Abstract
This study investigated two issues in youth risk assessment that may be important to juvenile justice agencies: (1) whether there are age-related differences that might impair the predictive accuracy of risk assessment across adolescence and (2) whether dynamic risk factors provide a unique contribution to risk assessment. The study tracked new petitions over an average 14.5-month follow-up for a large sample (n = 674) of adjudicated young offenders who received the Structured Assessment of Violence Risk for Youth (SAVRY). Findings indicated that age did not moderate the association between the SAVRY and reoffending between youth age 12 and under, age 13 to 15, and age 16 to 18. Dynamic risk factors had incremental predictive validity over static factors for each type of recidivism (e.g., violent, nonviolent) except probation violations. Implications to juvenile justice agencies include the critical importance of including dynamic risk factors in risk assessment tools of youth and the generalizability of these tools across age.
Notes
This research was funded by the John D. and Catherine T. MacArthur Foundation.
1. AUC values range from 0 (perfect negative prediction) to 1.0 (perfect positive prediction), with .50 indicating chance prediction. Values below .70 may be considered to be small; values in the range of .70 to .75 have been viewed as moderate; and values above .75 typically are regarded as large (e.g., CitationDouglas, Yeomans, & Boer, 2005). However, such descriptors are tentative, as no definitive classification scheme exists.
2. Given space constraints in the table, results of the between-group AUC difference tests were not included but are available from the first author.
3. Each of the regression analyses were replicated by using just the older two age groups while eliminating the age 12 and younger group as a check to determine whether the small sample size of this group affected the results. The pattern of results was the same.