Creating a Categorical Understanding of Adolescent Delinquency

» Posted by on Jan 13, 2014 in Alpha Blog | 0 comments

 by Steven Bengis, David S. Prescott, and Joan Tabachnick



Can understanding categories of delinquency lead to more effective intervention, treatment, and prevention programs?


The Research

In a just-released 2013 paper, Kristina Childs and Christopher Sullivan used data gathered on 1,124 adolescents in a short-term longitudinal study from the Project on Human Development in Chicago Neighborhoods (PHDCN). The researchers compared two models used to explain adolescent behaviors. The first was a dimensional model. This would involve a single underlying trait, a “general proneness to deviance” that would generate a variety of deviant behaviors from less to more severe.

The other was a categorical model, which would involve a difference in the pattern and level of problem behaviors that creates sub-groups of teens engaging in different kinds of risky behaviors. The authors also considered a hybrid of both approaches that might better explain differences in behaviors.

Childs and Sullivan analyzed several self-report measures of risky adolescent behavior, including different areas of substance use, delinquency, and sexual behaviors. They concluded that using a categorical model involving four classes (e.g., drug use and delinquency group; experimenter group; abstainers; high-risk & delinquency) most accurately explained behaviors. Their research showed that many risk factors for engaging in risk behaviors differ across these sub-groups of adolescents (e.g., risk factors for the “drug use and delinquency” group include low social support, antisocial peers, and having a family member with criminal and legal problems. The authors’ research also showed distinct patterns of change when using the categorical model, e.g., they found the most prevalent change was the move from low risk/abstinence to increased but nonetheless non-serious experimentation (15% of the sample). However, nearly 10% of the sample escalated from low risk/abstinence or experimentation to higher risk behaviors of drug use and delinquency and 7% persisted in the high risk behaviors.

Implications for Professionals

The above research demonstrates the critical importance of differentiating sub-groups in the adolescent population when considering risk, treatment implications, prevention, and early and intensive intervention approaches. Professionals should take great care when recommending intervention and treatment approaches to ensure that the recommendations are supported by research, and in this case follow what works best with various sub-types of youth. Incorporating treatment elements that fail to promote successful outcomes with a given sub-group can either waste valuable fiscal and human resources or possibly lead to increased risk.

Implications for the Field

This study did not target the sub-group of youth who engage in sexually abusive behaviors alone or those who engage in both sexually abusive and non-sexually abusive delinquent actions. But we know from other research, that many youth who abuse sexually profile very similarly to the general delinquent population. While approximately 7-12% of sexually abusive youth are known to re-offend sexually (depending on their treatment status), up to 50% may re-offend in other criminal ways (for more information, see the NEARI newsletters of May 2008 and November 2012 at

Thus, our intervention programs must incorporate interventions that the criminology literature suggests are effective for general delinquency and we need to consider the important differences in the sexually abusing sub-group (See NEARI’s January 2013 summary of James Worling’s review of abuse-related sexual interests in teens). By more accurately identifying sub-groups of youth whose behavioral patterns place them at risk to sexually abuse, we may be able to develop more targeted and individualized prevention and early intervention programs. We may also continue to advocate for use of limited resources to target more effectively those likeliest to re-offend. Finally, we can develop different types of interventions that more accurately accommodate individual differences.


Data collected as part of the Projects on Human Development in Chicago Neighborhoods (PHDCN) were used to examine (1) the underlying structure of adolescent problem behavior, (2) continuity and change in patterns of problem behavior across mid to late adolescence and, (3) the risk and protective factors related to observed patterns of behavior. The results suggested that a 4-class categorical model best represents the pattern of responses to behavioral items used to measure delinquency, substance use and risky sexual practices.


  • Childs, K.K., & Sullivan, C.J. (2013). Investigating the underlying structure and stability of problem behaviors across adolescence. Criminal Justice and Behavior, 40, 57-79.