Bias in participant ratings of psychotherapy process: An initial generalizability study.

Abstract

Rater bias has long been considered a source of error in observer ratings but has been ignored by process researchers using participant ratings. In particular, rater variance, or differences in generalized favorable or unfavorable perceptions of others, represents a neglected source of error in studies using participant ratings. The author presents an initial generalizability study of 2 commonly used psychotherapy process measures and discusses the implications for statistical power of studies using these measures and for interpretation of research findings. Results suggest that bias may be a substantial source of error in process ratings, accounting for 4 to 8 times as much variance in observed scores as intrarater error.