Write a 1,050 word paper describing the types and functions of research designs. Address each of the following in your

Write a 1,050 word paper describing the types and functions of research designs. Address each of the following in your paper:

An explanation of internal and external variable-related validity factors
The strengths and limitations of experimental research designs
A summary of two contemporary examples of experimental research conducted within the criminal justice field within the past 10 years

Format your paper consistent with APA guidelines.

***Create a 2 slide power point on a research article presentation. I have attached the file that contains the article in which the slides will be about.

Slide 1 will discuss:

A of the methodology

Slide 2 will discuss

An evaluation of the strengths and limitations of the research


J Exp Criminol (2014) 10:399–428
DOI 10.1007/s11292-014-9210-y
Community-oriented policing to reduce crime, disorder
and fear and increase satisfaction and legitimacy
among citizens: a systematic review
Charlotte Gill & David Weisburd & Cody W. Telep &
Zoe Vitter & Trevor Bennett
Published online: 2 August 2014
# Springer Science+Business Media Dordrecht 2014
Abstract
Objectives Systematically review and synthesize the existing research on communityoriented policing to identify its effects on crime, disorder, fear, citizen satisfaction, and
police legitimacy.
Methods We searched a broad range of databases, websites, and journals to identify
eligible studies that measured pre-post changes in outcomes in treatment and comparison areas following the implementation of policing strategies that involved community
collaboration or consultation. We identified 25 reports containing 65 independent tests
of community-oriented policing, most of which were conducted in neighborhoods in
the United States. Thirty-seven of these comparisons were included in a meta-analysis.
Results Our findings suggest that community-oriented policing strategies have positive
effects on citizen satisfaction, perceptions of disorder, and police legitimacy, but limited
effects on crime and fear of crime.
Conclusions Our review provides important evidence for the benefits of community
policing for improving perceptions of the police, although our findings overall are
ambiguous. The challenges we faced in conducting this review highlight a need for
Electronic supplementary material The online version of this article (doi:10.1007/s11292-014-9210-y)
contains supplementary material, which is available to authorized users.
C. Gill (*) : D. Weisburd : Z. Vitter
Center for Evidence-Based Crime Policy, Department of Criminology, Law and Society, George Mason
University, 4400 University Dr. MS6D12, Fairfax, VA 22030, USA
e-mail: cgill9@gmu.edu
D. Weisburd
Institute of Criminology, Faculty of Law, Hebrew University, Jerusalem, Israel
C. W. Telep
School of Criminology and Criminal Justice, Arizona State University, Phoenix, AZ, USA
T. Bennett
Centre for Criminology, University of South Wales, Pontypridd, UK
400
C. Gill et al.
further research and theory development around community policing. In particular,
there is a need to explicate and test a logic model that explains how short-term benefits
of community policing, like improved citizen satisfaction, relate to longer-term crime
prevention effects, and to identify the policing strategies that benefit most from
community participation.
Keywords Community policing . Crime prevention . Evaluation research .
Legitimacy . Meta-analysis . Problem solving . Systematic review
Introduction
Community-oriented policing (COP) is a philosophy of policing that emphasizes
community involvement in crime prevention efforts, in contrast to the focus of traditional policing on law enforcement and order maintenance. Previous narrative reviews
have found limited effects of COP on reducing crime, but suggest that it may have
benefits for other outcomes, including citizen satisfaction and trust in the police
(Sherman and Eck 2002; Skogan and Frydl 2004; Weisburd and Eck 2004). There
are a number of challenges in assessing the effectiveness of COP, including a broad
interpretation of its scope and, subsequently, substantial heterogeneity in the types of
strategies that are classified as COP, and the lack of a clear logic model or accepted
structure for implementation. We therefore conducted a systematic review to attempt to
identify and synthesize the various approaches that have been classified as COP.
This study sought to investigate the extent to which community-oriented policing
impacts crime, disorder, fear of crime, citizen satisfaction with police, and police
legitimacy, as well as identify the most effective strategies police can use in collaboration with the community to prevent crime. Our results reinforce earlier recognition of
the benefits of community policing for increasing citizen satisfaction and trust in the
police. Community policing strategies are also associated with a significant reduction in
citizens’ perceptions of disorder. In line with other reviews, we do not find an overall
significant effect on crime, and COP does not appear to reduce fear of crime.
Identifying the specific components that are implemented in collaboration with the
community, such as problem solving, may be key to refining COP as an intervention
and assessing its effectiveness. We begin with a review of the prior literature on
community-oriented policing and some of the challenges of conceptualizing and
measuring COP interventions, before discussing the systematic review and metaanalysis methodology and results. We conclude with a discussion of the implications
of our findings for future research and the refinement of the community-oriented
policing model.
Background
Community-oriented policing (COP) is a law enforcement philosophy comprising three
key components: community partnerships, organizational transformation, and problem
solving (Office of Community Oriented Policing Services 2012; Skogan 2006a). COP
is based on the premise that the police are not limited to traditional law enforcement
Community-oriented policing: a systematic review
401
powers in carrying out their work, and should draw on community involvement and
input to define, prioritize, and address crime problems (Weisburd and McElroy 1988).
This approach requires the traditional hierarchy of the police department to be ‘flattened’ in order to delegate decision making to the frontline officers who directly engage
the community (Cordner 1999; Office of Community Oriented Policing Services 2012;
Trojanowicz et al. 1998; Weisburd et al. 2003a). Thus, COP is not simply about
improving relationships between police and citizens. It is also a problem-solving
process that draws upon citizens’ expertise in identifying and understanding the social
issues that create crime, disorder, and fear (Trojanowicz et al. 1998). Community
members and the police officers who serve them, as well as the police organization
more broadly, are therefore the ‘co-producers’ of public safety. In this way, COP is
distinct from problem-oriented policing (POP; see Clarke 2002; Goldstein 1990; Scott
2000; Weisburd et al. 2010). The primary focus of POP is to find effective solutions to
problems, which may or may not involve the community, while problem solving within
the COP framework emphasizes community outreach and engagement and does not
necessarily rely on thorough problem analysis approaches such as the SARA model.
COP is one of several policing innovations that became widely adopted in the 1990s
after several decades of dissatisfaction with ‘standard’ police practices. Traditional
policing approaches in the United States, also described as the ‘professional model of
policing,’ have consisted of generic, reactive strategies to prevent or respond to crime
(e.g., Weisburd and Eck 2004). These strategies largely emphasize inputs and outputs
(for example, resource management and meeting targets) rather than longer-term
outcomes like effectiveness, legitimacy of the police, and citizen satisfaction.
Standard policing can also be insular, relying heavily on traditional law enforcement
powers with very little input from or collaboration with non-police institutions and the
community.
During the 1970s, rising crime and challenges to the effectiveness of a broad
range of criminal justice practices (e.g., Martinson 1974) led to criticisms of the
standard model of policing. The criminal justice system more generally had been
criticized as being out of touch with and unsupported by the communities it served
(see, e.g., the recommendations of the President’s Commission on Law Enforcement
and Criminal Justice 1967). Around the same time, several high-profile research
studies suggested that two key elements of the standard model of policing, preventive
patrol and rapid response, had little impact on crime rates (Kelling et al. 1974;
Spelman and Brown 1984; see also Weisburd and Braga 2006). Much as Martinson’s
findings about the limitations of research on rehabilitation led to “nothing works”
becoming the defining characteristic of the entire field, these studies raised questions
about the effectiveness of the police in general. By the 1990s, many scholars believed
that the police could do little to impact crime (e.g., Bayley 1994; Goldstein 1990;
Gottfredson and Hirschi 1990).
This “crisis” in American policing arguably provided the foundation for the changes
and innovations that led to developments such as COP (Weisburd and Braga 2006).
Community-oriented policing recognized what scholars had begun to discover in
observational studies (e.g., Reiss 1971), that much of the police role did not involve
crime fighting. Rather, police work involved order maintenance, service provision,
reduction of fear, and conflict resolution (Kelling and Moore 1988; Skogan and Frydl
2004; Skogan and Hartnett 1997; Weisburd and Braga 2006). Community-oriented
402
C. Gill et al.
policing advocates brought to the fore aspects of policing that had been deemphasized
through the characterization of the police solely as crime fighters. They also responded
to the crisis in community support and legitimacy of policing at the time by placing the
community as central to the police task, recognizing the community in a broad sense as
the foundation of social institutions like the police and criminal justice system
(Sherman 1997a).
COP has become extremely popular, particularly in the United States but also in
other European countries (notably the United Kingdom) and Australia. In the U.S.,
Skogan and Frydl (2004: 232) report a major investment in COP by police agencies and
communities in the two decades prior to their review (see also Skogan 2006a; Weisburd
and Eck 2004). In a 1997 Police Foundation survey, all police departments in U.S.
municipalities with populations greater than 100,000 who responded reported that they
had adopted COP, with 85 % of the total sample claiming that they had adopted or
planned to adopt it (Skogan 2004). Similarly, in a later survey, over 90 % of police
departments in large urban areas indicated that they employed fully trained communityoriented policing officers (Hickman and Reaves 2001; see also Mastrofski et al. 2007).
Much of this popularity in the United States was driven by the creation of the COPS
Office (the U.S. Department of Justice’s Office of Community Oriented Policing
Services) in 1994. The COPS Office has provided significant funding to encourage
agencies to implement COP. However, more recently, there has been an overall decline
in the percentage of departments using full-time community-oriented policing officers.
This drop-off has been most pronounced in smaller departments serving populations of
less than 50,000 (Reaves 2010). In the United States at least, the impact of the
economic downturn on the budgets of police departments and federal agencies may
be partly responsible for the lower levels of investment compared to a decade ago, but
some more fundamental challenges with the implementation of COP may also have
played a role.
Stone and Travis (2011) suggest that COP has lost some of its momentum because
some police departments felt unsure of what to ask of communities and felt the
“transformative” power of the program did not live up to expectations. Mastrofski
et al. (2007), in a national survey of U.S. police leaders, found that over half of
respondents found obtaining sufficient resources to “do [community-oriented policing]
right” very or extremely challenging, and over 40 % found it difficult to obtain the
support of frontline officers for COP activities (see also Skogan and Hartnett 1997).
The challenge may lie in the fact that, unlike other police innovations like hot spots or
problem-oriented policing, COP is a philosophy or guiding framework for
implementing strategies, and not a strategy in itself. Trojanowicz et al. (1998), in a
revised edition of their seminal book on community-oriented policing, state that
“[c]ommunity policing is not just a tactic that can be applied to solve a particular
problem, one that can be abandoned once the goal is achieved.” Weisburd et al. (2003b)
also suggest that COP’s demand for decentralization presented a threat to the traditional
military model of policing, which may have led police to look to other innovations such
as Compstat.
Furthermore, there are no specific criteria for implementing COP because the
mission of each police department is assumed to be guided by the community it serves
(Morabito 2010). Thus, the organizational transformation element of COP is crucial.
Yet, it appears that few police departments that claim to be ‘doing COP’ have
Community-oriented policing: a systematic review
403
implemented all three key components (Morabito 2010; Trojanowicz et al. 1998). COP
is often introduced at the unit level, as a set of tactics employed by individual police
officers, in specific beat areas, or by specialized teams (e.g., Weisburd et al. 1988). A
diverse range of strategies have been employed under the auspices of COP over the past
few decades—and have fallen in and out of favor as community-oriented strategies
during that time—including problem-oriented policing (which, as we note above, does
not necessarily require community collaboration), foot patrol, newsletters, door-to-door
surveys, education programs in schools, neighborhood watch, and multi-agency partnerships with municipal organizations and community members (Mastrofski et al.
1995; Skogan 2006b; Weisburd and Eck 2004).
Mastrofski et al. (2007: 224) note that the resulting heterogeneity renders the COP
approach “vague and difficult to execute,” and that many of these tactics have not been
rigorously tested (see also Eck and Rosenbaum 1994; Skogan and Frydl 2004).
Weisburd and Eck (2004) found in a narrative review of community-oriented policing
studies that the different strategies appear to have variable effects on crime. For
example, community meetings, foot patrol, and providing information about crime
prevention to the public had little effect on crime, while door-to-door visits reduced
crime and fear of crime, and general improvements in police-community interactions
also reduced fear and concern about crime.
The outcomes of COP programs may also have proved disappointing because there
is no clear theory of change linking police–community collaboration to reduced crime.
While crime reduction is now a desired goal of COP, it was not a key reason for the
adoption of community-oriented policing (Skogan 2006a)—as discussed above, COP
emerged from a desire to broaden the mandate of the police at a time when their
effectiveness at controlling crime was in doubt. Community-oriented policing identified a host of roles for the police beyond crime fighting, including reducing fear,
responding to social and physical disorder in the community, and forging positive
relationships with residents to enhance police legitimacy, while rejecting law enforcement as the “core function” of police (Mastrofski et al. 1995). Crime control as a key
outcome measure of COP interventions most likely arose from the growth of
community-oriented policing in the 1990s, when the COPS Office spurred the development of programs that treated crime as a key issue (see also Klockars 1985). Skogan
(2006a) notes that government agencies are concerned with crime control and want
COP to be one tool for that purpose, but no police department likely adopted COP as
the primary method of reducing crime and disorder. This may explain why prior
narrative reviews (e.g., Sherman and Eck 2002; Skogan and Frydl 2004; Weisburd
and Eck 2004) suggest that, despite the massive investment in COP, its impact on crime
has been somewhat limited.
On the other hand, COP may have a positive impact on non-crime control outcomes
related to community relations and trust in the police, such as fear of crime, legitimacy,
and satisfaction with policing. The community-oriented approach, which promotes
positive relationships and collaboration between police and citizens over police imposition of enforcement and control over citizens, may foster procedural justice (fair,
consistent, respectful, and accountable policing), enhanced trust in police, and higher
citizen ratings of their performance. Given that citizens who trust and accept the
authority of the police are more likely to obey the law (Sunshine and Tyler 2003;
Tyler 1990), citizens’ perceptions of legitimacy and police effectiveness may be
404
C. Gill et al.
important antecedents to crime control. Some recent research (e.g., Jackson and
Sunshine 2007; Kochel 2012; Wells et al. 2006) also suggests that residents’ assessments of the effectiveness and quality of policing and the behavior of police are related
to collective efficacy—the “willingness [of residents] to intervene for the common
good” (Sampson et al. 1997: 919)—which may in itself be a precursor to crime control
(e.g., Weisburd et al. 2012).
The diversity of approaches to community-oriented policing reflects the substantial
heterogeneity found across ‘community crime prevention’ programs in general (Gill
2014). Consistent with the broader body of research in this area (Welsh and Hoshi
2006), few studies of community-oriented policing programs explicitly define “the
community.” It is usually assumed to mean the general population of the target area, but
the definition and role of the community in crime prevention varies depending on the
context. Gill (2014) notes that “community” may refer to a physical setting, such as a
neighborhood or other area with administrative or informal boundaries, at which
interventions are targeted; a more abstract concept of place or belonging such as ‘social
fabric,’ shared culture, or support networks; or the actual people within a place who can
be mobilized in crime prevention efforts. Even this last more concrete definition poses
challenges: even at the micro-geographic level, there may be multiple ‘communities’
delineated by cultural or socio-demographic factors, roles and use of space (e.g.,
business owners vs. residents, night vs. daytime economy), behavioral norms, and so
on. Community engagement in crime prevention may be active or passive, direct or
indirect.
The potential for these different definitions to influence the effectiveness of community crime prevention programs has not been studied. Furthermore, the vague
definition of community found in the COP literature may dilute the ability of these
studies to identify effects, particularly since the community is supposed to drive COP
goals. In general, the literature on evidence-based crime prevention, whether in the
context of policing high-crime places, or individual offender treatment, suggests that
highly focused interventions targeted at specific populations and risk factors produce
the most consistent reductions in crime (e.g., Andrews et al. 1990; Lowenkamp et al.
2006; Lum et al. 2011; Weisburd and Eck 2004).
Community-oriented policing, as it is done in practice, clearly faces a number of
definitional and implementation challenges. As Weisburd and Eck (2004) note, it is
surprising that there has been so little systematic study of such a widely-implemented
policy that has been popular enough to have an entire federal agency created to provide
it with resources. This may be explained by the lack of fidelity to the ‘ideal’ model of
COP and the resulting morass of strategies that have come to define the approach; the
lack of clarity around implementation and intended goals of COP; and the lack of
definition of ‘community.’ We believe a systematic review and assessment of the
existing evaluation research on community-oriented policing can help to make sense
of some of this confusion.
We recognize at the outset that very few studies have examined ‘model’ COP
programs in which a set of problem-solving strategies have evolved from a philosophical shift in departmental priorities toward community-oriented goals, and that most
police departments have only adopted “the language of community-oriented policing”
(Trojanowicz et al. 1998: 2). Thus, our goal is to evaluate COP as it has come to be
understood in practice, rather than the theoretical ideal, in order to bring together a
Community-oriented policing: a systematic review
405
‘fractured’ literature1 and understand the circumstances under which COP approaches
might be effective. Ultimately, this may help refine the approach and enhance fidelity to
the philosophical model. Our objective is to synthesize the published and unpublished
literature on the effectiveness of the various strategies and policies collectively termed
‘community-oriented policing’ in an effort to answer the following research questions:
1. To what extent do community-oriented policing strategies reduce crime and
disorder in the target areas?
2. To what extent do community-oriented policing strategies reduce fear of crime and
improve citizen satisfaction with police and perceived legitimacy of the police?
3. Do the effects of community-oriented policing vary according to the particular
strategy/combination of strategies used?
Methods
Inclusion criteria
The following criteria were developed to identify studies for inclusion in the review:
1. The study must evaluate a policing strategy that involved, at minimum, some
type of consultation or collaboration between the police and local citizens for
the purpose of defining, prioritizing, and/or solving problems (e.g., door-todoor visits by officers, providing information to the community, setting up a
local police substation, or developing a crime prevention partnership).
Interventions that also involved problem solving and/or organizational change
in the police department (e.g., decentralization, streamlining of management,
increased responsibility at street level, or community-oriented training and
recruitment policies), which are closer to accepted definitions of COP (Skogan
2006a), are included. As discussed above, we recognize that this criterion
dilutes the accepted definition of the community-oriented policing model, but
our aim is to identify the broader range of approaches that have been implemented within a community collaboration framework. Studies related to COP
but not involving a direct program evaluation (for example, an assessment of
the impact of federal funding for hiring of community police officers on
crime) were not eligible.
2. The study design must be a quantitative analysis including a control or comparison
group and both pre- and post- (or pre- and during-) intervention measures of
outcomes, such as a randomized controlled trial or quasi-experiment.
Comparison areas may receive ‘policing as usual,’ the precise definition of which
varies by study, or another policing intervention that does not include a community
consultation element. Time series studies where outcomes are measured multiple
times before and after the intervention, with adjustment for secular trends, are also
eligible. Simple pre–post studies are excluded due to the risk of historical bias with
such studies.
1
We thank an anonymous peer reviewer for this useful description.
406
C. Gill et al.
3. For official crime outcomes, the unit of analysis of the study must be a police
jurisdiction or area within the jurisdiction, such as a community, neighborhood,
precinct, beat, or ‘micro-geographic’ area such as a street segment (Weisburd et al.
2012). For outcomes involving citizen perceptions, such as fear of crime or
satisfaction with police, individual citizens living within target and comparison
areas are the unit of analysis.
4. The study must have assessed at least one outcome related to crime or disorder
(e.g., arrests, police calls for service, incident reports, victimization reports); citizen
satisfaction with police; fear of crime; and citizen perceptions of physical and
social disorder and police legitimacy (measured through citizen surveys).
5. Studies from any country2 published from 1970 onwards (reflecting the time period
during which interest in COP increased) are eligible.
Search strategy for identification of relevant studies
Our initial searches were conducted in summer 2011 and updated in summer 2012. We
used several strategies to conduct a thorough search for eligible studies: a keyword
search of online databases; a review of the bibliographies of prior studies and reviews
of community-oriented policing (e.g., Sherman 1997b; Sherman and Eck 2002; Skogan
and Frydl 2004; Weisburd and Eck 2004); forward searches for works that have cited
key community-oriented policing studies (e.g., Goldstein 1987; Greene and Mastrofski
1988; Kelling and Moore 1988; Skogan and Hartnett 1997; Trojanowicz et al. 1998); a
review of abstracts from leading policing and criminology journals; and searches of the
websites of research and professional agencies. A full list of online databases,
journals, and agency websites, and the keywords used to search them, are
reported in Appendix A.
Details of study coding categories
We coded a variety of pertinent information from each study. A full coding protocol is
available on request from the lead author. We recorded basic reference information,
details of the target and comparison sites, and the components of the COP intervention,
as well as other police activities in the target and comparison sites. Details of the study
design, unit of analysis, and measurement of specific outcomes, and the sample size
were coded. We detailed any implementation difficulties reported by the study authors
and information about the methodological quality and potential biases, such as attrition.
To calculate effect sizes, we recorded the pre- and post-intervention outcome measure
statistics in the target and comparison areas, including the statistical test(s) used and any
reports of statistical significance. Finally, we noted the authors’ conclusions about
whether the intervention was effective. Given the complexity of COP interventions,
we also captured quotations from the studies for qualitative analysis to ensure we did
not miss any important characteristics.
2
Studies were not excluded on the basis of language, but we lacked resources to conduct our search in
languages other than English.
Community-oriented policing: a systematic review
407
Statistical procedures
We used meta-analysis to combine quantitative findings from studies that reported
sufficient data to calculate an effect size. We used several different methods for
calculating effect sizes, depending on the outcome measure (crime data from geographic units of analysis called for different methods than attitudinal/perceptional data
obtained from citizen surveys) and the presentation of results in the original study.
Wherever possible, we follow standard procedures for effect size calculation and metaanalysis as set out in Lipsey and Wilson (2001) (deviations from these methods are
described below). For ease of interpretation, all effect sizes are presented in the analysis
as odds ratios (OR), where OR > 1 indicates a favorable outcome for the treatment
group relative to the comparison. Where odds ratios were not calculated directly, we
performed conversions using the Comprehensive Meta-Analysis (CMA) software
(Borenstein et al. 2005), which was also used to conduct all analyses.3
Studies of policing interventions at places present particular challenges for effect
size calculation because interventions are often targeted at a single geographic area and
outcomes compared to a single control area. In such studies, there is only one
observation in each group, so many standard measures of effect size (which rely on
means and standard deviations) cannot be used. Studies using time series analysis
present similar difficulties because they are also effectively single subject designs and
the statistical tests from time series models such as ARIMA are not analogous to tests
such as Student’s t. Most of the studies that measured official crime outcomes also
presented pre- and post-intervention counts of incidents, calls for service, etc., for
treatment and comparison areas.
Following prior meta-analyses of place-based interventions (e.g., Bowers et al. 2011;
Braga and Weisburd 2012; Weisburd et al. 2010), we use what Farrington et al. (2007)
term the ‘relative effect size’ to obtain a measure of the relative change in crime counts
that is analogous to an odds ratio (although it should be noted that this is not a true odds
ratio). This method has limitations, notably that we cannot assume that the standard
error of the relative effect size can be calculated in the same way as the standard error of
an odds ratio. 4 We adjust for this limitation by following Farrington et al.’s (2007)
conservative approach: we use the square of the odds ratio standard error formula to
increase the size of the confidence intervals. Thus, we caution that the confidence
intervals and p values presented in the analysis of official crime outcomes later in this
review are only approximations and should not be interpreted as precise. The formulas
for calculating the relative effect size (RES) and its variance (SERES) are presented in
Fig. 1.
Effect sizes for person-based outcomes, such as fear of crime, perceptions of
disorder and legitimacy, and satisfaction with police, were simpler to calculate using
standard methods (as were crime outcomes in studies that tested interventions at
multiple places, such as street segments or multiple beat areas, and reported mean
crime counts and standard deviations that were used to calculate standardized mean
3
Note that meta-analysis is performed on the log odds ratio, but we present findings as odds ratios for
simplicity.
4
If we assume that crime counts follow a Poisson distribution, the standard error would be the same as for the
odds ratio because Poisson is a generalization of the binomial distribution. However, this assumption is not
realistic at places: counts tend to be overdispersed (see Bowers et al. 2011).
408
C. Gill et al.
where a, b, c, and d represent crime counts:
Treatment area
Comparison area
Pre-intervention
a
c
Post-intervention
b
d
Fig. 1 Formulas for relative effect size and adjusted variance
differences and converted to odds ratios). Person-based outcomes were typically binary
measures (yes/no questions), reporting the number or proportion of people surveyed
who felt safe, trusted the police, and so on, which allowed us to calculate a true odds
ratio (OR). Because we are interested in pre–post changes in perceptions (e.g., the
proportion of respondents who reported trusting the police before and after the intervention), we report an adjusted OR calculated by subtracting the pre-test log OR from
the post-test log OR for person-based outcomes.5
Mean effect sizes across studies were calculated in CMA and weighted using the
inverse variance weights to account for the greater precision of effect size estimates
from larger samples. We assumed a random effects model for the meta-analysis,
accounting for the diversity of approaches comprising community-oriented policing
interventions across the eligible studies.
A number of studies in our review included multiple comparisons and/or multiple
outcome measures for the same construct. Meta-analysis assumes that each effect size
in the analysis is based on a statistically independent sample. In a number of studies,
multiple independent comparisons were available for analysis (for example, a program
was implemented in several neighborhoods, each with its own separate comparison
neighborhood, and results for each treatment–comparison pair were reported separately
in the same study). However, other studies contained comparisons that violated the
independence assumption (for example, two different treatments in two neighborhoods
were both compared to the same control neighborhood). In these cases, we picked a
treatment–comparison pair at random for inclusion in the meta-analysis. Some studies
also contained multiple outcome measures for the same construct, inclusion of which
would also violate statistical independence. While there are methods available for
including all measures in the analysis (Hedges et al. 2010), we elected to prioritize
outcomes for the purpose of this review, given the overall complexity of the intervention and inconsistency in measures across studies.
5
The method for calculating the variance of this adjusted OR varies depending on whether the same people
were interviewed in the pre- and post-intervention surveys.