Skip to main content

Applications of growth-mixture modelling in violence research

Recent theoretical and empirical work in criminology supports the classification of individuals into groups, where the groups differ in trajectories of offending over time. Over the last three years we have employed growth-mixture modeling (semiparametric group-based modeling; Nagin, 1999) to identify and examine naturally occurring offense trajectories from childhood through adolescence and young adulthood (e.g., Chung, et al. 2002). In this presentation, I provide several illustrations of applications of this procedure in prevention research including: (1) static predictors of violence trajectories; (2) gang membership as a time-varying predictor of violence trajectories; and (3) trajectory-to-trajectory analysis: trajectories as predictors and outcomes. The sample is from the Seattle Social Development Project (J. David Hawkins, PI), a longitudinal panel study of 808 youth interviewed annually from 1985 (age 10) to 1991 (age 16), and again in 1993 (age 18), 1996 (age 21), and 1999 (age 24). The sample, which was selected to over-represent students from schools serving high crime neighborhoods, is gender-balanced, ethnically diverse, with high retention rates (95% of the still-living sample were interviewed at age 24). Analysis methods include semi-parametric group-based modeling for estimating developmental trajectories, examining the impact of a time-varying predictor, and examining transition probabilities between sets of trajectories, and multinomial logistic regression for examining the impact of static predictors on violence trajectory membership.

Chung, I.-J., Hill, K. G., Hawkins, J. D., Gilchrist, L. D., & Nagin, D. S. (2002). Childhood predictors of offense trajectories. Journal of Research in Crime and Delinquency, 39(1), 60-90.

Nagin DS. (1999) Analyzing developmental trajectories: A semiparametric, group-based approach. Psychological Methods, 4:139-57.