Meta-Analysis Snapshot
|
Literature Coverage Dates |
Number of Studies |
Number of Study Participants |
| Meta Analysis 1 |
1971-2010 |
16 |
178 |
Meta Analysis 1Bowers and colleagues (2011) conducted a meta-analysis to 1) assess the effect of geographically focused policing initiatives on crime and disorder, and 2) evaluate displacement and diffusion (i.e., whether crime generally increased or decreased in catchment areas more than control areas) following the implementation of a geographically focused policing intervention. However, the focus of this CrimeSolutions review was solely on the overall effect of geographically focused policing initiatives on crime.
To be included in the meta-analysis, the study must have 1) evaluated a focused policing intervention that fell into the categories described above in the Practice Description, 2) used some quantitative measure of crime and/or disorder, 3) reported original research findings (i.e., systematic reviews or meta-analyses were not included), and 4) included an intervention that was geographically focused on an area smaller than a city or a region. The meta-analysis also included both published and unpublished studies that met the criteria, and which could have been conducted at any point in time or at any location.
The search strategy to retrieve relevant studies included a keyword search of electronic databases, a review of bibliographies of existing displacement reviews and those on the effectiveness of focused policing initiatives, forward searches for works that cited key publications, a review of research reports of professional research and policing organizations, and a hand search of pertinent journals. The original search was carried out between December 2009 and January 2010 and yielded more than 2,731 studies.
Following coding and screening, two reviewers determined that 44 studies were eligible for review. Of the 44 studies, five were randomized controlled trials, and 39 were quasi-experimental designs. Most of the 44 studies used simple pre- and posttest research designs (57 percent), 43 percent used more complex pre- and posttest assessments with at least one control area, and 14 percent used a separate catchment for the control area. Thirty of the interventions had taken place in the United States, 10 in the United Kingdom, three in Australia, and one in Sweden. Of the others, 23 took place in purely residential environments, nine in multiple types of environments, four in various environments, four in mixed areas containing both residential and retail, two in strictly retail locations, and one in an educational and recreational environment.
The most common type of outcome data used in the included studies was recorded crime (35 studies), seven studies used calls for service data, four used arrest data, and one used data generated through observation methods. The studies also varied in terms of the size of the physical area that received the intervention. Of the 44 studies, 24 covered a large area, nine covered a medium area, and 11 covered a small area. In terms of the type of intervention, 12 used problem-oriented policing programs, seven used directed patrol studies, four used hot spots policing, 10 looked at police crackdowns, two looked at intelligence-led policing, five used community-oriented policing, and two adopted a broken windows approach. Of the remaining studies, one used a civil abatement intervention, and one used a police-led initiative to alter the physical environment.
The authors used odds ratios to estimate the effect of geographically focused policing initiatives on the treatment areas. In addition to calculating individual estimates of relative effect size for each treatment area in each study, they also calculated mean effect sizes across studies.
Of the 44 studies, only 16 provided individual effect sizes for both treatment and control areas. Therefore, only those 16 studies were included in the meta-analysis.
