Congruence research in behavioral medicine: methodological review and demonstration of alternative methodology

Abstract

Researchers in behavioral medicine are using methods to study the effect of congruence between two predictors (e.g., patient and provider preferences for patient-decision making) on outcomes (e.g., patient satisfaction and adherence) that may compromise the validity of their results and conclusions. The current paper reviews the methods used in behavioral medicine for the study of congruence effects and the problems associated with those methods-namely, that difference-score-based and artificial-group-based methods increase the risk of both Types I and II error and constrain the theoretical hypotheses that researchers are able to assess. The current paper explains and demonstrates a possible alternative method, polynomial regression, that may be used in some instances of congruence research and that avoids many of the problems of currently used methods; the current paper provides the first analysis of behavioral-medicine data using this method. Methodological advancement in health-related congruence research can help advance theory and optimize intervention-design.