IS CONSTRUCT EQUIVALENCE SUFFICIENT TO ESTABLISH CROSS-CULTURAL GENERALIZABILITY: AN ARGUMENT FOR MODEL EQUIVALENCE
Priester, P.E., University of Wisconsin-Milwaukee, USA
Construct equivalence is the current standard used to support the existence of cross-cultural generalizability for a psychological construct. The current standard for demonstrating evidence of construct equivalence involves accurate translation of an instrument (using blind-back and translation-back techniques), norm development, cross-cultural validation (Bracken & Barona, 1991); and performing analyses to examine whether a similar latent factor structure exists (McCrae & Costa, 1997; Vittorio- Caprara, Barbaranelli, Bermudez, Maslach, and Ruch, 2000). This author suggests that construct equivalence is a necessary but not sufficient condition for demonstrating the existence of cross-cultural generalizability. A valid translation of an instrument, population- specific norms, and the presence of a comparable latent factor structure of a construct alone do not meet an acceptable standard to warrant assertions of the existence of cross-cultural generalizability. Psychological research and theory evolution are concerned with predicting the relationships between constructs, not merely their existence. This paper suggests that international psychologists wanting to assert the existence of cross-cultural generalizability should meet the standards for "model equivalence." A proposed strategy for testing model equivalence of a psychological construct will be offered. This approach incorporates meta- analytic research strategies to determine whether the relationships between variables from two data sets are equivalent. By first meeting the standard of construct equivalence, and then satisfying the standard of model equivalence, researchers can authoritatively claim whether cross-cultural generalizability is feasible. A meta-analysis that explored generalizability questions related to psychotherapy research will be offered as an example of the application of this research synthesis strategy (Priester, 2002). This approach will be differentiated from Validity Generalizability theory as well.