Understanding the Latent Structure of Dynamic Risk: Seeking Empirical Constraints on Theory Development Using the...

The present study is part of a larger project aiming to more closely integrate theory with empirical research into dynamic risk. It seeks to generate empirical findings with the dynamic risk factors contained in the Violence Risk Scale-Sexual Offense version (VRS-SO) that might constrain and guide the further development of Thornton's theoretical model of dynamic risk. Two key issues for theory development are (a) whether the structure of pretreatment dynamic risk factors is the same as the structure of the change in the dynamic risk factors that occurs during treatment, and (b) whether theoretical analysis should focus on individual dynamic items or on the broader factors that run through them. Factor analyses and item-level prediction analyses were conducted on VRS-SO pretreatment, posttreatment, and change ratings obtained from a large combined sample of men (Ns = 1,289-1,431) convicted and treated for sexual offenses. Results indicated that the latent structure of pretreatment dynamic risk was best described by a three-factor model while the latent structure of change items was two dimensional. Prediction analyses examined the degree to which items were predictive beyond prediction obtained from the broader factor that they loaded on. Results showed that for some items, their prediction appeared to be largely carried by the three broad factors. In contrast, other items seem to operate as funnels through which the broader factors' predictiveness flowed. Implications for theory development implied by these results are identified.