A methodological examination of trajectories of antisocial behavior
MetadataShow full item record
This is a methodological examination of the results of all articles that have examined the developmental nature of antisocial behavior by incorporating semi-parametric group based modeling or growth mixture modeling analyses into longitudinal studies. In this exploratory study, the author sought to answer the following research question: Do differences in sample, design, or measurement characteristics predict the differences in empirical results found among the articles? Each article that used semi-parametric group based modeling or growth mixture modeling was coded for sample characteristics, including gender of the sample and sample size; design characteristics, including number of assessments and interval in years between assessments; measurement characteristics, including number of scale items on the dependent variable, number of response categories of the dependent variable, and breadth of the dependent variable; and outcome characteristics, including the number of trajectories found and the presence or absence of abstainer, low-level, desister, and chronic trajectories. These independent variables were chosen based on differences observed while reading these articles. The dependent or outcome variables were chosen based on the number and types of trajectories predicted by Moffitt (1997). Sample, measurement, and design characteristics were used to predict outcomes using logistic and multinomial logistic regression. Sample size, number of assessments, assessment intervals, number of scale items, number of response categories, and breadth of the scale used to model antisocial behavior predicted the number of trajectories found or the particular types of trajectories found. Therefore, differences in sample, design, and measurement characteristics are at least partially responsible for differences in empirical results using semi-parametric group based modeling. Follow-up analyses investigated whether age is related to the number of trajectories found and to the presence or absence of abstainer, low-level, desister, and chronic trajectories. Results indicated that the presence (or absence) of abstainer, desister, and chronic trajectories was also related to whether not the participants were assessed during adulthood.