“…Predictive reliability and accuracy remain poor, and significant numbers of false positive predictions continue to be made, given the relatively low prevalence of serious violent conduct in psychiatric patients. Advances in predictive methodology can be anticipated with improved characterization of the etiologies and syndromes of violent behavior (Huber, Roth, Appelbaum, & Ore, 1982;Prins, 1980;Craig, 1982;Elliott, 1982;Mungas, 1983) and their precipitants (Masuda, Cutler, Hein, & Holmes, 1978;Levinson & Ramsay, 1979;Langevin, Paitich, Orchard, Handy, & Grusson, 1982;Monahan & Klassen, 1982;Felson & Steadman, 1983). Use of probability statements rather than dichotomous predictions, differential base rates, and validated decision aids would also be helpful.…”