| ||Literature Coverage Dates||Number of Studies||Number of Study Participants|
|Meta-Analysis 1||1967 - 1983||8||1091|
|Meta-Analysis 2||1967 - 1992||9||946|
Aos and colleagues (2001) set out to evaluate the costs and benefits of certain juvenile and adult criminal justice policies, violence prevention programs, and other efforts to decrease various at-risk behaviors of youth. The authors used meta-analytic techniques to estimate the degree to which a program or policy can be expected to influence the outcome of interest (i.e. criminality).
The review focused on evaluations that measured a program’s effect on criminal behavior. Program evaluations were gathered from a wide variety of sources, including studies that were published in peer-reviewed journals as well as other sources (such as government or private agency reports). Once an evaluation that met the criteria of inclusion was located, four types of information on the program’s effectiveness in reducing crime were recorded: (1) Did the program affect the percent of the population that offended or reoffend? (2) Of those that offended or reoffended, did the program change the average number of offenses? (3) Did the program affect the types of offenses of those that offended or reoffended? (4) Of those that offended or reoffended, did the program change the timing of offenses? Of the four effects, almost all of the evaluations provided information on the first effect. Far fewer evaluations reported information on the second effect, and even fewer reported information on the third effect. Almost none of the evaluations reported information on the fourth effect.
The mean difference effect size was calculated for each program. Adjustments were made to the effect sizes to account for small sample sizes, evaluations of “non-real world” programs, and the quality of the research design (the quality of each study was rated using the University of Maryland’s five-point scale; only studies that received a rating of ‘3’ or higher on the scale were included in the analysis). Once effect sizes were calculated for each program effect, the individual measures were added together to produce a weighted average effect size for a program or practice area. The inverse variance weight was calculated for each program effect and those weights were used to compute the average. A fixed effects model was used for the analysis.
The search resulted in the inclusion of eight studies of juvenile awareness programs. The eight studies included approximately 1,091 juvenile participants with a mean age of 15. The group of eight studies included published and unpublished evaluations that spanned from 1967 to 1983. No information was provided on the gender and racial/ethnic breakdown of the studies’ samples, or on the location of the programs.Meta-Analysis 2
Petrosino, Petrosino and Buehler (2004) meta-analyzed the results from nine randomized trials that compared youth who participated in juvenile awareness programs to youth who were assigned to a no-treatment control condition. A comprehensive literature search was conducted to identify published and unpublished studies through November 2003.
To be included, studies had to use random or “seemingly” (i.e. quasi) random procedures (i.e. alternating every other case to one group or odd/even assignment) to assign participants to either the treatment or control groups. Studies were included if they only involved juveniles (i.e. children 17 years of age or younger, or overlapping samples of juveniles and young adults such as ages 13-21). Studies also had to examine interventions that featured as their main component a visit by program participants to a prison facility. Finally, studies had to include at least one outcome of subsequent offending behavior that could be measured by such indices as arrest, conviction, contact with police, or self-reported offenses.
The nine studies included 946 juveniles or young adults, and were conducted in eight different states across the country, including Michigan, Illinois, Virginia, Texas, New Jersey, California, Kansas, and Mississippi. The studies spanned the years 1967 to 1992. Five of the studies were unpublished and disseminated in government documents or dissertations, while the other four were found in academic journals or book publications. The average age of juvenile participants in each study ranged from 15 to 17. Only one study (conducted in New Jersey) included girls in the sample. The racial composition of study participants was diverse, ranging from 36 percent to 84 percent white (there was no information provided on the racial/ethnic breakdown of non-white youths included in the studies).
The crime outcomes for official measures at the first follow-up period were reported. Each analysis focused on proportion data (i.e. the proportion of each group reoffending). Because the data from the studies mostly reported dichotomous outcome measures of crime, odds ratios and confidence intervals were calculated for each study. The authors assumed both random effects model and fixed effects models for treatment effects across the studies (because the results from each model were very similar, only the mean odds ratio from the random effects model is reported here). Although the initial analysis included nine studies, one study (Finckenauer 1982) was excluded because of randomization problems.