Such a calculation involves a formula (Gibertini, Branden- berg, & Retzlaff, 1986; see Millon, 1994, pp. 41–43) in which prevalence rates must be inserted (derived from knowledge regarding a specific client population) along with sensitivity and specificity data (available in the MCMI-III manual). Such a calculation provides practitioners with an estimate of the extent to which the instrument performs beyond merely BR levels. For example, if the prevalence or BR of antisocial personalities is .25 but the positive predictive power of the MCMI is .76, the difference (.76 − .25) of .56 indicates that the incremental validity of the instrument is .56 above merely BR (prevalence) or chance predictions. This emphasis on levels of certainty, with its implications for actual clinical decision making, is one of the strong features of the MCMI.