Experimental Construct Validity in the Evaluation of Cognitive and Behavioral Treatments for Depression

Distilled versions of Beck's cognitive and Lewinsohn's behavioral treatments for depression were crossed in a 2 × 2 design that included combined and high-demand control treatments as well. Multivariate and univariate analyses of pretreatment, midtreatment, posttreatment, and follow-up data revealed that the Cognitive Treatment factor produced a consistent and durable impact on a battery of devices reflecting cognitive manifestations of depression; some generalization to the behavioral domain occurred as well. The Behavioral Treatment factor failed to produce improvement within the corresponding behavioral assessment battery or on any cognitive device. Postmortem analysis of a full syndrome measure suggested possible evidence favoring each factor. Both conditions generated equivalent demand characteristics and counselor ratings of client adherence to treatment. No interactions involving the treatments occurred. The obtained pattern of convergent and divergent outcomes indicates considerable construct-valid strength for cognitive therapy applied to a moderately depressed population. Possible reasons for behavior therapy's comparatively weak showing are discussed.