Testing measurement equivalence in a patient satisfaction instrument.

The purpose of this methodological study was to examine five sources of measurement equivalence (configural invariance, metric invariance, scalar invariance, invariant uniquenesses, and invariant factor variances) in perceptions of patient satisfaction for two groups of patients, whose satisfaction was measured at different times, for men compared with women, and for minority populations compared with Whites. The sample consisted of a total sample of 1,897 patients of whom 1,070 were females, and 1,228 were White. Using confirmatory factor analysis and structural equation modeling, we found that the 10-item Likert-type scale performed well across time but required additional attention if it is to be used to compare satisfaction between genders and racial groups.