Survey on the Use of Clinical and Mechanical Prediction Methods in Clinical Psychology

We surveyed 491 American Psychological Association division 12 (clinical psychology) members regarding their professional use of clinical and mechanical data combination (CC and MC) in making clinical predictions; 183 (37%) responded. This is the first report of CC and MC utilization frequency known to us. Nearly all respondents used CC in practice (98%), while fewer used MC (31%). Respondents gave reasons why they did not use MC, the most common being that it is conceptually misguided. In addition to computing odds ratios for variable comparisons, we constructed a multivariate regression model to predict use of MC, using a bootstrapping method that returns an estimate of the model's cross-validated validity. This procedure returned the null model: no predictor was powerful enough to replicate upon cross-validation. The extent to which MC was discussed in graduate school significantly mediated nearly all other relationships between variables of interest. Implications of these findings are discussed.