Continuous versus Categorical Models of Delinquency Risk

Two groups of participants, one a nationally representative sample with roughly equal numbers of male and female participants (N?=?8,984) and the other a sample of mostly male adjudicated delinquents (N?=?1,354), were used to test whether risk factors for delinquency are organized continuously or categorically. A continuous (variable-centered) model was created using factor scores from a one-factor confirmatory factor analysis and a categorical (person-centered) model was constructed using posterior probabilities from a two-class finite mixture modeling analysis. In both samples the continuous model correlated significantly better with subsequent offending than did the categorical model, a finding that was replicated in males from both samples and in females from the nationally representative sample. The current findings suggest that risk factors are better construed as points along a continuum rather than as properties of distinct groups or types. These results further suggest that the etiology of offending, in the form of risk factors, is general/additive rather than specific/selective. The implications of these results for theory development, clinical practice, and future research are discussed.