Increasing measurement precision in psychotherapy research

Abstract

This chapter addresses the critical importance of measurement in psychotherapy research and presents an overview of item response theory (IRT) and bifactor models as a more comprehensive measurement model approach for psychotherapy outcome research. The focus of the chapter is on parametric IRT models; however, it is well recognized that nonparametric IRT models offer flexibility in terms of data analyses, based on less restrictive assumptions that typically lead to the inclusion of more items in the scale. Statistical methods are impervious to the level of measurement and response scale imprecision. The test characteristic curve (TCC) illustrates the issues faced in relying on raw scores. Rasch and IRT models provide a method of associating raw scores with interval-level latent trait units. The goal of the chapter is to raise awareness of the many issues associated with the measures used in supporting the decisions made about how psychotherapy works, how clients change, and how effective psychotherapy is in terms of its outcomes.