Predictors of School Dropout, Juvenile Delinquency, and Adult Criminality in a Southern U.S. State
Methods: This study examines 919,692 individuals born between 1980 and 1992, with outcomes available through 2012. Administrative state agency data are used, from a Deep South state with high levels of poverty and generally negative health, educational, social, criminal justice, and economic outcomes. This study is an investigative project, so we do not engage in a priori speculation regarding the relative impacts of one risk factor versus another. Multivariate logistical regression models are employed, examining the impact of several risk-related predictors on encountering the Office of Juvenile Justice (OJJ), committing a felony offense leading to OJJ involvement, encountering the Department of Corrections (DOC), only being on probation in the DOC, being incarcerated in the DOC, and dropping out of school.
Results: Though additional factors are statistically significant, findings across models indicate that being a male (Odds Ratios [ORs] ranging from 2.6 to 9.7, depending on the model), failing a grade (ORs from 1.6 to 1.8), and being on free lunch (ORs from 1.5 to 1.7) are consistently the strongest predictors of negative delinquent and criminal outcomes. For models examining school discipline, failing a grade (ORs from 4.1 [suspension] to 4.4 [expulsion]), failing multiple grades (ORs of 1.9 for both models), and being on free lunch (ORs from 1.4 to 1.5) are the strongest, consistent predictors of dropping out.
Conclusions: Findings highlight the relative importance of risk factors leading to negative outcomes and provide suggestions for selectively targeting areas that require the most attention. Notable disparities between those at-risk versus those who are not at-risk, some of the economic costs associated with high risk children, and suggestions for next steps with the research are explored in this research effort. Implications for social work professionals working among at-risk children and juveniles are examined.