The power and utility of latent growth modeling applied to longitudinal data for assessing the developmental synchrony of offending and victimization

In this dissertation I describe and empirically illustrate how a powerful data-analytic technique--Latent Growth Modeling (LGM)--can be used by empirical researchers who are interested in detection and analysis of change in longitudinal data. In doing so, I provide insight into the application of LGM to real empirical data--the National Youth Survey--to address the issue of co-development of offending and victimization at the individual level. Throughout the dissertation, I examine 1) whether there is co-development of offending and victimization over time, and 2) whether temporal co-development of offending and victimization varies across different sub-groups--gender groups and statistically identified classes of individuals--boys and girls--with distinct offending-victimization trajectories. The results of this dissertation clearly show that offending and victimization do not evolve independently; rather, their developmental trajectories are related to one another across gender groups and across statistically identified classes of boys and girls with qualitatively different offending-victimization paths. The results of my analyses indicate that for all latent classes of boys and girls, with one exception, if offending escalates with time, the victimization escalates as well; whereas, if deviant behavior declines with time, victimization declines too, and this relationship, in most cases, is stronger for boys. There was one group of girls for which a decline in the level of offending was statistically associated with an increase in the level of victimization