Race Matching in Predicting Relational Therapy Outcome: a Machine Learning Approach

Abstract

ABSTRACT This study explores the relationship between therapist-client race/ethnicity matching on client treatment outcomes and whether other demographic factors contribute to treatment outcomes in a training clinic. An ANCOVA was conducted to examine the differences between race match and mismatch groups. A random forest algorithm was used to determine how racial matching conditions and other factors, such as gender, predict treatment outcomes. We found significant relationships between therapist-client race/ethnicity matching conditions and treatment outcomes for clients who received at least 10 sessions of therapy. However, results of the random forest algorithm indicated that race/ethnicity matching is one of the weakest predictors of treatment outcomes. Clinical implications and the limitations of the study are discussed.