Variations in Juvenile Offending in the Deep South: Demographic, Behavioral, Geographic and School-Related Predictors

Background and purpose: The relationship between early childhood risk factors and negative life outcomes has been examined extensively in the extant literature; however, the relative impacts of various risk factors on negative outcomes (in particular, delinquent behavior and juvenile justice involvement) are still not fully understood, and this understanding could have a critical impact on social policy and the most effective selection of targeted interventions in social, developmental, educational, and associated domains.  This study examines these relative impacts through examination of demographic, behavioral, geographic, and school-related factors as they impact juvenile justice contact of varying magnitudes across a large, broad-based sample of youth in the Deep South.

Methods: In this exploratory study, the sample includes public school students examined from 1996 through 2012 (N=615,515). All eligible participants were included in this study.  Data were obtained through state administrative databases in a Deep South state with high levels of poverty, crime, and underachievement among at-risk populations.  Dependent variables represent Juvenile Justice (JJ) contact.  Four models are examined, each exploring a different aspect of JJ contact: 1) all JJ contact; 2) felony JJ offense contact; 3) misdemeanor JJ offense contact; and 4) status JJ offense contact.  Predictor variables include rural and urban school setting, African American race, gender, school failure, free lunch subsidies (socio-economic proxy), age at first school contact, birth year cohort control, average yearly school absences, prior JJ involvement, and school expulsion.  Hierarchical linear models employing logistic regression examine outcomes of interest (outcomes are coded as 0 = no JJ contact of interest, and 1 = JJ contact of interest), with first school attended included as the nesting variable/random effect.  In addition to conventional statistical results, predictive margins are also examined for the purpose of additionally exploring the practical significance of findings.

Results: Noteworthy findings are that school expulsion (odds ratios [ORs] ranging from to 2.34 to 3.52, depending on the model), male gender (ORs ranging from 1.11 to 7.63), prior LAOJJ contact (ORs ranging from 1.16 to 1.98), and grade failure (ORs ranging from 1.96 to 2.08) are major, frequently statistically significant predictors, though their relative impact varies based on the severity of offense.  Further, being African American (ORs ranging from 1.20 to 1.64) loses much of its practical significance in all models once other factors are taken into account.

Conclusions and Implications: This study can provide social work professionals with knowledge to effectively direct attention and assistance toward interventions where they are likely to do the most good, an important concern in a reality of limited resources juxtaposed with difficult social challenges.  Specifically, these findings speak to considerations regarding the relative importance of gender, race, and rural/urban environment within and across models, suggesting that programs focusing on pro-social behaviors, peer and family support, and school attachment/achievement may be particularly important for specific, identified subgroups, depending on the JJ context.