The results reported in the RREP study indicate that the original relapse taxonomy of the RP model has only moderate inter-rater reliability at the highest level of specificity, although reliability of the more general categories (e.g., negative affect and social pressure) was better. The model’s predictive validity also was modest; however, the definition of the key relapse episodes utilized in these studies failed to clarify whether these were voluntary change episodes or simply a return to drinking following a short period of abstinence that did not represent a serious attempt to quit drinking. Therefore, the RREP studies do not represent a good test of the predictive validity of the taxonomy.
