Sorensen and Johnson (1996) recently used cluster analysis to develop a new
classification system for incarcerated juvenile delinquents based upon their responses to the Jesness Inventory and the:MMPI. The new classification system appeared to be a refinement of a previous system derived by factor analysis (Quay, 1987a). Sorensen and Johnson then used discriminant analysis to examine the generalizability of the system to a second sample, with good results. They concluded that their new system provides for more accurate classification of delinquent subtypes than was previously possible and recommended that other researchers first identify the five new subtypes, anticipating refinements in predictability that would perhaps account for previously nonsignificant findings. The purpose of this study was to investigate the usefulness of the Sorensen.and Johnson classification system in three ways. First, in order to test the construct validity of the system, it was applied to a new, non-incarcerated sample of juvenile delinquents treated for substance dependence in a community-based treatment center. Second, In order for the system to remain useful to current and future researchers and clinicians, it had to be demonstrated that scores from the newer, adolescent-focused version of the :MMPI, the MMPI-A, could be used to generate the same subtypes. Finally, an
investigation of the system's clinical utility was made by conducting exploratory testing of between-group SUbtype differences on treatment outcome measures (successful completion oftreatment, length of abstinence from substances, and criminal recidivism). The results indicated that the system accurately classified the non-incarcerated delinquents. Some of the Mlv.lPI-A generated subtypes, however, appeared to differ in important ways from their :MMPI counterparts. Four Chi-Square and two ANOVA analyses produced uniformly negative findings with regard to between-group comparisons on the outcome measures. Due to the methodological limitations of the study, however, these results are equivocal, and the predictive utility of the system still remains in need of investigation. Strengths and shortcomings of the new classification system and limitations of the present study are discussed, and the author offers suggestions for future research.
Files are restricted to Pacific University. Sign in to view.