What Is the Question? A Comment on "Investigating Treatment Mediators When Simple Random Assignment to a Control Group Is Not Possible"

Debate about psychotherapy research methods in the last decade has helped to highlight important issues related to the clinical utility of randomized controlled trials. Out of this debate, significant recent advances have been made in using randomized designs to examine treatment efficacy in settings more akin to those in the "real world" where most patients access mental health care. Doss and Atkins (2006) accurately note several potential limitations of treatment-treatment designs for testing mediation, especially when the same mediator may be operating in both active treatments, and they propose several ways that correlational methods can be useful in pursuing mediational hypotheses. These proposed methods are discussed and, using several specific psychotherapy research examples, the continuing relevance of randomized designs in answering specific questions about causality is underscored.