Meta-Analysis Snapshot
|
Literature Coverage Dates |
Number of Studies |
Number of Study Participants |
Meta Analysis 1 |
1993-2007 |
11 |
2166 |
Meta Analysis 1Joliffe and Farrington (2007) conducted a meta-analysis to assess the effectiveness of interventions with violent adult males on general reoffending and violent reoffending. Studies on domestic violence, sexual offending, and offending by individuals with a personality or mental disorder were excluded.
To be included in the meta-analysis, studies had to have met six criteria. First, the study had to investigate the effects of an intervention or treatment. Second, the intervention had to include adult males who committed violent offenses. For the meta-analysis, a violent offender was defined as a person identified as violent either by official contacts with the criminal justice system or through self-reports. Third, the study had to measure at least one quantitative offending outcome variable that allowed the direction of the effect to be determined (i.e., whether the outcome favored the treatment or comparison group). Fourth, the study design had to involve a comparison that contrasted one or more interventions with one or more comparable control conditions. Random assignment designs were eligible, but one-group pretest–posttest studies were not. Nonequivalent comparison group designs could be eligible; however, such comparisons had to include 1) matching of the treatment and control/comparison groups prior to treatment on a recognized risk variable for offending (e.g., prior offending history); 2) a pre-intervention measure (pretest) of at least one offending outcome variable on which the treatment and control/comparison groups could be compared; or 3) some other demonstration of comparability of the groups. Fifth, the study had to include a sample size of at least 25 individuals per condition, or 50 total. And sixth, the study had to be published between 1975 and 2006.
The search strategy included searches of electronic databases (such as Criminal Justice Abstracts and PsychLIT), research registries, references from relevant articles, studies that cited relevant articles, and hand searches of relevant journals (such as Criminal Justice and Behavior and International Journal of Offender Therapy and Comparative Criminology). The searches resulted in identification of 1,955 studies that were potentially relevant. Of those studies, 11 were included in the final review, and their quality was assessed using the Maryland Scientific Methods Scale. One study was rated as Level 5 (randomized controlled trial), eight studies were rated Level 4 (quasi-experimental), and two studies were rated Level 3 (two comparable groups). The studies were published from 1993 through 2007 in the United Kingdom, Canada, United States, and New Zealand. The 11 studies had a total sample size that ranged from 61 to 571, and the mean age of 29 years (ages ranged from 23 to 36).
The study used modified least squares regression to measure the independent influence of various features on effect sizes of reoffending and violent reoffending. This analysis was also used to measure the impact of the intent-to-treat group versus the treatment completers. Random effects models were used to analyze the results, and effect sizes were calculated as Cohen’s d.