| ||Literature Coverage Dates||Number of Studies||Number of Study Participants|
|Meta-Analysis 1||1977 - 2014||51||132366|
Spruit and colleagues (2016) conducted a meta-analysis to examine whether there was a relationship between sports participation and juvenile delinquency. Juvenile delinquency was operationalized as criminal behavior (i.e., violation of the law) by a minor, outside the sports context. This meta-analysis excluded other types of deviant behaviors such as behavioral problems, status offenses, and antisocial behavior (i.e., substance use or aggression).
To be included, studies had to report a relationship between sports participation and juvenile delinquency that allowed for calculation of an effect size. The studies had to include a sample where the mean age was between 12 and 18. The studies had to include both athlete and nonathlete samples and both delinquent and nondelinquent samples or samples of the general population of adolescents. Finally, the variables of interest had to be measured on the individual level.
A comprehensive search of nine electronic databases (including ScienceDirect, Web of Knowledge, Ovid, Picarta, Wiley, Google Scholar, ProQuest, EBSCOhost, and Narcis) for studies published before October 2015 was conducted. ProQuest Dissertations and Theses databases were also searched for unpublished studies. Reference sections of review studies on leisure participation and behavioral problems, and publication lists of some experts on sports and antisocial behavior, were also checked.
The initial search resulted in 414 articles, which also contained review and qualitative studies. Ultimately, a total of 51 studies met the inclusion criteria, which included 48 independent samples, 431 effect sizes, and 132,366 participants.
Studies were coded as either cross-sectional (i.e., measured the relationship between sports participation and juvenile delinquency at one point in time) or longitudinal (i.e., took the developmental aspect of the relationship between sports participation and juvenile delinquency into account). Of the eligible studies, 42 were cross-sectional, 8 were longitudinal, and 1 was mixed.
Only 29 studies reported information on race/ethnicity; of those studies, the percentage of minority youths ranged from 0 to 100 percent. Almost all studies reported on gender. Four studies included only boys, two studies included only girls, and the rest included both genders. No other demographic information on the sample was provided
The effect sizes were calculated and transformed into correlation coefficient r (where a positive correlation would indicate that athletes are more delinquent than nonathletes, and a negative correlation would indicate that athletes are less delinquent than nonathletes). It should be noted that the analysis included multiple effect sizes per study. To deal with the interdependency of effect sizes, the authors employed a three-level random effects model to account for three levels of variance, including the sampling variance for each effect size (level 1), the variance between effect sizes within a study (level 2), and the variance between the studies (level 3).