Self-efficacy and relapse in smoking cessation programs.

The utility and validity of Bandura's self-efficacy theory and Marlatt's theoretical model of relapse were evaluated in a study of 78 cigarette smokers from two different cessation programs. Pretreatment, posttreatment, and follow-up assessments of smoking behavior, self-efficacy, and mood states were obtained. Efficacy state was found to be significantly enhanced as the result of both treatment programs. Subjects' scores on the seven clusters from the posttreatment efficacy state inventory were used as predictor variables in a multiple regression analysis to predict which subjects would relapse and how long, on a dichotomous time variable, they would remain abstinent before relapse. Multiple correlations, corrected for shrinkage, were .57 for relapse and .69 for time to relapse. A microanalysis revealed an extremely high correspondence between the cluster of smoking situations in which relapsing subjects experienced a low degree of selfefficacy and the situation in which relapse first occurred (weighted kappa = .89). Analysis of mood and efficacy data during follow-up indicated that relapsing subjects demonstrated aspects of a cognitive dissonance reaction and a personal attribution effect that were consistent with Marlatt's description of the abstinence violation effect. Research and clinical implications of the findings are discussed.