Latent class analysis of early onset of behavior problems: relevance for predicting offending across the life-course

ABSTRACT Research has demonstrated the importance of early-life behavioral issues for predicting risk for offending throughout the life-course. However, there has yet to be any study that has examined how early behavior problems may cluster together and how these clusters predict offending across the life-course. Latent class analysis was used to identify behavior classes comprising the onset of the childhood behavior problems: cheating, disrupting class, being drunk/stoned, stealing, and fighting. Analysis of variance was used to identify between-class differences in risk/protective factors. Age–crime curves were identified for groups to measure offending throughout the life-course. Results indicated that a three-class model best fits the data (Low EOBP, FDC, DCSF). Groups differed in risk/protective factor profiles. Age–crime curves indicated that participants assigned to the DCSF group demonstrated more serious offending in adolescence and that the FDC and DCSF groups demonstrated more serious offending in early adulthood. Identification of individuals reporting disturbing class, fighting, and stealing in childhood may be at elevated risk for more serious offending in adolescence. This may place them at risk for encountering snares in the life-course.