Community policing and youth offending: a comparison of large and small jurisdictions in the United States

ABSTRACT

Over the last two decades, the implementation of community-oriented practices has been a primary focus of American policing. Supporters of this approach suggest that a partnership between police, community members, and service agencies can strengthen the welfare of young individuals in the community. Relying on Census, Uniform Crime Report, and Law Enforcement Management and Administrative Statistics data, the current study aims to identify the association between community policing practices and arrests of young individuals for violent offences. Findings suggest that the relationship between violent crime arrest counts and implementation of COP strategies is relative to the size of the jurisdiction and age group being examined. These effects are inversely related to violent crime arrest counts among juveniles and young adults in large jurisdictions, suggesting the implementation of COP is successful in reducing levels of violence in urban settings. Based on these results, policy implications and future directions for research are discussed.

KEYWORDS:

• Community policing
• delinquency
• violent offending

Introduction

Over the last two decades, the implementation of community-oriented practices has been a primary focus of American policing. Supporters of this approach suggest that a partnership between police, community members, and service agencies can strengthen the welfare of vulnerable groups in the community, such as youth (Cordner, 2015). Young individuals, however, are often seen as ‘targets’ rather than ‘assets’ to this policing model (Forman, 2004, pp. 2; see also Ilan, 2018). Due to concerns regarding the incidence and seriousness of offending among young individuals (Puzzanchera, 2014) and their increased likelihood of coming in contact with the police (Brick, Taylor, & Esbensen, 2009) there is a need to assess the effectiveness of community-oriented policing in reducing violent crimes perpetrated by juveniles and young adults. Although extensive research has been conducted on community-oriented policing (COP), much remains to be known regarding the impact of COP strategies on specific age groups. While extant literature has largely focused on overall crime rates, most of these studies have overlooked the potential impact of COP on offending among young individuals.

Disaggregation by age may demonstrate variation in the relative effectiveness of COP strategies. Furthermore, previous studies have maintained focus on the effects of COP in urban areas to the neglect of comparatively small jurisdictions (National Institute of Justice, 2004; Rukus, Warner, & Zhang, 2018; Sozer & Merlo, 2013). The lack of information regarding the effectiveness of COP in small jurisdictions is concerning given that most local police departments serve smaller populations (Pelfrey, 2007; Reaves, 2015). Assessing (in)equality in the effects of COP strategies across jurisdictions of varying size, therefore, represents an equally important problem in criminal justice research. To address these gaps in research, the current study advances the understanding of the effectiveness of COP strategies by comparing and contrasting their relative influence, and those of potentially confounding community and department characteristics, on counts of violent offences committed by juveniles and young adults in samples of large (n = 364) and small police jurisdictions (n = 626). We accomplish this through the application of negative binomial regression analyses of data procured from the Law Enforcement Management and Administrative Statistics (LEMAS), American Community Survey (ACS), and Uniform Crime Report (UCR). Our findings reveal several important distinctions in the relative effectiveness of COP strategies on violent offences committed by juveniles and young adults across jurisdictions. This provides both a context and foundation for additional research on the relationship between COP and crime.

Review of the literature

As described by the U.S. Department of Justice, ‘community policing is a philosophy that promotes organizational strategies that support the systematic use of partnerships and problem-solving techniques to proactively address the immediate conditions that give rise to public safety issues such as crime, social disorder, and fear of crime’ (COPS, Office of Community Oriented Policing Services, 2014a). As COP enjoys increasing consideration as an integral component of professional policing (Maguire & Mastrofski, 2000), significant resources have been allocated to the implementation of COP strategies (Maguire & King, 2004). Since 1994, the Office of Community Oriented Policing Services (COPS Office) has allocated nearly $14 billion for the purposes of increasing the number of community-policing officers, improving technology, supporting crime prevention initiatives, and providing training and technical support to the development of community policing (COPS, Office of Community Oriented Policing Services, 2014b). The vast majority of the nation’s law enforcement agencies have reported engaging in at least some form of COP activity (Hickman & Reaves, 2006). By the end of the 2013 fiscal year, the COPS Office had funded the hire of approximately 126,000 additional officers for more than 13,000 law enforcement agencies across the nation (COPS, Office of Community Oriented Policing Services, 2014b).

COP and contemporary social problems

Contemporary events have highlighted the need for increased implementation of COP principles – as evidenced by the well-publicized conflicts that exist between many police agencies, particularly in urban areas, and the communities they serve. As a result, federal officials have recognized the importance of supporting and strengthening the ties between the public and law enforcement agencies (The White House, 2014). The U.S. Department of Justice’s investigation of the Ferguson Police Department (FPD), for example, revealed that the lack of proper implementation of community policing principles had contributed to distrust and damaged relationships between the community and police. Specifically, a ‘systemic problem’ regarding reluctance among young individuals to work with the FPD was recognized (Civil Rights Division, 2015, p. 80). To amend Ferguson’s police practices and restore community trust in law enforcement, the Justice Department’s investigators called for increased officer training related to mechanisms of community policing.

Despite such efforts, consensus is lacking on whether COP strategies are effective in reducing crime, which serves as a long-term goal of the paradigm (see Gill, Weisburd, Telep, Vitter, & Bennett, 2014 for a systematic review of community policing goals). While several studies describe an inverse association between COP and crime reduction (e.g. Eck & Maguire, 2000), others have found limited evidence of such a relationship (e.g. MacDonald, 2002; Sozer & Merlo, 2013; Zhao & Thurman, 2004). As a consequence, researchers have raised concerns regarding the crime reducing the viability of the community policing model (MacDonald, 2002; Weisburd & Eck, 2004). The nature of the relationship between COP and violent offending by juveniles and young adults, in particular, remains at the centre of such debates (Forman, 2004; Myers, 2002).

Young individuals and the police

Criminological research has often suggested that young individuals are at increased risk of offending, victimization, and to be subject to police contact and arrest (e.g. Brown, Novak, & Frank, 2009; Murphy, 2015; Snyder & Sickmund, 2006). A multitude of studies have suggested that the onset of criminal activity starts in late childhood/early adolescence, peaks during late adolescence and early adulthood, and gradually decreases with age (Farrington, 1986; Hirschi & Gottfredson, 1983; Moffitt, 1993; Sampson & Laub, 1992). Correspondingly, adolescents and youth account for a substantial portion of police-citizen contacts (Puzzanchera, 2014). At the macro-level, however, recent studies challenged the status of age structure as a covariate of violent crime (Burgason et al., 2014; Krivo & Peterson, 2000). An alternative conceptual approach suggests that some youth become isolated from local mainstream socializing institutions, such as school and employment (Ousey & Lee, 2002). The works of McCall, Land, Dollar, and Parker (2013) and Thomas and Shihadeh (2013) support this notion, each has reported increased levels of violence in communities maintaining larger percentages of young individuals who are disassociated from school and work.

While researchers have suggested that community policing and problem-oriented policing programs are likely to positively influence perceptions of police among youth (Anderson, Sabatelli & Trachtenberg, 2007; Friedman, Lurigio, Greenleaf, & Albertson, 2004; Leroux & McShane, 2017), this demographic is often overlooked in community policing practices (Forman, 2004). Yet, research suggests that since youth identities are in the process of development, encounters with the law can have a significant impact (Murphy, 2015). Studies on perceptions of police have suggested that younger members of the community tend to have less favourable attitudes towards the police than their older counterparts (Brown & Benedict, 2002; Frank, Brandl, Cullen, & Stichman, 1996; Hurst & Frank, 2000; Reisig & Correia, 1997). As described by Friedman et al. (2004) and Taylor, Turner, Esbensen, and Winfree (2001), such findings are consistent with the disproportionately frequent negative contact between police and youth. As a consequence, attempting to incorporate youth in community-oriented policing represents a salient challenge for police departments (Hurst & Frank, 2000). Recognizing the need for such positive interaction, the President’s Task Force (2015) has recommended community support of police practices to protect youth at risk of violent offending and victimization, as well as the development of programs to promote collaboration and positive interaction between youth and the police. Similarly, the importance of including youth in community decision-making has been strongly emphasized (President’s Task Force on 21st Century Policing, 2015). As described by Warner, Beck, and Ohmer (2010), however, the success of such endeavours may be confounded by community characteristics.

Disparities between large and small jurisdictions

COPS Office efforts to support and encourage the adoption of community policing remain widespread, with the expressed purpose of facilitating increased effectiveness and responsiveness to community needs among both large and small jurisdictions (COPS, Office of Community Oriented Policing Services, 2014b; National Institute of Justice, 2004). Generally speaking, COP practices tend to be well received in both urban and rural locales (Zhao & Thurman, 2003). As noted by Sozer and Merlo (2013), however, potential differences between large and small jurisdictions, both in terms of crime-generating structural characteristics and agency operations, should not be overlooked.

Structural theories of crime have consistently demonstrated the deleterious effects of socio-economic deprivation, residential instability, and ethnic heterogeneity on crime (Bursik & Grasmick, 1993; Sampson & Groves, 1989; Shaw & McKay, 1969). These structural characteristics are theorized to disrupt the ability of communities to regulate themselves through informal social controls. While structural perspectives have historically been tested in urban settings, the work of Osgood and Chambers (2000) suggested that hypothesized relationships between structural characteristics and violent offences committed by youth manifest in rural communities. Other studies, however, suggest that such findings may not be salient, highlighting potentially important distinctions between large and small communities in terms of the relationship between community structure and crime. Wells and Weisheit (2004), for example, noted that economic improvements may yield crime increases within rural communities.

Large and small jurisdictions differ significantly in terms of both department characteristics and operations (Cordner & Scarborough, 2003; Hawkins & Weisheit, 2003; Reaves, 2015). Consequently, distinctions may also exist between large and small agencies in relation to the implementation and practice of COP strategies and, in turn, the results of incorporating COP practices into their respective operations. Furthermore, extant literature has suggested that rural and urban agencies differ greatly in terms of needs-orientation. Officers serving in small jurisdictions frequently maintain close social ties with residents and are responsive to the needs and priorities of the community (Weisheit, Wells, & Falcone, 1994).In contrast, officers serving in urban jurisdictions enjoy comparatively lower levels of community interaction and support. In addition, they are responsive to the needs and priorities of the department (Weisheit et al., 1994). Officers in small agencies, then, are likely to ‘do’ community policing regardless of formal adoption by the department (Sozer & Merlo, 2013).

Such disparities extend to evaluations of COP practices. Regarding small jurisdictions, a study of five communities by Adams, Rohe, and Arcury (2005), for example, revealed that implementation of COP prompted lower levels of fear of crime and increased community attachment among residents. The findings of recent study by Rukus et al. (2018), suggested that community policing was related to youth services in suburbs and rural areas. Similarly, a study of a single community by Brand and Birzer (2003) indicated that implementation of COP was associated with improved police services and resident perceptions of the police department, as well as reductions in crime – a finding that stands in conflict to that of a similar case study conducted by Prine, Ballard, and Robinson (2001).

The effects of COP in larger jurisdictions are also somewhat tenuous. While a case study of a large Southern community by Lord, Kuhns, and Friday (2009) indicated that adoption of COP had no effect on citizen perceptions of police or fear of crime, Rukus et al. (2018) found community policing had a positive effect on community participation in metro areas. Similarly, Zhao, Scheider, and Thurman (2002) found that COP funding was associated with significant reductions in violent crime in large cities. Despite increased efforts by the COP Office to fund community policing in small jurisdictions, Zhao and Thurman (2004) reported that grant efforts have been more effective in reducing crime in larger jurisdictions. While this finding may be related to difficulties of modifying and implementing strategies initially designed for urban locales (Rosenbaum, 1986), it also suggests the need to evaluate the relationship(s) between COP and disaggregated offences – particularly those committed by populations at increased risk of involvement in the crime.

Considering the relative influence of COP on arrest

Overall, there is little consensus within the extant literature concerning the relative crime-reducing effectiveness of COP amongst specific age groups across jurisdictions of varying size. Previous works have largely focused on perceptions of COP and fear of crime. While improved community relations and decreased fear of crime are certainly goals of the COP paradigm, so too is the reduction of the crime itself. The dearth of literature as it relates to the correlation COP implementation of criminal offending, particularly among youth, precludes holistic assessment of the effectiveness of COP. Moreover, prior analyses of COP have often been restricted to individual communities and small sample sizes. As a consequence, degree to which the relationship between COP and violent crime is (in)consistent by age group and jurisdiction size remain unknown. These limitations in the extant literature prompt the following research questions:

Research Question 1: What is the relationship between community-oriented policing and arrests for violent offences among young individuals?

Research Question 2: How does the relative influence of community-oriented policing on arrests of young individuals for violent crimes vary by jurisdiction size?

Data, measures, and methods

The current study examines the influence of community policing strategies on arrest counts for violent offences committed by juveniles and young adults across large and small jurisdictions. To accomplish this, we draw on arrest data taken from the Uniform Crime Report (UCR) for 2008–2010, census estimates of demographic, social, and economic characteristics from the 2007–2011 American Community Survey (ACS), and police agency characteristics from the 2007 Law Enforcement Management and Administrative Statistics (LEMAS). While the LEMAS is designed to be representative of all state and local law enforcement agencies, the current study is limited to an examination of municipal law enforcement agencies. After eliminating jurisdictions to maintain insufficient data, the final sample includes 364 large agencies and 626 small agencies. Roughly half of all municipal law enforcement agencies for which data are available through the LEMAS survey.¹ In accordance with previous literature (e.g. Roth & Ryan, 2000; Sozer & Merlo, 2013; Wells & Weisheit, 2004), large agencies are identified as those serving populations of 50,000 or more and small agencies as those serving populations under 50,000.

Dependent variables

The use of the UCR allows the present study to examine disparities in the effects of COP not only in terms of agency size, but ages of offenders. Consistent with the work of Osgood and Chambers (2000), we examine violent crime as 3-year counts of UCR Part 1 violent offences (homicide, robbery, and aggravated assault), which are disaggregated by age. Violent juvenile offences are considered those homicides, robberies, and aggravated assaults committed by individuals between the ages of 10 and 17. In contrast, violent offences by young adults are considered those committed by individuals between the ages of 18 and 24. Descriptive statistics for the outcome measures are presented in Table 1. There is considerable variation in average arrest counts among juveniles, ranging from a low of six arrests to a high of 1,793 arrests in large jurisdictions and 0 to 292 arrests in small jurisdictions.

Table 1. Means and standard deviations for all measures

	Large agencies (n = 364)		Small agencies (n = 626)
	Mean	SD	Mean	SD
Juvenile Arrests	153.89	233.07	16.43	25.10
Young Adult Arrests	323.32	476.80	33.54	44.63
COP Scale	6.47	2.29	3.49	2.52
COP Officers (%)	13.22	25.23	12.93	26.73
Police Per Capita	1.83	0.61	2.69	1.74
Police Education	0.28	0.45	0.19	0.39
Field Personnel (%)	73.38	10.43	77.71	13.00
Concentrated Disadvantage	0.01	0.65	−0.01	0.79
Renters (%)	42.58	10.94	38.25	12.50
Blau Index	0.48	0.15	0.35	0.18
Adult Arrest Rate	7.43	3.50	8.00	6.51
Pop 15–24 (%)	15.97	4.95	14.28	5.62
Institutionally Isolated Youth (%)	2.99	1.67	3.19	3.87
Population (Ln)	11.59	0.65	9.40	0.93

Primary explanatory variable

Community policing, both as a philosophy and organizational strategy, should influence all aspects of an agency; extending beyond the simple designation of personnel to engage in community policing practices. In accordance with this perspective, as well as an awareness that agencies are not required to follow specific criteria in the implementation of COP (Morabito, 2010), a summary measure of implementation of COP strategies is created. To create this measure, 10 distinct items² from the ‘community policing’ section of the 2007 LEMAS survey are dichotomized (0 = Not Implemented; 1 = Implemented) and summed. Importantly, the items selected for inclusion in the summary measure encompass the three primary tenets of COP: collaborative police-community partnerships, agency’s management and organizational structure, and problem-solving processes (as defined by the COPS, Office of Community Oriented Policing Services, 2014a). A greater value on the scale indicates a higher level of implementation of community policing strategies across agency operations. As displayed in Table 1, the large jurisdictions included in the sample adopted, on average, a wider variety of COP strategies in comparison to small jurisdictions. Specifically, large agencies adopted an approximate average of six COP strategies compared to an average of three strategies adopted by small agencies.

Control variables

In addition to the primary explanatory variable, we control for numerous community- and department-level characteristics. Scale of place is accounted for with the total population size, converted to its natural logarithm. As suggested by previous studies, high levels of structural disadvantage, residential instability, and ethnic heterogeneity may disrupt the ability of communities to regulate themselves (e.g. Bursik & Grasmick, 1993; Sampson & Groves, 1989) and inhibit the development of effective collaboration between residents and police (Mastrofski & Greene, 1993). To control for these effects, we draw upon data pertaining to structural characteristics of communities maintained in the 2007–2011 ACS 5-year summary file. Indicators of structural disadvantage include measures of poverty, female-headed households, unemployment, and low educational attainment, while the per cent of renters is treated as a proxy of residential instability. Racial/ethnic heterogeneity is measured relying on Blau’s (1977) index of intergroup relations (1- Σ Pi²), where i represents the various racial/ethnic group proportions in the population.³ The diversity index ranges from 0 to 1 in which lower scores indicate homogeneity while scores approaching one indicate increasing heterogeneity. Preliminary analyses reveal high levels of multicollinearity among aggregate indicators of structural disadvantage for both samples. For large jurisdictions, an obliquely related factor analysis indicates measures of poverty, female-headed households, unemployment, and low educational attainment converge on a single dimension with an Eigen-value in excess of 1 and factor loadings in excess of .714 (α = .82). Identical analyses for small jurisdictions revealed the same measures converge on a single dimension with an Eigen-value in excess of 1 and factor loadings in excess of .643 (α = .80).

As has been demonstrated by previous works, the presence of crime-prone groups has also been linked with elevated levels of community violence. Specifically, the relative prevalence of youth (Hirschi & Gottfredson, 1983) and young individuals who become isolated from mainstream socializing institutions (McCall et al., 2013; Thomas & Shihadeh, 2013) have frequently been correlated with violent offending. To control for the effects of these populations, we include measures of age structure and institutionally isolated young individuals. Consistent with prior research, age structure is measured as the per cent of the population between the ages of 15 and 24, while institutionally isolated youth is measured as the per cent of persons between the ages of 16 and 19 who are simultaneously unemployed, not in labour force, and not enrolled in school. To control for community-level violence, we include the 3-year rate of adults age 25 and over arrested for Part 1 violent offences.

Department-level characteristics are drawn from the 2007 LEMAS, including indicators of COP officers, proportion of field operations personnel, officers per capita, and department educational requirements. Each of these measures was selected due to its potential influence on daily operations and police-community relations (Sozer & Merlo, 2013; Weisheit et al., 1994). The percentage of COP officers is created by dividing the number of sworn personnel specifically assigned to engage community-policing activities by the overall number of sworn officers with general arrest powers and multiplied by 100. Proportion of field operations personnel is measured by dividing the number of police personnel assigned to the street by the total number of agency employees. Officers per capita, which serves as a proxy of formal social control capacity, is computed by dividing the number of sworn police officers by population number and multiplying it 1,000. Finally, department educational requirement is measured as a dichotomous measure whereby a value of 1 is assigned if the department requires at least ‘some college’ within 2 years of hire and ‘0ʹ if the department requires either a high school degree or no formal educational requirement has been implemented. Means and standard deviations for each of the control measures are displayed in Table 1. Preliminary diagnostic analyses reveal an absence of multicollinearity between the independent variables, as evidenced by VIFs of less than 2.2 in the large jurisdiction sample and 2.0 in the small jurisdiction sample.

Analytical technique

Consistent with the recommendations of Osgood (2000), we use a Poisson-based estimator to predict variation in arrest counts across jurisdictions and age groups. Population size, converted to its natural logarithm, is specified as the ‘offset’ variable. While the standard Poisson model assumes the mean and variance of an outcome measure to be equal, preliminary analyses involving post-Poisson goodness-of-fit tests reveals that the variance exceeds the mean for each of the four outcome measures. Such over-dispersion suggests that standard Poisson modelling is inappropriate, which prompted the use of a negative binomial estimator. Robust standard errors are utilized in each model as a means of accounting for potential non-independence in the data. While negative binomial regression is an important first step toward assessing potential disparities in the relative influence of COP on offending across age groups and jurisdictions, we recognize that more comprehensive assessment requires a statistical test of the significance of the differences between model coefficients. To accomplish this, we test the inequality of coefficients using the method developed by Clogg, Petkova, and Haritou (1995).⁴

Results

The results of negative binomial analyses predicting variation in violent crime arrests among juveniles in large and small jurisdictions are presented in Table 2. Regarding the primary explanatory variable, increases in the implementation of COP strategies correspond with lower levels of arrests for violent offences among juveniles in large jurisdictions. To put these results into context, a one standard deviation increase in the COP scale is, on average, associated with an 8.8% decrease in arrests among juveniles. These results do not extend to small jurisdictions. Application of Clogg tests further highlights this inconsistency, as 3 of 6 model comparisons reveal statistically significant differences. Specifically, the level of COP implementation exhibits significantly different effects on arrests of juveniles for violent offences in large jurisdictions in comparison to small jurisdictions.

Table 2. Negative binomial analyses predicting violent arrest counts

	Large agencies (n = 364)		Small agencies (n = 626)
	Juveniles	Young Adults	Juveniles	Young Adults
COP Scale	−.040***	−.016†	.017	.014
	(.011)	(.009)	(.015)	(.010)
Percent COP	.033†	.013	−.009	.013
Officers (Ln)	(.019)	(.017)	(.026)	(.018)
Police Per	−.107*	−.144***	.027	.046*
Capita	(.051)	(.037)	(.036)	(.023)
Police	.039	−.074†	.101	−.005
Education	(.056)	(.044)	(.077)	(.061)
Field	−.001	.000	−.001	−.003
Personnel (%)	(.003)	(.002)	(.003)	(.002)
Concentrated	.426***	.324***	.514***	.342***
Disadvantage	(.060)	(.047)	(.066)	(.045)
Blau Index	1.028***	.854***	.889***	.445***
	(.170)	(.129)	(.200)	(.140)
Adult Arrest	.052***	.078***	.027***	.049***
Rate	(.008)	(.008)	(.006)	(.005)
Percent	−.282*	−.011	−.175	.013
15–24 (Ln)	(.129)	(.088)	(.145)	(.089)
Percent	.012***	.015***	.002	.012***
Renters	(.003)	(.002)	(.004)	(.003)
Institutionally	.005	.005	−.008	.000
Isolated Youth (%)	(.017)	(.015)	(.010)	(.007)
Constant	−7.120***	−7.563***	−7.287***	−7.482***
	(.434)	(.325)	(.473)	(.293)

Unstandardized coefficients with robust standard errors in parentheses. †p ≤ .10; *p ≤ .05; **p ≤ .01; ***p ≤ .001

In contrast to the varying association between COP implementation and violent crime arrest counts among juveniles, the influence of the per cent of COP designated officers is more uniform. Within large jurisdictions, the relationship between per cent of COP officers and violent crime arrests among juveniles achieves marginal statistical significance. Each standard deviation increase in COP officers, on average, related to a 4.1% increase in juvenile arrests. The results of Clogg tests, however, demonstrate that this effect is not significantly different in comparison to the remaining models, all of which fail to achieve statistical significance. Overall, these findings demonstrate the relatively uniform null influence of COP officers while providing a layer of support for the notion that the effects of implementing COP strategies and COP officers are conceptually and empirically distinct.

The relationship between departmental education requirement and violent crime arrest counts is also relatively consistent. Findings indicate that this relationship is null across regression models. The use of Clogg tests indicates this effect is differentially experienced at a statistically significant level only in comparison to violent crime arrests among juveniles in small jurisdictions. The final department-related control, per cent of personnel assigned to field activities, fails to exhibit a statistically significant relationship with outcome measures in any of the four regression models.

Although the relative influence of COP officers is almost uniformly null, capacity for formal social control, measured here as police per capita, exhibits inconsistent and contrary effects. In large jurisdictions, police per capita are shares a statistically significant inverse relationship with violent crime arrests among both juveniles and young adults. In small jurisdictions, police per capita share a null relationship with violent crime arrest counts among juveniles but a statistically significant positive relationship with violent crime arrest counts among young adults.

Analyses also reveal a number of similarities and distinctions regarding the relative effects of community characteristics on violent crime arrests among youth. The index tapping levels concentrated disadvantage shares a positive and statistically significant relationship with violent crime arrest counts across age groups and jurisdictions. Specifically, a standard deviation increase in levels of concentrated disadvantage corresponds, on average, with increases of 31.9% and 23.4% in violent crime arrests among juveniles and young adults in large jurisdictions. In small jurisdictions, a standard deviation increase in the concentrated disadvantage index is correlated, on average, with a 49.8% increase in violent crime arrest counts among juveniles and 30.9% increase in violent crime arrest counts among young adults. Clogg tests indicate that the estimated effects of aggregate levels of concentrated disadvantage on violent crime arrest are substantively consistent across 4 of 6 model comparisons, with the only statistically significant differences observed between violent crime arrest counts among young adults in large jurisdictions compared to arrests of juveniles in small jurisdictions and arrests of juveniles in small jurisdictions compared to arrests of young adults in small jurisdictions.

Also, in accordance with structural perspectives of violent crime are results as they relate to ethnic heterogeneity as measured by the Blau index of dissimilarity, which shares a positive and statistically significant positive relationship with violent crime arrest counts across all regression models. Clogg tests, however, reveal statistically significant inequality in coefficients across 3 of 6 models comparisons. The estimated effect of the Blau index is significantly different on violent crime arrest counts for juveniles in large jurisdictions compared to young adult arrests in small jurisdictions. Similar inequality is observed in terms of arrest counts for young adults in large jurisdictions compared to those in small jurisdictions, as well as between arrest counts for juveniles and young adults in small jurisdictions.

While less consistent than that of concentrated disadvantage and ethnic heterogeneity, the relative influence of residential instability, measured as the per cent of renter-occupied homes, is generally consistent with structural perspectives. The per cent of renter-occupied homes shares a positive and statistically significant relationship with violent crime arrest counts for juveniles and young adults in large jurisdictions, as well as young adults in small jurisdictions. Clogg tests reveal statistically significant differences in the relative effects of residential instability across 3 of 6 model comparisons. Compared to juvenile arrests in small jurisdictions, the measure of renter-occupied homes exhibits a stronger association with arrests among juveniles and young adults in larger jurisdictions. Similarly, the effects of residential instability are more strongly related to arrests of young adults than juveniles in small jurisdictions.

As was described in the review of the literature, the relative prevalence of youth (Moffitt, 1993; Sampson & Laub, 1992) and institutionally isolated youth (McCall et al., 2013; Thomas & Shihadeh, 2013) might each be expected to maintain strong and positive relationships with community levels of violence. Our results, however, are not supportive of either perspective. The prevalence of institutionally isolated youth, measured as the per cent of 16- to 19-year olds, who are simultaneously unemployed, not in labour force, and not enrolled in school, fails to exhibit a statistically significant relationship with violent crime arrests counts in any of our regression models. In contrast, the prevalence of youth, measured as the per cent of the population between the ages of 15 and 24, shares a statistically significant but inverse relationship with violent crime arrest counts among juveniles in large jurisdictions. Specifically, a standard deviation increase in the relative prevalence of this population corresponds with a 7% decrease in violent crime arrest counts. This relationship appears relatively isolated to juvenile arrests in large jurisdictions, as demonstrated by the results of Clogg tests, which reveal that the influence of the relative prevalence of youth is stronger for juvenile arrests in large jurisdictions than violent crime arrest counts for young adults in large and small jurisdictions.

Finally, results of regression analyses demonstrate the robust effect of aggregate levels of adult violence, which shares a positive and statistically significant relationship with crime counts across all models. In relation to large jurisdictions, standard deviation increases in aggregate levels of adult violence correspond with increases of 20.1% and 31.4% in violent crime arrest counts among juveniles and young adults. Similar increases of 19.6% and 37.4% are observed in arrest counts among juveniles and young adults in small jurisdictions. The results of Clogg tests, however, reveal that this effect is differential across jurisdictions and age groups. The relative influence of aggregate levels of adult violence is weaker for violent crime arrest counts among juveniles in comparison to young adults in both large and small jurisdictions. Similarly, this effect this stronger for violent crime arrest counts for juveniles and young adults in large jurisdictions compared to juveniles in small jurisdictions; and young adults in large jurisdictions compared to those in small jurisdictions.

Discussion and conclusions

The present study advances the literature by examining the differential effects of community policing on crime arrest counts juveniles and young adults across large and small jurisdictions. With the exception of Sozer and Merlo (2013), the methodologies of previous works on this topic have not allowed for such comparisons. This notable omission ignores important distinctions between large and small jurisdictions (Cordner & Scarborough, 2003; Hawkins & Weisheit, 2003; Reaves, 2015), which risks fostering the inaccurate notion that capacity for implementation, practice, and success in COP are consistent regardless of agency size. The use of crime measures that are not disaggregated by age promotes similar problems in terms of assessing the effectiveness of COP strategies. As our findings demonstrate, the relationship between crime arrest counts and the implementation of COP strategies is relative to the size of the jurisdiction and age group being examined. These effects are inversely related to crime arrest counts among juveniles and young adults in large jurisdictions, suggesting the implementation of COP is successful in reducing levels of violence in urban settings. As might be expected, this effect was most pronounced for violent crimes committed by juveniles. In contrast, levels of COP implementation appear unrelated to violent crime arrests in small jurisdictions.

As was noted in the review of the literature, small jurisdictions frequently practice COP philosophies without formal adoption of COP strategies outlined in LEMAS (Sozer & Merlo, 2013). Similarly, small jurisdictions, compared to large jurisdictions, maintain stronger social ties with community members and higher levels of responsiveness to community needs (Weisheit et al., 1994). The null effect of the COP scale within small jurisdictions, therefore, may simply be an artefact of strong levels of cooperation, trust, and familiarity between police and community members (including youth) that tend to exist in non-urban locales.

While the inclusion of potentially confounding department and community variables are examined for the purposes of isolating the effects of COP, their relative influence also merits further discussion. First, the general null impact of the per cent of COP designated officers across all models strongly supports the notion that community policing must extend beyond simple designation of personnel to the implementation of a variety of practices if it is to be successful as an organizational strategy in reducing violent crime. The positive relationship between the per cent of officers designated as community policemen and juvenile arrests for crimes of violence in large jurisdictions is, at face value, also indicative of this assumption. Alternatively, we acknowledge the possibility that this relationship may be the product of increased cooperation between civilians and police – which is certainly a goal of COP (Gill et al., 2014). Given that that relationship achieved marginal significance only in terms of arrests among juveniles in large jurisdictions, however, we are prompted to accept the former explanation. Simply stated, the title of ‘Community Police Officer’ matters little if the officer is not engaged in community policing activities. Second, similarities between large and small jurisdictions in terms of educational requirements and percentage of field personnel extended to their general null effects across statistical models. The null influence of educational requirements, with the exception of the marginally significant inverse relationship shared with arrests of young adults in large jurisdictions, is particularly interesting, as it might be expected that a more highly educated police force would produce a number of benefits, including reduced crime. It is noteworthy, however, that educational requirements are uniformly low across the samples included in this study. Moreover, the use of alternative outcome measures (i.e. confidence in police, use of force reports, or fear of crime) in similar regression models may yield markedly different results. Our findings, therefore, should not be interpreted as evidence of the need to discount the potential importance of college education among police officers.

Third, the relationship between arrests and formal social control capacity is statistically significant in three of the four regression models; but while this relationship is inverse in large jurisdictions, formal social control capacity is positively related with violent arrest counts among young adults in small jurisdictions. As is the case with per cent of COP officers, these findings may be attributed to varying levels of cooperation between community members and police. In support of this assertion, Weisheit et al. (1994) described stronger social ties and community support for police in small jurisdictions compared to large jurisdictions. A less encouraging but equally relevant possibility is that increasing levels of formal social control may promote elevated levels of crime in small jurisdictions. As described by Lee (2008), rural areas, in comparison to urban locales, tend to maintain stronger kinship and network ties associated with informal social control. Increased formal social control, particularly in the form of enforcement by police, may disrupt the efficacy of these networks in communities that are otherwise able to regulate themselves; amplifying social dislocations including violent crime. While our data do not allow us to determine the precise nature of the relationship between formal social control capacity and violent crime arrests, our findings are suggestive of the need for police agencies to carefully consider community characteristics while developing department strategies.

Fourth, our findings as they relate to concentrated disadvantage, ethnic heterogeneity, and residential instability are relatively consistent across regression models and with structural perspectives of informal social control and youth violent offending (Bursik & Grasmick, 1993; Sampson & Groves, 1989; Shaw & McKay, 1969); lending support to the argument by Osgood and Chambers (2000) that the structural covariates of crime, particularly among juveniles, are ubiquitous across areas of varying size. Our results as they relate to the relative prevalence of youth and institutionally isolated youth, however, are inconsistent with previous works. We do, however, suggest caution in the interpretation of these findings and encourage examination beyond direct effects. Specifically, we encourage future researchers to examine the potential mediated and moderated relationships between indicators of social disorganization, institutionally isolated youth, age structure, COP strategies, and violent crime. It is possible, for example, that the success of COP in reducing crimes committed by youth is dependent on the degree to which juveniles and young adults have been isolated from mainstream socializing institutions within the community.

In addition to the recommendations and limitations noted above, two additional concerns are offered. First, the cross-sectional nature of this study does not allow for a determination of causality. Indeed, COP may be more frequently implemented in communities where relations between police and civilians are comparatively strong. Second, the generalizability of our findings is limited to violent crimes committed by a specific segment of the population, which in turn limits our assessment of the efficacy of COP. Future studies, therefore, should feel encouraged to engage in longitudinal assessments of COP across a wide array of offence categories. Moreover, we strongly recommend that future studies endeavour to extend our methodology to offences disaggregated by race and ethnicity. Such assessment of COP lends itself not only to the development evidence-based policy but the implementation of practices that build trust and respect between police and the communities they serve. As recent events in the United States have illustrated, these concepts are frequently lacking in the areas where they are most needed.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Notes on contributors

Rocio Alejandra Paez

Rocio Alejandra Paez is a Visiting Assistant Professor in the Department of Sociology and Criminology at the University of Arkansas. Her current research interests include include juvenile delinquency, the intersection of race, immigration, and crime and the role that networks and institutions play in neighborhood crime. Her most recent works have been published in the American Journal of Criminal Justice and Deviant Behavior.

Rick Dierenfeldt

Rick Dierenfeldt is an assistant professor in the Department of Social, Cultural, and Justice Studies at the University of Tennessee at Chattanooga. His research interests include the intersection of structure, culture, race, gender, and crime, as well as policy evaluation in the fields of policing and corrections. His recent publications have appeared in the Journal of Criminal Justice, Journal of Interpersonal Violence, Crime and Delinquency, and Deviant Behavior.

Notes

1. The 2007 LEMAS dataset contains information on a sample of 3,095 law enforcement agencies, including state police agencies, sheriff’s departments, and municipal police departments. Nearly 70% (2,095) of this sample consisted of local agencies.

2. 1) agency’s mission statement includes a community policing component, 2) agency actively encouraged patrol officers to engage in SARA-type problem-solving projects, 3) agency conducted a citizen police academy, 4) agency maintained or created a formal, written community policing plan, 5) agency gave patrol officers responsibility for specific geographic areas/beats, 6) agency included collaborative problem-solving projects in the evaluation criteria of patrol officers, 7) agency upgraded technology to support the analysis of community problems, and 8) agency partnered with citizen groups and included their feedback in development neighbourhood/community policing strategies, 9) agency conducted or sponsored a survey of citizens on crime, fear of crime, or satisfaction with police services, and 10) agency maintained a community policing unit with full-time personnel.

3. Seven racial/ethnic categories, White, Black, American Indian, Asian, Native Hawaiian or Pacific Islander, Hispanic, and Other, are used to construct this index.

4. The coefficient comparison strategy is a significance test that takes the form: Z = b1 – b2 / √ (SEb₁² + SEb₂²).