Psychotherapy outcome research: Methods for reporting variability and evaluating clinical significance

The purpose of this article is to suggest some new directions for the presentation and reporting of data in psychotherapy outcome research. Statistical comparisons based on group means provide no information on the variability of treatment outcome, and statistical significance tests do not address clinical significance. Although psychotherapy research has begun to address these issues, it has done so unsystematically. New standards and conventions are needed to serve as criteria for classifying therapy subjects into categories of improved, unimproved, and deteriorated based on response to treatment. A two-fold criterion for determining improvement in a client is recommended, based on both statistical reliability and clinical significance. Statistical procedures for determining whether or not these criteria have been met are discussed.