An evaluation of the internal and external validity of the Intrex and Interpersonal Adjective Scales.

摘要

Interpersonal behaviors have long been recognized to play important roles in the expression of psychopathology, adaptive development, effective research, and effective intervention. By implication, it is desirable to provide clinicians and researchers with models and measures that effectively conceive, operationalize, and assess interpersonal variables. The Interpersonal Adjective Scales (IAS) and the Structural Analysis of Social Behavior (SASB) Intrex are designed to assess interpersonal variables. The current research examined important aspects of each measure. Two studies were conducted. The first study focused on aspects internal validity, the second on aspects of external validity. The internal validity study recreated the two-dimensional structures of the measures by factor analyzing IAS and Intrex self-reports (N = 233) and by analyzing naive judges' (undergraduates, N = 136) ratings of IAS and Intrex items for expression of circumplex dimensions. Factor analytic results tended to better confirm SASB model structure than had an earlier Benjamin study (N = 80) and to confirm IAS model structure for normed IAS scores. Factor analyses of raw IAS scores did not yield a two-factor solution. Content analysis of naive judges' ratings of the measures indicated that both models tend to confirm posited structures in some ways and show difficulties confirming them in others. Both IAS and Intrex variables achieved the correct circumplex order. However, spacing among variables and in relation to circle center tended to be uneven. Within each cluster, IAS items tended to show substantial variance. As expected, IAS variables which relied heavily on the use of unusual words deviated more substantially from expected locations. Intrex ratings of behavior in reaction to others produced relatively good variable spacing. Ratings of behavior toward others and toward self tended to associate autonomy giving with affiliation and control with hostility. The second study focused on provision of useful treatment information. Two hundred and twenty-seven students and patients were assessed on the IAS, Intrex, MCMI--III, PAI, and NEO-PI-R. Canonical analysis indicated that both measures associated broadly with representations of psychopathology. Both measures provided information that clinicians and researchers could readily apply to their work. As expected, the Intrex's structure typically provided clinically related information with greater specificity.