Abstract
This randomized experimental study evaluated a web-based debiasing intervention designed to improve clinical judgment in the assessment of adult major depressive disorder (MDD) in Brazil. The intervention, adapted from prior work on pediatric bipolar disorder, taught Bayesian probability estimates and other debiasing strategies. Brazilian psychologists (N = 535) were assigned to either the intervention or an active control condition featuring a brief educational video, and were assessed post-intervention (n = 299). Outcomes were decision-making errors and diagnostic accuracy in response to vignettes, which depicted diagnostic challenges that could elicit cognitive biases. Assignment to intervention was associated with significantly fewer decision-making errors and greater diagnostic accuracy. Intervention effects were stronger for the vignette involving diagnostic risk estimation, as supported by intervention-by-vignette interaction effects. Intervention effects across vignettes were not statistically significant when only categorical diagnostic tasks were considered. Years of clinical practice and postgraduate training were largely unrelated to outcomes, and decision-making errors were not associated with treatment decisions. Overall, findings provide preliminary evidence of the transferability of this brief, web-based debiasing intervention to a distinct clinical and cultural context, although effect sizes were substantially attenuated compared with those observed in the original intervention.