The predictive validity of a situational judgement test and multiple-mini interview for entry into postgraduate training in Australia

BACKGROUND: Evidence for the predictive validity of situational judgement tests (SJTs) and multiple-mini interviews (MMIs) is well-established in undergraduate selection contexts, however at present there is less evidence to support the validity of their use in postgraduate settings. More research is also required to assess the extent to which SJTs and MMIs are complementary for predicting performance in practice. This study represents the first longitudinal evaluation of the complementary roles and predictive validity of an SJT and an MMI for selection for entry into postgraduate General Practice (GP) specialty training in Australia. METHODS: Longitudinal data was collected from 443 GP registrars in Australia who were selected into GP training in 2010 or 2011. All 17 Regional Training Providers in Australia were asked to participate; performance data were received from 13 of these. Data was collected for participants’ end-of-training assessment performance. Outcome measures include GP registrars’ performance on the Royal Australian College of General Practitioners (RACGP) applied knowledge test, key feature problems and an objective structured clinical exam. RESULTS: Performance on the SJT, MMI and the overall selection score significantly predicted all three end-of-training assessments (r = .12 to .54), indicating that both of the selection methods, as well the overall selection score, have good predictive validity. The SJT and MMI provide incremental validity over each other for two of the three end-of-training assessments. CONCLUSIONS: The SJT and MMI were both significant positive predictors of all end-of-training assessments. Results provide evidence that they are complementary in predicting end-of-training assessment scores. This research adds to the limited literature at present regarding the predictive validity of postgraduate medical selection methods, and their comparable effectiveness when used in a single selection system. A future research agenda is proposed.