Bringing the psychotherapist back: basic concepts for reading articles examining therapist effects using multilevel modeling

Abstract

Psychotherapy research examining the influence of psychotherapists on clients' clinical outcomes can provide valuable insights for enhancing psychotherapists' effectiveness. Psychotherapy data often have a hierarchical structure--multiple clients treated by the same psychotherapist. As such, researchers are more commonly turning to the use of advanced statistical methods, namely multilevel modeling (MLM), to address this complexity. In this article, we describe MLM for consumers of psychotherapy studies so they can better understand and evaluate studies that employ this method. We provide an example study that illustrates how traditional statistical methods may conceal meaningful findings. Also, we describe commonly utilized applications of MLM in psychotherapy research, such as variation among psychotherapist effects in outcomes, variation in the relationship between predictors and outcomes that can be attributed to psychotherapists, and sample size concerns.