Reliability Generalization of Working Alliance Inventory Scale Scores

Abstract

Reliability generalization (RG) was used to study five versions of the Working Alliance Inventory (WAI), including scores from 12 different scales. Sixty-seven internal consistency estimates, six interrater reliability estimates, and four study characteristics were analyzed. In general, reliability estimates of WAI scale scores appear to be robust. Mean reliability estimates ranged, in this sample of studies, from .79 to .97, with a modal estimate of .92. Variability in reliability estimates was, based on simple bivariate correlations, associated with client and therapist sample size for WAI total scores (observer version). Implications for measuring alliance using the WAI and conducting future RG studies on psychotherapy process measures are discussed.