The body of research exploring the impact of person-focused strategies on community outcomes is relatively small, even compared to the evidence for problem-solving and place-based strategies. There are only a handful of studies on interventions that primarily used SQF, traffic stops, or repeat offender practices; there are none on focused deterrence interventions. Most of the studies involve qualitative or correlational designs, making it hard to draw causal inferences. In evaluating person-focused strategies, it is important to distinguish effects on the person targeted for these interventions (suspects and offenders) from effects on the larger stakeholder community that is intended to benefit from the intervention. The studies that measure impacts on targeted persons all show marked negative associations between experiencing a given strategy and the attitudes and orientations of those who experienced the interventions. The studies that measure the impact on the larger community do not indicate a clear pattern of outcomes. The long-term and jurisdictionwide community consequences of person-focused proactive strategies remain untested.

COLLATERAL CONSEQUENCES FOR SOCIETY OF PROACTIVE POLICING

As discussed above (and in the next chapter), much of the literature assessing community reactions to proactive policing strategies focuses on community evaluations, orientations, and behavior toward the police. However, an emerging body of literature examines the indirect, or collateral, consequences of proactive policing practices on community characteristics such as public health and civic and institutional engagement. Although most of this literature is correlational, it nevertheless raises important questions regarding the impact of proactive policing policies on communities.

Moreover, much of this literature examines the implications of such practices for poor and non-White communities. Because of the concentration of proactive policing tactics in these neighborhoods, the literature stresses the importance of exploring these effects further, as these collateral consequences may be especially salient for particular neighborhoods and communities. We note that a large literature exists in the criminal justice field that assesses the consequences of arrest, imprisonment, and other criminal justice contact for health and mental health; employment and earnings; and families, communities, and society writ large. For example, incarceration is strongly correlated with negative social, economic, and health outcomes not only for prisoners and former prisoners but also for their families (National Research Council, 2014). Moreover, a number of collateral consequences have been documented for people who are arrested and convicted of crimes but not incarcerated. These consequences include effects on people’s employment and business opportunities and on access to government benefits, including student loans and housing (see, e.g., Colgate-Love, Roberts, and Klingele, 2013). To the extent that proactive policing practices foster criminal justice contact and involvement, such consequences may also be said to derive indirectly from proactive policing.

Impact of Proactive Policing Practices on Health and Development

As noted above, many scholars have suggested that the criminal justice system adversely affects physical and mental health (see, e.g., Golembeski and Fullilove, 2005; Johnson and Raphael, 2009; Western, 2006). Less well studied is the specific effect of proactive policing practices on health and development. However, an emerging public health literature suggests that involuntary police contact may threaten the health of individuals stopped by the police—for instance, in SQF stops. The adverse health effects may arise from the physical nature of some stops, which present risks of physical injury; from emotional trauma associated with unwarranted accusations of wrongdoing; and from contacts associated with racism, which may cause stigma and stress responses and depressive symptoms (see, e.g., Bylander, 2015; Garcia and Sharif, 2015; Nordberg et al., 2015; Shedd, 2015). Alternatively, people targeted for such interventions may also be at greater risk for health problems due to “third variable” vulnerabilities, such as limited wealth and education. On the other hand, much of this literature also acknowledges that policing may improve individual and population health by improving public safety and building feelings of security.

Geller and colleagues (2014) conducted a population-based survey of young men in New York City in 2012–2013 to understand the extent and nature of their experiences with the New York City Police Department’s SQF tactics and the association between these contacts and dimensions

of respondents’ mental health. The survey found that young men who reported police contact also reported higher anxiety scores (controlling for demographic characteristics and criminal involvement). Anxiety symptoms were correlated to the number of times the young men were stopped and to how they perceived the encounter, and these correlations were statistically significant. Those respondents who reported more police intrusion also tended to have greater anxiety levels. The statistical associations between respondents’ experiences with the police and their mental health were strong and largely robust (consistent) across samples and models—particularly among respondents reporting stops carried out in an intrusive fashion. Geller and colleagues (2014) concluded that such associations between police intrusion and mental health, as observed in a population-based sample of young men reporting high rates of contact with the police, raise public health concerns for the individuals and communities most aggressively targeted by the police. However, the authors are careful to note that the cross-sectional nature of the data analyzed does not lend itself to causal claims and that the causal direction of the relationship is in fact uncertain. That is, it is possible that the respondents’ mental health influenced their perception of their interactions with the police and that those prone to the greatest anxiety and stress tended to exaggerate their experiences, or that respondents displaying mental health symptoms might have attracted greater reasonable suspicion or may have responded to police questions in ways that escalated their encounters.

Another study examined the impact of proactive policing practices on adolescent development. Jones (2014, p. 36) argued that targeted policing practices do more than shape young men’s perceptions of the police; they also shape their “life space, affecting what they do, where, and with whom.” This ethnographic study of adult and adolescent Black men in a San Francisco neighborhood, where police implemented problem-oriented policing interventions along with other targeted law enforcement practices, found that routine exposure to proactive policing practices had the potential to influence normative adolescent development. Pointing to areas where future research is needed, Jones (2014) questioned whether the penetration of police practices into young men’s peer and family networks gives neighborhood youth a criminalized identity, keeping them linked to the juvenile or criminal justice system and to peers who are more deeply committed to delinquency or criminal behavior. The question, in other words, is whether policing practices make it more difficult for young people to drift out of delinquency.

In addition, a study by Desmond and Valdez (2013) focused on the harm to the urban poor associated with the application of coercive third party policing strategies. The authors examined the use of Milwaukee’s nuisance-property ordinance that charged or threatened landlords with a

substantial fine for repeated tenant behavior that police authorities deemed to be a nuisance (e.g., noise, domestic violence, frivolous use of 911, family trouble). Analyzing all nuisance property notifications (identifying a property as a nuisance, the property owner’s abatement response, and the police response to the owner’s response) and all nuisance-eligible properties over a 2-year period, and controlling for a variety of neighborhood socioeconomic factors, property code violations, and the crime rate, researchers found that properties in predominantly Black neighborhoods were disproportionately likely to get cited for nuisances and those located in integrated Black neighborhoods were most likely to be judged a nuisance. A substantial portion of citations (almost one-third) for nuisance incidents were based on domestic violence, and the typical property owner response was eviction, a response encouraged by the police who reviewed landlord responses to threats to punish them. In addition, landlords threatened and discouraged tenants from summoning police assistance and instead encouraged them to solve the problem themselves (e.g., by making the abusive party in the domestic relationship move out) or to refer the problem to the landlord or another nonpolice entity. Most of the landlords who received a nuisance citation for domestic violence responded with efforts to evict, either formally or informally, or they threatened eviction if the tenant summoned the police again. The researchers noted, “the nuisance property ordinance has the effect of forcing abused women to choose between calling the police on their abusers (only to risk eviction) or staying in their apartments (only to risk more abuse)” (Desmond and Valdez, 2013, p. 137).

Noting that Milwaukee’s ordinance was similar to ordinances of other American cities, the authors argued that this coercive form of third party policing was implicated in “the reproduction of racial, economic, and gender inequalities” (Desmond and Valdez, 2013, p. 137) that disproportionately exposes women in poor Black neighborhoods to higher rates of eviction, which in turn causes homelessness, loss of wealth, residential instability, unemployment, and a variety of mental health problems. But because this descriptive study lacked a comparison to the distribution of harms where no such coercive third party policing program was present, the results do not provide evidence confirming a causal impact of third party nuisance abatement programs on these important societal outcomes.

Similarly, other scholars have considered the health impacts of racism and perceived racism on individuals and communities. This literature is important because much of the discussion surrounding proactive policing strategies has focused on its targeting of non-White communities. The public health literature indicates that racism as a social condition is a cause of health and illness (Link and Phelan, 1995; Ford and Airhihenbuwa, 2010; Gee and Ford, 2011; Jones, 2001; Williams and Mohammed, 2013; Brondolo et al., 2009; Dressler, Oths, and Gravlee, 2005). For example,

Lewis and colleagues (2006) found a positive association between discrimination and coronary artery calcification in Black women, while McLaughlin, Hatzenbuelher, and Keyes (2010) found that experiences of discrimination were correlated with elevated levels of psychological distress. Therefore, if proactive policing strategies are implemented in a discriminatory fashion or are perceived to be discriminatory (see Chapter 7 of this report), there may be resulting public health consequences for the communities experiencing that discrimination or perceiving policing activity to be discriminatory.

Summary. The committee concluded that existing studies of the collateral consequences of more aggressive policing styles are informative in suggesting the importance of focusing more research on potential public health consequences of proactive policing strategies. However, the research to date does not allow the committee to draw evidence-based conclusions regarding these potential consequences. Future studies of proactive policing should include measures that examine potential negative consequences of policing interventions on physical and mental health for both individuals and the communities where such interventions are implemented.

Impact of Proactive Policing on Civic and Institutional Engagement

Another emerging body of literature considers the indirect effect of proactive policing practices on civic and institutional engagement and political life. In this area, scholars have suggested that involvement with the criminal justice system may have an impact on levels of civic engagement. Justice and Meares (2014), for example, hypothesized a link between policing and political life by arguing that the criminal justice system, through encounters between police officers and citizens, educates those it contacts in what it means to be a citizen. This “education,” they suggested, has the potential to incite radicalization, resistance, and solidarity, as well as anger, insecurity, and despair. And Weaver and Lerman (2010) concluded that contact with the criminal justice system is associated with weakened attachment to the political process and increased negative perceptions of government.

In another study, which examined the consequences of SQF tactics on civic engagement by assessing non-emergency calls for service or information requests (“311 calls”) from 2010 to 2011 in New York City, Lerman and Weaver (2014a) showed that the concentration of police stops, at the block-group level, was associated with higher levels of community engagement. However, there was also a negative correlation between the number of stops that featured searches or the use of force, especially if the stop did not result in an arrest, and incidence of neighborhood-level outreach to local government. These results suggest that the nature and perception of policing practices may affect levels of civic engagement in urban communities.

In a subsequent study, Lerman and Weaver (2014b) estimated the magnitude of the relationship between encounters with law enforcement, Americans’ attitudes toward government and democratic values, and their likelihood of voting or engaging in other forms of citizen participation. Using a nationally representative survey and in-person interviews, they assessed the relationship between increased use of SQF tactics and citizens’ attitudes and behavior that were associated with their experiences with police. For example, the authors found that people who had been arrested but never convicted were 16 percent less likely to “feel like a full and equal citizen” and were 20 percent less likely to believe that “everyone in the U.S. has an equal chance to succeed.” Moreover, people who had been stopped and questioned by police or arrested for a crime but never convicted were about 10 percent more likely to express distrust of government. Lerman and Weaver (2014b) concluded that these attitudes contribute to disengagement from the democratic process, an action that is not passive but rather is a conscious effort at non-engagement. That is, these respondents believed non-engagement to be the best strategy against intrusive law enforcement—to intentionally stay invisible, to actively avoid authorities, and to keep a low profile.

Though not focused specifically on proactive policing strategies, Brayne (2014) assessed the impact on institutional engagement of being stopped by the police or of being arrested but not convicted. The study found that individuals who had been stopped by police, arrested, convicted, or incarcerated were less likely to interact with surveilling institutions, including medical, financial, labor market, and education institutions, than their counterparts who had not had criminal justice contact. That is, they exhibited behaviors of “system avoidance”: the practice of individuals avoiding institutions that keep formal records and therefore heighten the risk of surveillance and apprehension by authorities. Using data from the National Longitudinal Study of Adolescent Health and controlling for sociographic, behavioral, and other pertinent factors (e.g., possession of medical insurance), the study found that individuals who had been stopped by the police had 33 percent higher odds of not obtaining medical care when needed and that those who had been arrested (but not convicted) had 29 percent higher odds. The results also showed that arrest (without conviction) or conviction were statistically significant negative predictors of institutional attachment (obtaining medical care, possessing a bank account, and being in school/working), but being stopped by police was not a significant predictor of bank account ownership or of being in school or working. Thus, Brayne (2014) concluded that fear of surveillance and subsequent system avoidance, rather than sociodemographic characteristics or behavioral characteristics, may shape individuals’ behavior and involvement with institutions that are consequential for future outcomes. Most notably, her “difference-

in-differences” identification strategy has a plausibly causal interpretation, as she found no such change in avoidance behavior with regard to social groups that did not keep records, such as attending church or participating with volunteer groups, avoidance behaviors one might expect if people were becoming generally less active in the community, rather than specifically avoiding record-keeping institutions.

Summary. The impact of policing on civic and institutional engagement is an emerging area of study, and the committee did not have an evidence base adequate for drawing conclusions. But the limited number of studies to date do suggest the potential for research to offer insights into the impacts of proactive policing approaches, and policing more generally, on civic participation. Again, we think this area should be a more widely examined subject of research in the future.

CONCLUSION

Throughout this chapter, we have noted the limitations of the existing research base for assessing the community outcomes of proactive policing strategies that have developed primarily as crime-fighting strategies. The modest number of studies assessing the impact of problem-solving, place-based, and especially person-focused proactive strategies on community outcomes calls both for caution at present in drawing conclusions and for more research. These implications are emphasized in Chapter 8. Nonetheless, the extant research does allow the committee to draw several specific, narrow conclusions regarding the impacts of proactive policing approaches that focus on crime control.

Place-Based Proactive Strategies

There is only an emerging body of research evaluating the impact of place-based strategies on community outcomes, including both quasi-experimental and experimental studies. Place-based strategies in the studied interventions were often co-implemented and integrated with tactics typical of other approaches (such as problem solving, community based, and person focused), making it difficult to know how much of the effects were attributable to their place-based character. However, the consistency of the findings of these studies leads the committee to draw the following conclusion:

The committee notes that its conclusion regarding the absence of negative short-term effects on community outcomes is in contrast to a growing narrative that presumes or expects such strategies will have community impacts (see Chapter 1).

Problem-Solving Proactive Strategies

The research literature on community impacts of problem-solving proactive policing interventions is relatively large compared to the other approaches reviewed in this chapter. Much of this literature relies on quasi-experimental designs. However, a few well-implemented randomized experiments also provide information on community outcomes. Because problem-solving strategies are so often implemented in tandem with practices typical of community-based policing (i.e., community engagement), it is difficult to determine what role the problem-solving aspect plays in community outcomes, relative to the impact of the community engagement practices in the intervention. Nevertheless, the committee was able to draw the following conclusions:

Person-Focused Proactive Strategies

The body of research evaluating the impact of person-focused interventions on community outcomes is relatively small, even when compared to the evidence base for problem-solving and place-based strategies. There are a handful of studies on SQF, traffic stops, and repeat offenders but none on focused deterrence. Most of the studies involve qualitative or correlational designs, making it hard to draw causal inferences. In evaluating

person-focused strategies, it is important to distinguish effects on the person targeted for these interventions (suspects and offenders) from effects on the larger stakeholder community that is intended to benefit from the interventions. The studies that measure impacts on targeted persons all show marked negative associations between exposure to the strategy and the attitudes and orientations of those who experienced the interventions. The studies that measure the impact on the larger community show a more complicated pattern, but overall do not indicate a clear pattern of outcomes.