“…A number of studies have demonstrated a correspondence between behavior seen on role-play tests and that seen in naturalistic analogs (Blumer & McNamara, 1982; Merluzzi & Biever, 1987; Wessberg, Mariotto, Conger, Conger, & Farrell, 1979), whereas others have found only a modest relationship between behavior on standard role-play procedures and in matched criterion situations (Bellack, Hersen, & Lamparski, 1979; Higgins, Alonso, & Pendleton, 1979; Kern, Miller, & Eggers, 1983). The few studies conducted with psychiatric patients have also yielded inconsistent results (Bellack, Hersen, & Turner, 1978; Helzel & Rice, 1985; Wessberg et al, 1981). For the most part, role-play tests appear to account for a moderate proportion of variance in ratings derived from naturalistic test criteria, with molar measures (e.g., overall social skill) accounting for more variance than molecular-response components (e.g., eye contact and speech disruptions).…”