Skip to main content
SearchLogin or Signup

The Effect of Police Layoffs on Crime: Results from a Natural Experiment Involving New Jersey’s Two Largest Cities

The current study tests the effect of police layoffs on crime through a natural experiment involving Newark and Jersey City, New Jersey’s two largest municipalities.

Published onJul 24, 2020
The Effect of Police Layoffs on Crime: Results from a Natural Experiment Involving New Jersey’s Two Largest Cities
·

Abstract

The current study tests the effect of police layoffs on crime through a natural experiment involving Newark and Jersey City, New Jersey’s two largest municipalities. In response to severe budget shortfalls resulting from the economic recession beginning in 2008, officials in both Newark and Jersey City explored police layoffs as a potential component of their cutback strategies. In November 2010, the Newark Police Department terminated 13% of the police force while Jersey City officials were able to avert any layoffs from occurring. The current study uses monthly Part 1 crime counts spanning from 2006 to 2015 to measure the effect of the police layoffs on crime in Newark. Findings of time series generalized least squares regression models using panel data indicate the police layoffs were associated with significant increases of violent crime, murder, robbery, and larceny theft in Newark as compared to Jersey City. This study contributes to the literature on the relationship between police force size and crime by demonstrating sudden, drastic reductions in police force size may generate significant crime increases.

Introduction

In early 2008, the United States economy began a significant downturn, representing the 10th economic recession since World War II (Wiseman, 2011). Over the approximately 18 months the recession lasted, budget shortfalls required strategic cutbacks on the part of American police agencies (COPS, 2011; Wiseman, 2011). In many instances, such cutbacks included police officer layoffs, with estimates suggesting that approximately 10,000 law enforcement jobs were lost during the recession (COPS, 2011). Researchers have previously focused their attention on the impact of the great recession on policing, exploring issues such as the impact of police officer layoffs on agency operations (COPS, 2011; PERF, 2010), cutback management strategies for police to consider (Wiseman, 2011), and the contextual factors associated with agencies either enacting or averting officer layoffs in the face of fiscal distress (Giblin and Nowacki, 2018). However, the effect of police layoffs on crime, to our knowledge, has yet to be subjected to empirical evaluation.

In considering potential casual mechanisms linking police layoffs with changes in crime levels, the literature on police force size provides a logical source for exploration. Reviews of the literature have found that increasing police force size does not reduce crime (Lee et al., 2016; Sherman and Eck, 2002) or improve organizational performance (Skogan and Frydl, 2004). However, it is important to note that the majority of studies on police force size have tested incremental changes in manpower, speaking to the marginal effect of police (Sherman and Eck, 2002). Research considering the absolute effect of police, in which police force sizes are drastically and quickly reduced (Sherman and Eck, 2002), shows that the reduction of manpower is significantly related to crime occurrence. In addition, while research suggests that police strategy is a more important consideration for crime prevention than police force size (Lee et al., 2016), it is important to note that manpower levels and changing strategy are not always mutually exclusive propositions. Simply put, sufficient manpower may be necessary to institute evidence-based crime prevention practices.

The current study seeks to contribute to the literature through a natural experiment testing the effect of police layoffs on crime. In 2008, the Newark Police Department (NPD) and Jersey City Police Department (JCPD), New Jersey’s two largest police forces, were faced with the prospect of police layoffs due to severe budget cutbacks. The police union and city officials in Jersey City were able to come to a contract agreement to avert officer layoffs (Porter, 2011). No such agreement was achieved in Newark, leading to the termination of 13% of the NPD on November 30, 2010 (Star Ledger, 2010). Our analysis found that crime significantly increased in Newark as compared to Jersey City following the NPD layoffs. We conclude the paper with a discussion of the study implications for public policy, evidence-based policing, and contemporary policing research. We begin with a review of the empirical literature that informed the current study.

Review of Relevant Literature

Research into the relationship between police force size and crime rates is one that has spanned decades, methodologies, and statistical inquiry. A recent systematic review and meta-analysis (Lee et al., 2016) analyzed 229 findings across 62 studies conducted from 1971 to 2013. The analysis was concerned with whether a change in the size of an existing police force had an impact on crime. Lee and colleagues (2016) estimate a small, nonsignificant effect size suggesting police force size to be unrelated to crime. Lee et al. (2016) further contextualized their findings by comparing police force effect size to those from meta-analyses of contemporary policing strategies: problem-oriented policing (Weisburd et al., 2010), neighborhood watch (Bennett et al., 2006), hot spots policing (Braga et al., 2014), and focused-deterrence (Braga and Weisburd, 2012). This exercise showed that the effect of police force size was, in the words of Lee et al. (2016), “miniscule” in comparison to these policing strategies. The cumulative findings of their study led Lee et al. (2016) to two main conclusions: 1) that research on the effect of police force size on crime has “exhausted its utility” and 2) changing policing strategy is likely to have a greater impact than adding more police.

Upon further inspection of the literature, the finding that increasing police force size does not decrease crime may be explained by analyzing how officers allocate their time on duty. Researchers have found the majority of officer’s time is spent on activities other than fighting crime (Black, 1971; Cumming et al., 1965; Payne, 2017; Reiss, 1973; Whitaker, 1982). Given that most incidents police are responding to are not criminal in nature, perhaps we should not expect officer levels to significantly impact the crime rate.

However, research published since Lee et al. (2016) suggests there may be a relationship when focusing on crime type, swiftness of force level change, and deployment strategy. While the body of empirical research has found that adding police does not universally decrease crime, it does not mean that reducing the size of the police force will be without consequence in all instances (Payne, 2017). Recent studies by Chalfin and McCrary (2018), Mello (2019), and Kaplan and Chalfin (2019) suggest that the question researchers should be asking is whether police are impacting violent crime – particularly murder – and property crime in the same way. Chalfin and McCrary (2018) utilize various statistical models to account for simultaneity bias found within the quasi-experimental literature, and reject violations of classical measurement error. Findings from their generalized method of movements (GMM) model suggest that police reduce murder to a greater extent than they do assault, larceny, and even burglary. Similarly, Mello (2019) finds violent crime to be more responsive than property crimes to increases in the size of the police force. Specifically, one additional officer results in a decrease of 0.11 murders, 0.53 rapes, and 1.98 robberies.

In a similar vein, Kaplan and Chalfin (2019) show that the addition of one officer has the ability to stop approximately seven index crimes – six property and one violent. A particularly noteworthy contribution of the study is the finding that crime reductions occurred absent any increases in arrest rates or prison commitments. At a time when mass incarceration continues to be a societal issue, the study finds that increasing the number of police, does not lead to increased incarceration, representing what the authors call, a “double dividend” – a decrease in crime and incarceration rates.

The aforementioned studies measured the effect of marginal force level changes on crime. The question then becomes what happens when change is drastic, and in the negative direction? Such a change in police force size speaks to what Sherman and Eck (2002) term the absolute effect of police, which is informed by research on police strikes. Five of the six police strike studies reviewed by Sherman and Eck (2002) found major increases in both violent and property crime following police strikes (Andenaes, 1974; Clark, 1969; Makinen and Takala, 1980; Russel, 1975; Sellwood, 1978). The lone exception was the study by Pfuhl (1983), which found that strikes across 11 cities had neither a statistically significant or systematic impact on rates of the reported crimes in question. However, Sherman and Eck (2002: 302) noted that 89% of the “strike” period in Pfuhl’s study actually consisted of non-strike days, which confounds the measurement of cause and effect. Nonetheless, we should note that the overall body of research on police strikes is methodologically weak with no study rating higher than 2 on the Maryland Scientific Methods Scale (Farrington et al., 2002; see Sherman and Eck, 2002: 303). As this is below the minimally interpretable research design (level 3), which involves a comparable control condition, it cannot be completely ruled out that crime would have increased even without a strike.

When discussing the general lack of evidence in support of police force size as a crime control mechanism, it should be noted that recent decades have seen the rise of evidence-based policing, which advocates for the strategic deployment of officers to be rooted in scientific evidence about best practices (Lum and Koper, 2017; Sherman, 1998). The evidence-based policing matrix maps existing evaluations of police practices, ranking specific interventions along three dimensions: nature and type of target, the degree to which the strategy is reactive or proactive, and the strategy’s level of focus (Lum et al., 2011; also see chapter 3 in Lum and Koper, 2017). Results indicate initiatives that are proactive, focused, and place-based are more likely to result in reduced crime and disorder when compared to initiatives concentrating on individuals, or that that are reactive.

The evidence-based policing literature indicates that increasing police presence within effective practices, such as hot spots policing, is much more likely to result in crime reduction than the standard model of policing, which research has long shown to be ineffective (Lum et al., 2011; Sherman and Eck, 2002; Skogan and Frydl, 2004). In relation to research on police force size, the evidence-based policing literature supports Lee et al.’s (2016) conclusion that changing police strategy is likely to have a larger crime control effect than increasing the size of a police force. However, an important caveat is that a given department may only be able to implement evidence-based practices when there are a sufficient number of cops on the force.

An example of such circumstance is the Flint, MI police department (FPD) (Terrill et al., 2014). During the study period, Flint was one of the most violent cities in the nation, suffered extreme socioeconomic distress, and lost approximately 50% of its sworn officer personnel (almost 300 to under 150 officers). This loss of officers meant that FPD was unable to implement hot spots policing, disorder reduction strategies, or CompStat due to a lack of resources. Additionally, officers once assigned specifically to community policing were shifted to basic patrol following the budget cutbacks. Less than three hours of an average eight-hour shift was spent engaging in police-citizen encounters (Terrill et al., 2014).

A similar situation was observed in Newark, NJ, which directly informs the current study. In the summer of 2008, the NPD began Operation Impact, a foot-patrol saturation intervention modeled after an NYPD strategy of the same name. On a nightly basis, Operation Impact deployed 12 foot-patrol officers in an approximately quarter-square mile area of the city. An evaluation by Piza and O’Hara (2014) found that the foot-patrols generated a significant reduction in overall violence as compared to two separate control areas (Piza and O’Hara, 2014). Despite the intervention proving beneficial, the NPD’s operational budget was significantly reduced, which resulted in the cancellation of Operation Impact patrols until it was completely phased out in 2010 over looming police layoffs (see Piza and O’Hara, 2014, 713).

In light of the mixed findings regarding the effect of police force size on crime, and its sensitivity to crime type, swiftness of force level change, and deployment strategy, our study seeks to add to the discussion by investigating how drastic and immediate police force change enacted through layoffs affects crime. The research question is explored in the context of New Jersey’s two largest municipalities, Newark and Jersey City. In November 2010, the Newark Police Department (NPD) terminated 13% of the police force while Jersey City officials were able to avert layoffs. The unique timing of this situation provided the opportunity for a natural experiment to measure the effect of police layoffs on crime across two similar municipalities. The current study expands upon prior research that primarily focused on marginal or incremental change to police force size.

Methodology

Study setting

Newark and Jersey City are the two largest municipalities in New Jersey, with populations of 277,140 and 247,597, respectively, according to the 2010 decennial census. Ethnic minorities account for the majority of the citizenry, with only 12.2% of residents in Newark and 22.6% of residents in Jersey City identifying as White alone. In 2010, Newark and Jersey City boasted the two largest police forces in the state, employing 1,308 and 831 officers, respectively (FBI, 2011a). Both cities exhibited 2010 overall part 1-, violent-, and property-crime rates well above average for New Jersey municipalities with at least 50,000 residents. According to the 2010 UCR, the 32 New Jersey municipalities with 50,000 or more residents reported an average Part 1 crime rate of 2,764 per 10,000 residents, a violent crime rate of 479 per 10,0000 residents, and a property crime rate of 2,285 per 10,0000 residents. Newark’s Part 1, violent, and property crime rates were 4,313, 1,029, and 3,284 per 10,000 residents while Jersey City’s Part 1, violent, and property crime rates were 3,203, 749, and 2,454 per 100,00 residents (FBI, 2010b). Nonetheless, as will be discussed subsequently, Newark and Jersey City experienced general reductions in crime over the preceding years. This is important in the context of the current study, as research suggests cities with diminishing crime rates may be more likely to consider police layoffs in the face of fiscal constraints (Giblin and Nowacki, 2018).

Over the recent decade, urban revitalization has become a key focus of officials in both Newark and Jersey City. In Newark, substantial revitalization efforts occurred within and surrounding the downtown area following the opening of the Prudential Center arena in 2007 (Piza et al., 2020: 671). Given its close proximity to Manhattan, Jersey City has continuously attracted additional residents from New York City with significant revitalization efforts occurring in pockets of neighborhoods throughout the city (Dow, 2018). Such occurrences mirror recent trends observed throughout the United States. Between 2000 and 2010, there has been more growth in the city-center as compared to the suburbs, among the nation’s 50 most populous regions, with more affluent individuals flocking to poor, city center neighborhoods; a process referred to as gentrification (MacDonald &Stokes, 2020).

While each of the variables are at differing levels across Newark and Jersey City, each city’s relative levels remain similar throughout the study period. The cities exhibit similar trends in terms of racial composition (percent White, Black, and Hispanic), education (percent with a Bachelor’s Degree), unemployment, housing (percent vacant or owner occupied and median home value), and income. As an example, while the median home values remain higher for Jersey City compared to Newark during the study period, both experience similar changes, with decreases in 2008 following the start of the economic recession, and an increase in 2013, which coincided with the end of the housing market crash (see Figure 1).1 Given the relative stability observed in each city, sociodemographic factors should have played a minimal role in any crime changes detected in the statistical analysis.

Figure 1: Median home values in Newark and Jersey City.

[CHART; see end of this Pub]

Similar to police departments around the United States, the NPD and JCPD were challenged by the economic downturn beginning around 2008. In early 2009, officials in both Newark and Jersey City suggested that police layoffs may be necessary to offset budget deficits, leading to prolonged negotiations between the city officials and their respective police unions. In Jersey City, the police union ultimately agreed to a contract providing an 11% raise over 4 years instead of the previously requested raise of 13% and approved a higher copay for medical prescriptions, which averted layoffs. The agreement saved the jobs of approximately 10% of the JCPD (Porter, 2011).

Negotiations in Newark were not as fruitful. The Newark Police union agreed to about $2.7 million in concessions and $6 million in pay deferrals, but rejected the city’s proposal for an overtime cap and 5 days of unpaid leave (Ford, 2010). The lack of a contract agreement resulted in the termination of 167 officers, approximately 13% of NPD officers, which took effect on November 30, 2010. The layoffs brought the per capita police officers in the NPD much more in line with that of JCPD. In 2010, the NPD had over 120 more officers per 100,000 residents than the JCPD (466.5 vs. 340.3). While the NPD officer rate was still larger than JCPD following the layoffs, the difference was much smaller (393.8 vs. 324.4). The difference became less pronounced from 2012 through 2015, as the JCPD added officers while the NPD gradually lost officers to attrition without funding any new police academies (see Figure 2).

Figure 2: Number of police officers (A) and officers per 100,000 population (B) in Newark and Jersey City

[CHART; see end of this Pub]

[CHART; see end of this Pub]

Evaluation design

The current study takes advantage of the unique timing of the fiscal crisis, budget negotiations, and police officer layoffs to conduct a natural experiment measuring the effect of police layoffs on crime in Newark. Natural experiments share certain characteristics with quasi-experiments, particularly the fact that the creation of experimental and control groups cannot be manipulated by the researcher. However, quasi-experiments make no presumption that the allocation of the intervention occurred in any random fashion. This differs from natural experiments, in which the allocation of the experimental condition occurs either a random or as-if random process, despite falling outside the control of researchers (Dunning, 2008). Demonstrating that treatment occurred in either a random or as-if random manner is an important first step for any natural experiment. Without doing so, researchers risk “conceptual stretching” by applying the term “natural experiment” to studies with weaker designs (Dunning, 2012).

We consider the situation observed with New Jersey’s two largest police departments as occurring in an as-if random process. In doing so, we rely on the causal-process observation method, which Collier et al. (2010: 184) describe as “an insight or piece of data that provides information about context, process, or mechanism.” Police unions and city administrators in both Newark and Jersey City were actively negotiating to avoid police officer terminations at the same point in time, with neither the NPD or JCPD having any motivation to self-select into the experimental group experiencing layoffs.2 Furthermore, as New Jersey’s two largest cities, Newark and Jersey City are more similar to one another than any of the other municipalities in the state in terms of police force size, pre-layoff police resources, and pre-layoff crime levels. Lastly, given that both cities considered police layoffs in response to budgetary concerns rather than issues of crime, the experimental variable (police layoffs) and dependent variable (crime rates) can be considered exogenous (Mitchell, 2015).

Analytical Approach

Study period

The current study incorporates crime data from the year 2006 through 2015 in recognition of the local context in Newark. In 2006, Cory Booker began his first term as Newark Mayor, with crime control as his administration’s top priority. Booker installed a new Police Director at the NPD who stressed an adherence to evidence-based practices, in particular proactive, place-based crime control strategies (Jenkins and DeCarlo, 2015). In January 2016, after Booker resigned as mayor following a successful campaign for the U.S. Senate, newly elected Mayor Ras Baraka merged Newark’s police department, fire department, and office of emergency management and homeland security into a single Department of Public Safety. While each entity maintained autonomy in determining strategy and daily operations, this merging of agencies allowed certain administrative duties to be shared across the disparate public safety missions, freeing up more personnel to perform front-line duties. In recognition of these contextual factors, restricting the study period to 2006 through 2015 allows us to best isolate the effect of layoffs without the potentially confounding factors of a changing agency mission (pre-2006) or creation of the Public Safety Department (post-2015).

Outcome measures

Our analysis incorporates a panel design to conduct a time-series generalized least squares regression analysis testing the effect of police layoffs on crime rates (per 100,000 population). Panel models have long been regarded as one of the best methods for estimating causation (Campbell and Stanley, 1967; Hsiao, 1986). Both the NPD and JCPD provided the research team with copies of the monthly Uniform Crime Reports (UCR) submitted to the New Jersey State Police from January 2013 to December 2015. This totaled 240 monthly reports (120 per agency).

In recognition of prior research suggesting the effect of police force size to differ across crime types, we summed the individual crime types to form measures of violent crime (murder, robbery, and aggravated assault) and property crime (burglary, larceny theft, and motor vehicle theft). We also test each disaggregate crime type individually.3 Due to the multiple outcome measures, our analysis incorporated a Holm-Bonferroni correction to adjust critical p. values in order to protect against the increased risk of Type I error that results from tests of multiple hypotheses (Holm, 1979). The Holm-Bonferroni procedure expands upon the traditional Bonferroni method for controlling for multiple statistical comparisons.

The Bonferroni method adjusts the critical p. value through the formula /m\propto /m where \propto is the original critical p. value (e.g. 0.05) and m is the total number of hypotheses tested. Any hypothesis with p.>/m\propto /m is rejected though the traditional Bonferroni approach. Given the highly conservative nature of the Bonferroni method, it minimizes Type I errors while simultaneously increasing the risk for Type II errors (Olejnik et al., 1997). The Holm-Bonferroni method maintains statistical power by establishing different \propto values for the tested hypothesis depending on the observed level of significance. Obtained p. values are first ordered from smallest to largest, p(1),…,p(m), and matched with the corresponding hypotheses, H(1),…,H(m). This Holm-Bonferroni procedure rejects all hypotheses with p(i) > \propto/(m-i+1), protecting against Type I error while maintaining statistical power through the sequential increase of the significance criterion.

Independent variables

A binary measure identifies each observation as pertaining to Newark (“1”) or Jersey City (“0”). This variable was then used to create an interaction term to operationalize difference-in-differences (DiD) measures to reflect the manner by which the police layoffs were implemented in Newark. The DiD term interacted “Newark” with the post-layoff period (December 2010 – December 2015) to measure how the termination of officers impacted crime levels. The DiD interaction term was used as the main independent variable of interest in all models.

We sought to control for potential confounding factors by introducing 15 additional variables in our statistical models. The lagged outcome measure (t – 1) and year variables control for prior levels of crime and the year of occurrence (which may reflect changes in yearly policy occurrences unrelated to the layoffs), respectively (e.g., Braga et al., 2012; Sampson et al., 1997). A continuous variable measuring the sequential order of the monthly time periods (January 2006 =1,…, December 2015 = 120) accounts for the linear trends in the data. The number of days in the month controls for the differing time spans of our analytical units. A dichotomous variable measures whether the month falls in the summer (June – August) to control for the time of year when crime tends to peak. Finally, we included the 10 aforementioned sociodemographic variables in acknowledgment that they are not identical in Newark and Jersey City: percent Black population, percent White population, percent Hispanic population, percent unemployed, median household income, percent under the poverty level, percent of adults with a college degree, percent of properties that are vacant, percent of properties that are owner occupied, and median home value. For both Newark and Jersey City, each sociodemographic variable was standardized to reflect its level relative to the average city-wide value observed throughout the study period.

Results

In Jersey City, violent crime steadily decreased over the 10-year study period. While Jersey City’s property crime rate slightly increased from 2007 to 2008, steady declines occurred in the subsequent years with all years following the NPD layoffs exhibiting property crime rates that were lower than any year prior to the layoff period. This contrasts with the crime trends in Newark. While property crime rates from 2013-2015 were generally lower than prior to the NPD layoffs, violent crime rates during this time frame were higher in Newark than any year prior to the NPD layoffs (see Figure 3).4

Figure 3: Violent crime and property crime rates (per 100,000 population).

[CHART; see end of this Pub]

[CHART; see end of this Pub]

Disaggregate crime types further show diverging trends in Newark and Jersey City (see Figures 4 and 5). In Jersey City, robbery and aggravated assault steadily declined over the 10-year study period while the murder rate remained relatively consistent. In Newark, robbery saw a drastic increases beginning around the time of the NPD layoffs. The robbery rate increase continued the first few years following the layoffs and, while robbery rates dropped after 2013, they were substantially higher than any year prior to the NPD layoffs. Changes in aggravated assault and murder in Newark were less pronounced over the study period. Burglary, larceny theft, and motor vehicle theft steadily decreased in Jersey City throughout the study period, with the steepest decrease observed for larceny theft. The observed trend was much more volatile in Newark. Motor vehicle theft steadily declined in Newark from 2006-2009 before sharply increasing following the NPD layoffs. However, motor vehicle rates from 2013-2015 were lower in Newark than any year prior to the to the NPD layoffs. Larceny theft rates exhibited much less pronounced decreases from 2006-2009 and increased following the NPD layoffs. Burglary rates remained steady from 2006-2009 before sharply increasing following the layoffs. However, beginning in 2012, burglary rates steadily decreased.

Figure 4: Violent crime rates (per 100,000 population).

[CHART; see end of this Pub]

[CHART; see end of this Pub]

Figure 5: Property crime rates (per 100,000 population).

[CHART; see end of this Pub]

[CHART; see end of this Pub]

Tables 1-8 display the results of the generalized least square regression models.5 In each model, the main covariate of interest is the DiD term interacting Newark with the post-NPD layoffs time-period (Newark x Post Layoffs). Changes are considered statistically significant when observed p. values fall below the corrected \propto values generated through the Holm-Bonferroni procedure.6 For both Newark and Jersey City, all monthly crime rates (i.e. the dependent variable) were standardized to reflect their level relative to the average city-wide rate observed across all 120 months in the study period. This was done to facilitate interpretation of the interaction term, with b values representing changes in the dependent variable in terms of standard deviation increases or decreases.

Table 1 and Table 2 report the results of the aggregate violent crime and property crime models, respectively. Following the NPD layoffs, the violent crime rate significantly increased by 1.06 standard deviations in Newark as compared to Jersey City (see Table 1). While property crime rates increases by 0.32 standard deviations in Newark, this finding did not maintain statistical significance following the Holm-Bonferroni correction (see Table 2).

Table 1: Time Series GLS Regression Model results for standardized monthly violent crime rate. 2006 – 2015

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

1.06

0.22

4.74

0.00*

0.62

1.50

Newark

-0.52

0.13

-3.90

0.00

-0.79

-0.26

Post layoffs

-0.40

0.26

-1.56

0.12

-0.90

0.10

Lagged Outcome

0.37

0.05

6.79

0.00

0.26

0.48

Days in Month

0.25

0.04

5.50

0.00

0.16

0.34

Summer

0.14

0.09

1.66

0.10

-0.03

0.31

Year

-0.56

0.16

-3.50

0.00

-0.87

-0.25

Sequential order

0.04

0.01

3.30

0.00

0.01

0.06

% Black population

-0.09

0.07

-1.29

0.20

-0.23

0.05

% White population

0.00

0.05

-0.08

0.94

-0.10

0.09

% Hispanic population

-0.05

0.16

-0.33

0.74

-0.36

0.26

% Unemployed

-0.10

0.06

-1.66

0.10

-0.23

0.02

Median income

-0.19

0.07

-2.73

0.01

-0.33

-0.05

% Under poverty

0.04

0.11

0.39

0.70

-0.18

0.27

% College

0.03

0.14

0.23

0.82

-0.24

0.30

% Vacant properties

0.02

0.05

0.32

0.75

-0.08

0.12

% Owner occupied

0.09

0.14

0.65

0.52

-0.18

0.35

Median home value

-0.29

0.12

-2.51

0.01

-0.52

-0.06

Wald (X)2

543.54

Log likelihood

-192.49

 

 

 

 

 

N= 238

*Interaction term statistically significant following Holm-Bonferroni correction.

Table 2: Time Series GLS Regression Model results for standardized monthly property crime rate. 2006 – 2015

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

0.32

0.15

2.11

0.04

0.02

0.63

Newark

-0.17

0.09

-1.82

0.07

-0.35

0.01

Post layoffs

-0.07

0.19

-0.38

0.71

-0.44

0.30

Lagged Outcome

0.49

0.05

10.46

0.00

0.39

0.58

Days in Month

0.39

0.03

11.41

0.00

0.32

0.46

Summer

0.36

0.06

5.59

0.00

0.23

0.49

Year

0.02

0.12

0.16

0.87

-0.22

0.26

Sequential order

-0.01

0.01

-0.69

0.49

-0.02

0.01

% Black population

0.00

0.05

0.09

0.93

-0.10

0.11

% White population

-0.08

0.04

-2.31

0.02

-0.15

-0.01

% Hispanic population

-0.18

0.12

-1.51

0.13

-0.41

0.05

% Unemployed

0.05

0.05

1.13

0.26

-0.04

0.14

Median income

0.02

0.05

0.35

0.72

-0.08

0.12

% Under poverty

0.09

0.08

1.03

0.30

-0.08

0.25

% College

-0.14

0.10

-1.36

0.18

-0.33

0.06

% Vacant properties

-0.06

0.04

-1.48

0.14

-0.13

0.02

% Owner occupied

0.01

0.10

0.11

0.91

-0.18

0.21

Median home value

0.06

0.08

0.76

0.45

-0.10

0.22

Wald (X)2

1168.27

Log likelihood

-120.08

 

 

 

 

 

N= 238

Three of the six disaggregate crime types exhibited statistically significant effects following the police layoffs (see Tables 3-5). Murder, robbery, and larceny theft increased 0.92, 1.06, and 0.66 standard deviations in Newark as compared to Jersey City, respectively. Burglary increased in Newark by 0.51 standard deviations, but this finding did not maintain statistical significance following the Holm-Bonferroni correction (see Table 6). DiD interaction terms for aggravated assault (Table 7) and motor vehicle theft (Table 8) did not approach statistical significance.

Table 3: Time Series GLS Regression Model results for standardized monthly murder rate. 2006 – 2015.

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

0.92

0.34

2.70

0.01*

0.25

1.59

Newark

-0.47

0.21

-2.22

0.03

-0.88

-0.06

Post layoffs

-0.21

0.43

-0.49

0.62

-1.06

0.63

Lagged Outcome

0.10

0.06

1.62

0.10

-0.02

0.23

Days in Month

0.13

0.08

1.70

0.09

-0.02

0.28

Summer

0.37

0.14

2.64

0.01

0.09

0.64

Year

-0.50

0.26

-1.89

0.06

-1.02

0.02

Sequential order

0.03

0.02

1.42

0.16

-0.01

0.06

% Black population

0.27

0.12

2.24

0.03

0.03

0.51

% White population

-0.02

0.08

-0.25

0.81

-0.18

0.14

% Hispanic population

0.60

0.27

2.23

0.03

0.07

1.13

% Unemployed

0.02

0.10

0.15

0.88

-0.19

0.22

Median income

0.09

0.11

0.82

0.41

-0.13

0.32

% Under poverty

-0.16

0.19

-0.81

0.42

-0.54

0.22

% College

0.08

0.23

0.35

0.73

-0.37

0.53

% Vacant properties

-0.13

0.09

-1.57

0.12

-0.30

0.03

% Owner occupied

-0.15

0.23

-0.67

0.50

-0.60

0.30

Median home value

0.16

0.19

0.83

0.41

-0.21

0.52

Wald (X)2

46.17

Log likelihood

-316.45

 

 

 

 

 

N= 238

*Interaction term statistically significant following Holm-Bonferroni correction.

Table 4: Time Series GLS Regression Model results for standardized monthly robbery rate. 2006 – 2015.

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

1.06

0.21

5.08

0.00*

0.65

1.47

Newark

-1.12

0.11

-9.86

0.00

-1.35

-0.90

Post layoffs

-0.36

0.23

-1.56

0.12

-0.81

0.09

Lagged Outcome

0.03

0.00

8.56

0.00

0.02

0.04

Days in Month

0.21

0.04

5.07

0.00

0.13

0.29

Summer

0.01

0.08

0.14

0.89

-0.14

0.16

Year

-0.68

0.14

-4.76

0.00

-0.96

-0.40

Sequential order

0.05

0.01

5.01

0.00

0.03

0.07

% Black population

-0.14

0.06

-2.19

0.03

-0.27

-0.02

% White population

-0.01

0.04

-0.18

0.86

-0.10

0.08

% Hispanic population

0.02

0.14

0.15

0.88

-0.26

0.31

% Unemployed

-0.10

0.06

-1.75

0.08

-0.21

0.01

Median income

-0.18

0.06

-2.80

0.01

-0.30

-0.05

% Under poverty

0.15

0.10

1.49

0.14

-0.05

0.36

% College

0.03

0.12

0.27

0.79

-0.21

0.28

% Vacant properties

0.04

0.05

0.90

0.37

-0.05

0.13

% Owner occupied

0.25

0.12

2.04

0.04

0.01

0.50

Median home value

-0.21

0.10

-2.06

0.04

-0.42

-0.01

Wald (X)2

693.50

Log likelihood

-169.27

 

 

 

 

 

N= 238

*Interaction term statistically significant following Holm-Bonferroni correction.

Table 5: Time Series GLS Regression Model results for standardized monthly larceny theft rate. 2006 – 2015.

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

0.66

0.21

3.06

0.00*

0.24

1.08

Newark

-0.36

0.13

-2.72

0.01

-0.61

-0.10

Post layoffs

-0.23

0.26

-0.90

0.37

-0.74

0.27

Lagged Outcome

0.38

0.05

7.18

0.00

0.28

0.48

Days in Month

0.34

0.05

7.31

0.00

0.25

0.43

Summer

0.43

0.09

4.77

0.00

0.25

0.60

Year

-0.02

0.17

-0.12

0.90

-0.35

0.31

Sequential order

0.00

0.01

-0.02

0.98

-0.02

0.02

% Black population

0.05

0.07

0.72

0.47

-0.09

0.19

% White population

-0.03

0.05

-0.65

0.52

-0.13

0.07

% Hispanic population

-0.03

0.16

-0.18

0.85

-0.35

0.29

% Unemployed

0.01

0.06

0.10

0.92

-0.12

0.13

Median income

0.04

0.07

0.56

0.57

-0.10

0.18

% Under poverty

0.01

0.12

0.07

0.95

-0.22

0.24

% College

-0.20

0.14

-1.40

0.16

-0.47

0.08

% Vacant properties

-0.03

0.05

-0.65

0.51

-0.14

0.07

% Owner occupied

0.16

0.14

1.15

0.25

-0.11

0.43

Median home value

-0.03

0.11

-0.25

0.80

-0.25

0.20

Wald (X)2

491.84

Log likelihood

-198.61

 

 

 

 

 

N= 238

Table 6: Time Series GLS Regression Model results for standardized monthly burglary rate. 2006 – 2015.

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

0.51

0.22

2.26

0.02

0.07

0.94

Newark

-0.25

0.14

-1.84

0.07

-0.52

0.02

Post layoffs

0.01

0.27

0.03

0.98

-0.53

0.54

Lagged Outcome

0.47

0.05

8.67

0.00

0.36

0.57

Days in Month

0.33

0.05

6.65

0.00

0.24

0.43

Summer

0.27

0.09

3.00

0.00

0.09

0.45

Year

-0.16

0.17

-0.94

0.35

-0.50

0.18

Sequential order

0.01

0.01

0.70

0.49

-0.02

0.03

% Black population

-0.17

0.08

-2.18

0.03

-0.32

-0.02

% White population

-0.01

0.05

-0.27

0.79

-0.12

0.09

% Hispanic population

-0.42

0.18

-2.34

0.02

-0.77

-0.07

% Unemployed

0.08

0.07

1.27

0.21

-0.05

0.21

Median income

0.10

0.07

1.39

0.17

-0.04

0.24

% Under poverty

0.35

0.13

2.77

0.01

0.10

0.60

% College

0.01

0.15

0.07

0.94

-0.28

0.30

% Vacant properties

0.03

0.05

0.59

0.55

-0.07

0.14

% Owner occupied

0.15

0.15

1.01

0.31

-0.14

0.44

Median home value

-0.04

0.12

-0.37

0.71

-0.28

0.19

Wald (X)2

426.72

Log likelihood

-210.58

 

 

 

 

 

N= 238

Table 7: Time Series GLS Regression Model results for standardized monthly aggravated assault rate. 2006 – 2015.

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

0.25

0.24

1.05

0.29

-0.22

0.72

Newark

-0.12

0.15

-0.84

0.40

-0.41

0.16

Post layoffs

-0.43

0.30

-1.42

0.16

-1.03

0.16

Lagged Outcome

0.19

0.07

2.70

0.01

0.05

0.32

Days in Month

0.17

0.05

3.22

0.00

0.07

0.28

Summer

0.38

0.11

3.48

0.00

0.17

0.60

Year

-0.05

0.19

-0.24

0.81

-0.42

0.33

Sequential order

0.00

0.01

-0.18

0.85

-0.03

0.02

% Black population

0.07

0.08

0.79

0.43

-0.10

0.23

% White population

0.04

0.06

0.71

0.48

-0.07

0.16

% Hispanic population

-0.35

0.19

-1.81

0.07

-0.72

0.03

% Unemployed

-0.16

0.07

-2.18

0.03

-0.31

-0.02

Median income

-0.27

0.08

-3.29

0.00

-0.43

-0.11

% Under poverty

-0.31

0.14

-2.24

0.03

-0.59

-0.04

% College

-0.09

0.16

-0.53

0.60

-0.40

0.23

% Vacant properties

0.01

0.06

0.17

0.86

-0.11

0.13

% Owner occupied

-0.27

0.16

-1.67

0.09

-0.59

0.05

Median home value

-0.42

0.14

-3.01

0.00

-0.69

-0.15

Wald (X)2

327.96

Log likelihood

-233.30

 

 

 

 

 

N= 238

Table 8: Time Series GLS Regression Model results for standardized monthly motor vehicle theft rate. 2006 – 2015.

 

 

 

 

 

95% C.I.

Covariates

b

S.E.

z

p.

Lower

Upper

Newark x Post layoffs

0.07

0.17

0.39

0.70

-0.27

0.40

Newark

-0.03

0.10

-0.26

0.80

-0.23

0.18

Post layoffs

0.11

0.21

0.53

0.60

-0.31

0.53

Lagged Outcome

0.38

0.05

7.07

0.00

0.27

0.48

Days in Month

0.29

0.04

7.40

0.00

0.21

0.36

Summer

0.29

0.07

4.15

0.00

0.15

0.43

Year

-0.08

0.13

-0.63

0.53

-0.35

0.18

Sequential order

0.00

0.01

-0.28

0.78

-0.02

0.02

% Black population

0.06

0.06

1.03

0.30

-0.05

0.18

% White population

-0.13

0.04

-3.04

0.00

-0.21

-0.04

% Hispanic population

0.01

0.13

0.07

0.94

-0.25

0.27

% Unemployed

-0.01

0.05

-0.11

0.91

-0.11

0.09

Median income

-0.10

0.06

-1.81

0.07

-0.22

0.01

% Under poverty

-0.02

0.10

-0.20

0.84

-0.21

0.17

% College

-0.20

0.11

-1.70

0.09

-0.42

0.03

% Vacant properties

-0.09

0.04

-1.98

0.05

-0.17

0.00

% Owner occupied

-0.16

0.11

-1.43

0.15

-0.39

0.06

Median home value

0.20

0.10

2.07

0.04

0.01

0.38

Wald (X)2

849.07

Log likelihood

-150.76

 

 

 

 

 

N= 238

Discussion and Conclusion

The current study contributes to the literature on the effect of police force size on crime. Our findings indicate that drastic reductions in police force size via police officer layoffs can generate significant crime increases. In contextualizing the differences between findings of the current study and the prior literature, it is important to note that increases in police force size observed in prior research have typically been incremental. For example, Worrall and Kovandzic (2007) found that COPS Office hiring grants, which were seen as a vehicle to increase manpower throughout the U.S., provided funding that represented less than one half of 1% of recipient police agency budgets. This incremental addition of police officers may help explain why prior increases in police force size have not significantly impacted crime.

Conversely, research on decreases in police force size have tended to be much more drastic than the incremental changes seen with increases. Research on police strikes show that crime significantly increases when a large proportion of police officers fail to report for duty (see Sherman and Eck, 2002: 302). While the reduction of officers in Newark was not as large as those associated with police strikes, the change is force size was sudden, with 13% of the force terminated on the same day. We find it important to further acknowledge that the termination of officers can generate poor morale amongst the entirety of a police force, which may directly impact the performance of individual officers and, tangentially, crime levels. The effect of layoffs on employee effectiveness was recently illustrated in a study conducted by Strunk et al. (2018), who found that teacher layoffs negatively impacted the performance and job commitment of teachers who were ultimately retained by the school district in Washington State. While we are unaware of any research that looks directly at this issue in policing, we find it reasonable to assume that police officers may be similarly vulnerable to stress introduced by the termination of their colleagues.

Weisburd and Telep (2010) note that in a time of declining resources, police are forced to do more with scarce resources while still producing results in the most efficient ways. Evidence-based policing allows officers to control crime and disorder in ways that are more effective and less costly than traditional response driven models (Bueermann, 2012). However, while these practices enjoy evidence of effectiveness, and may be more cost-effective than traditional police practices (Bueermann, 2012), they can only be implemented when agencies have the required resources to do so. A survey conducted by the Police Executive Research Forum (PERF) (2010) during the recession found that 51% of reporting police agencies suffered cuts in their total funding and that 59% of those agencies anticipated their budgets would see additional cuts over subsequent years. In the light of such cutbacks, committing to evidence-based practices may be challenging.

To maximize officer time in hot spots, the NPD would oftentimes deploy teams of officers with little to no responsibly for responding to 9-1-1 calls for service. These officers were primarily tasked with proactively policing Newark’s crime hot spots for the majority of their shifts. Taken as a whole, hot spots policing activities were a mainstay throughout Newark from 2006 through 2009, which corresponded to general crime rate decreases. The threat of the impending layoffs led the NPD to phase out the hot spots policing program in anticipation of the termination of the officers who staffed these units. The situation in Newark may not be unique, as the disbandment of specialized units is a common response to budget cuts. For example, a report from the National Institute of Justice (Wiseman, 2011) offered unit disbandment as a cutback management strategy, citing the disbandment of mounted patrol divisions in both Boston and San Diego in response to budget cuts. While such unit contraction alleviates costs associated with running the units, the police agency obviously loses any crime control functions of these units. This may have been the case in Newark, as units solely dedicated to proactive hot spots operations were disbanded in anticipation of the layoffs.

In considering the findings of our analysis, one may ask whether the crime increase could have been averted if the NPD remained committed to its hot spots policing program. The NPD may have continued to emphasize place-based policing by tasking their regular patrol officers with patrolling hot spots in between calls for service. This is especially the case in consideration of the Koper curve, which finds that crime reductions can occur with police patrolling hot spots in intervals as short as 15 minutes (Koper, 1995; Telep et al., 2014). However, conducting hot spots policing in such a manner may be easier said than done in a contemporary police agency dealing with impending officer layoffs. Given the time intensive nature of activities associated with the standard patrol activities, such as writing paperwork, transporting arrestees, and responding to non-crime calls for service, patrol officers may not have sufficient discretionary time to conduct proactive hot spots policing activities. This may especially be the case in police agencies suffering from police officer layoffs. For such agencies, adhering to even a Koper curve model of hot spots policing may be challenging. In such a case, police officer size and incorporating evidence-based policing strategies may not be mutually exclusive propositions. A study by the Police Executive Research Forum (2010) conducted during the recession found that many departments reduced the scope of many of their community services due to budget cuts. This exemplifies how budget constraints can negatively impact police operations, which directly relates to the issue of officer layoffs.

Despite the implications of the findings, this study, like most research, suffers from certain limitations that should be mentioned. In particular, we should note prior critiques of official sources of crime data, specifically the UCR, which we used in this study. The UCR is the primary source of crime in the U.S., and provided us a readily available data source to measure the impact of NPD’s officer layoffs. However, the UCR provides an incomplete picture of crime, as Part 1 crimes are the only incidents that are systematically reported to the FBI (Maxfield and Babbie, 2015). Given their reliance on official crime reports, UCR data may be influenced by police officer discretion (Warner and Pierce, 1993) and citizen distrust of the police (Kirk and Matsuda, 2011). Including additional outcome measures, such as calls for service, would have helped to overcome these limitations. Unfortunately, such data was not available to us.

In sum, we feel that the current study positively contributes to the literature in a number of ways. First, we took advantage of naturally occurring phenomenon in order to conduct a natural experiment involving New Jersey’s two largest police forces. We feel that our methodology can inform the work of policing scholars interested in studying the effect of “treatment” conditions that cannot be readily manipulated by researchers. Second, we used the causal-process method (Collier et al., 2010) to contextualize the police layoffs in Newark. This allowed us to more readily explore the potential causal mechanisms of Newark’s crime increase. Third, the current study, to our knowledge, is the first empirical test of the effect of police officer layoffs on crime. While typically a rare occurrence, the economic downtown in the mid 2000’s necessitated that a number of police agencies enact layoffs in order to balance fiscal budgets. We feel this provides an opportunity for a range of natural experiments to better understand the effect of the layoffs. While our focus was on crime, other potential outcomes of interest include office productivity, officer wellbeing, citizen fear of crime, and citizen perceptions of police legitimacy. We hope that criminologists continue to pursue this line of research.

References

Andenaes, J. (1974). Punishment and deterrence (pp. 9-10). Ann Arbor: University of Michigan Press.

Bennett, T., Holloway, K., and Farrington, D. P. (2006). Does neighborhood watch reduce crime? A systematic review and meta-analysis. Journal of Experimental Criminology2(4), 437-458.

Black, D. J. (1971). The Social Organization of Arrest. Stanford Law Review, 1087-1111.

Braga, A. A., and Weisburd, D. L. (2012). The effects of focused deterrence strategies on crime: A systematic review and meta-analysis of the empirical evidence. Journal of Research in Crime and Delinquency49(3), 323-358.

Braga, A., Hureau, D., and Papachristos, A. (2012). An ex post facto evaluation framework for place-based police interventions. Evaluation Review, 35(6), 592–626.

Braga, A. A., Papachristos, A. V., and Hureau, D. M. (2014). The effects of hot spots policing on crime: An updated systematic review and meta-analysis. Justice quarterly31(4), 633 663.

Bueermann, J. (2012). Being smart on crime with evidence-based policing. NIJ journal269, 12 15.

Campbell, D. and Stanley, J. (1967). Experimental and quasi-experimental designs for research. Rand McNally: Chicago, IL.

Chalfin, A., & McCrary, J. (2018). Are US cities underpoliced? Theory and evidence. Review of Economics and Statistics100(1), 167-186.

Clark, G. (1969). What happens when the police strike. New York Times Magazine11, 15-22.

Collier, D., Brady, H., and Seawright, J. (2010). Sources of leverage in causal inference: Toward an alternate view of methodology. In Brady, H. and Collier, D. (eds.) Rethinking Social Inquiry: Diverse Tools, Shared Standards.

Community Oriented Policing Services, Office of (COPS). (2011). The impact of the economic downturn on American police agencies. U.S. Department of Justice: Washington, DC.

Cumming, E., Cumming, I., and Edell, L. (1965). Policeman as philosopher, guide and friend. Social Problems, 12(3), 276-286.

Decarlo, J. and Jenkins, M. (2015). Labor unions, management innovation and organizational change in police departments. Springer Briefs in Policing. Springer: New York.

Dow, J. (2018). Jersey City busting at seams with population surge. Retrieved 5/18/20 from https://pix11.com/2018/11/14/jersey-city-bursting-at-seams-with-population-surge

Dunning, T. (2012). Natural experiments in the social sciences. A Design-Based Approach. Cambridge University Press: Cambridge, UK.

Dunning, T. (2008). Improving causal inference. Strengths and limitations of natural experiments. Political Research Quarterly, 61(2): 282-293.

Farrington, D. P., Gottfredson, D. C., Sherman, L. W., and Welsh, B. C. (2002). The Maryland scientific methods scale. Evidence-based crime prevention, 13-21.

Federal Bureau of Investigation (FBI)(2011a). New Jersey: Offenses known to law enforcement by city, 2010. Table 8 in Crime in the United States. Washington, DC: Uniform Crime Reporting Program.

Federal Bureau of Investigation (FBI)(2011b). New Jersey: Full-time law enforcement employees by city, 2010. Table 78 in Crime in the United States. Washington, DC: Uniform Crime Reporting Program.

Ford, J. (2010). Laid off Newark police turn in guns, badges. McClatchy-Tribune Business News.

Giblin, M. J., and Nowacki, J. S. (2018). Organizational decline and fiscal distress in municipal police Agencies. Police Quarterly21(2), 171-195.

Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics, 6: 65-70.

Hsiao, C. (1986). Analysis of panel data. Cambridge University Press: New York.

Jenkins, M. and DeCarlo, J. (2015). Police Leaders in the New Community Problem-Solving Era. Carolina Academic Press: Durham, North Carolina.

Kaplan, J., & Chalfin, A. (2019). More cops, fewer prisoners? Criminology & Public Policy18(1), 171-200.

Kirk, D. S., and Matsuda, M. (2011). Legal cynicism, collective efficacy, and the ecology of arrest. Criminology, 49(2), 443–472.

Koper, C. (1995). Just enough police presence: Reducing crime and disorderly behavior by optimizing patrol time in crime hot spots. Justice Quarterly, 12(4), 649-672.

Lee, Y., Eck, J. E., and Corsaro, N. (2016). Conclusions from the history of research into the effects of police force size on crime—1968 through 2013: A historical systematic review. Journal of Experimental Criminology12(3), 431-451.

Lum, C. and Koper, C. (2017). Evidence-based policing: Translating research into practice. Oxford, UK: Oxford University Press.

Lum, C., Koper, C. S., and Telep, C. W. (2011). The evidence-based policing matrix. Journal of Experimental Criminology7(1), 3-26.

MacDonald, J. M., & Stokes, R. J. (2020). Gentrification, land use, and crime. Annual Review of Criminology3, 121-138.

Makinen, T. and Takala, H. (1980). The 1976 police strike in Finland. Scandinavian Studies in Criminology, 7(1), 87-106.

Maxfield, M. and Babbie, E. (2015). Research methods for criminal justice and criminology. Third edition. Wadsworth/Thompson Learning: Belmont, CA.

Mello, S. (2019). More COPS, less crime. Journal of Public Economics172, 174-200.

Mitchell, O. (2015). Experimental research design. In W. G. Jennings (Ed.), The encyclopedia of crime and punishment. Malden, MA: Wiley-Blackwell doi:10.1002/9781118519639.wbecpx113.

Olejnik, S., Li, J., Supattathum, S., & Huberty, C. (1997). Multiple testing and statistical power with modified Bonferroni procedures. Journal of Educations and Behavioral Statistics, 22: 389-406.

Payne, T. C. (2017). How do you determine the right size of a police department: Don't look to crime rates. Alaska Justice Forum34, 3.

Pfuhl Jr, E. H. (1983). Police strikes and conventional crime: A look at the data. Criminology,

21(4), 489-504.

Piza, E., Wheeler, A., Connealy, N. and Feng, S. (2020). Crime control effects of a police sub-station within a business-improvement district: A quasi-experimental synthetic control evaluation. Criminology & Public Policy, 19(2), 653-684.

Piza, E., and O’Hara, B. (2014). Saturation foot-patrol in a high-violence area: A quasi experimental evaluation. Justice Quarterly31(4), 693-718.

Police Executive Research Forum (PERF). (2010). Is the economic downturn fundamentally changing how we police? Critical Issues in Policing Series. Washington, DC: Police Executive Research Forum.

Porter, D. (2011, January 28). Jersey City bucks recent wave of NJ police layoffs. Desert News.

Sampson, R. J., Raudenbush, S., and Earls, F. (1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918-924.

Sherman, L. W. (1998). Evidence-based policing. Ideas in American Policing. Washington, DC: Police Foundation.

Sherman, L., and Eck, J. (2002). Policing for crime prevention. In L. Sherman, D. Farrington, B. Welsh, and D. Mackenzie (eds.), Evidence-based crime prevention (pp. 295- 329). New York, NY: Routledge

Skogan, W. and Frydl, K. (eds.) (2004). Fairness and effectiveness in policing: The evidence. Committee to review research on police policy and practices. Committee on law and justice, division of behavioral and social sciences and education. Washington, DC: The National Academies Press.

Snow, J. ([1855]1965). On the mode of communication of cholera. John Churchill: London. Reprinted in Snow on Cholera. Humphrey Milford; Oxford University Press: London.

Star Ledger. (2010, November 30). Newark finalizes 167 police layoffs after union refuses Booker’s plea to return to negotiating table.

Strunk, K., Goldhaber, D., Knight, D., and Brown, N. (2018). Are there hidden costs associated with conducting layoffs? The impact of reduction-in-force and layoff notices on teacher effectiveness. Journal of Policy Analysis and Management, 37(4): 755-782.

Telep, C., Mitchell, R., and Weisburd, D. (2014). How much time should the police spend at crime hot spots? Answers from a police agency directed randomized field trial in Sacramento, California. Justice Quarterly, 31(5), 37–41.

Terrill, W., Rossler, M. T., and Paoline III, E. A. (2014). Police service delivery and responsiveness in a period of economic instability. Police Practice and Research15(6), 490-504.

United States Census Bureau. (2010). Quick facts from the United States census bureau. Retrieved 5/15/20 from https://www.census.gov/quickfacts/fact/table/jerseycitycitynewjersey,newarkcitynewjersey/PST045217

Warner, B. D., and Pierce, G. L. (1993). Reexamining social disorganization theory using calls to the police as a measure of crime. Criminology31(4), 493-517.

Weisburd, D. L., and Telep, C. W. (2010). The efficiency of place-based policing. Journal of Police Studies, 17, 247-262.

Weisburd, D., Telep, C. W., Hinkle, J. C., and Eck, J. E. (2010). Is problem-oriented policing effective in reducing crime and disorder? Findings from a Campbell systematic review. Criminology and Public Policy9(1), 139-172.

Wiseman, J. (2011). Strategic cutback management: Law enforcement leadership for lean times. U.S. Department of Justice, Office of Justice Programs, National Institute of Justice.

Worrall, J. L., and Kovandzic, T. V. (2007). COPS grants and crime revisited. Criminology, 45(1), 159-190.

Figures

Comments
0
comment

No comments here