“…As Table 1 reflects, several of the actuarial models (S&P, RASP-CC, RASP-Cap) were modestly successful in modeling violence at all severity levels (i.e., potential violence, assaults, and serious assaults) in their construction samples and predicting violence in other samples. Buffington-Vollum, Edens, and Keilen (2008), however, found that the S&P scale was modestly associated with general disciplinary misconduct and minor assaults, but not with serious assaults. This discrepancy is likely a function of the low prevalence (or base rates) of serious assaults.…”