CROSS-SITUATIONAL ASSESSMENT OF HETEROSOCIAL PERFORMANCE: A GENERALIZABILITY STUDY (SOCIAL SKILLS, ROLE-PLAYS, INTERMEDIATE LEVEL, MEASUREMENT)
Abstract
Researchers and clinicians interested in social competence rely upon role-play interactions to obtain samples of social behavior in more or less standardized situations for the assessment of performance capability. Measurement of performance usually includes global ratings of social skill and anxiety, and may include micro-measurement of specific behaviors such as talk, gaze, or gestures. Recently, measurement of intermediate level behaviors such as conversational content or use of hands has been advocated. Thus, the role-play is an essential component of the assessment process for both clinicians and researchers in this area. Many researchers and clinicians use a single role-play to estimate performance capability. There has been no research demonstrating the dependability of measurements based upon a single role-play. The present investigation examined the dependability or generalizability of global, intermediate, and molecular level measurements across five role-play situations in an effort to ascertain how many role-play samples are needed to obtain dependable estimates of social behaviors at each of these three levels of measurement. Forty-five male college students participated in five different role-plays which were videotaped. Global level measures were: skill, anxiety, and attractiveness. Intermediate level measures were: conversation structure, conversation content, verbal support of partner, personal conversational style, and use of head, facial expression, eyes, and hands. Micro-measures were: gaze, smiles, gestures, subject talk time, silence and confederate talk time. Results indicated that in order to obtain a generalizability-coefficient of .70, at least two role-plays are needed for skill (with four raters), and anxiety (with ten raters). Intermediate level measures were quite variable. Dependable estimates for these ratings required two situations for most behaviors, one situation for use of eyes and facial expression, and three situations for use of hands. All molecular level subject behaviors except gestures were generalizable if two situations were observed, gestures required three. Examination of variance components revealed large Person by Situation interactions due to differences in rank ordering of subjects in different situations. Relationships among measures are discussed as well as implications for future research.
Degree
Ph.D.
Subject Area
Psychotherapy
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.