Measuring predictive, summative and concurrent self-assessment: a study of GP trainees

Background: Self-regulation is essential for safe medical practice. Self-assessment ability has been shown to be poor in the predictive and summative modes but it has rarely been measured in the concurrent mode which may be closer to the all important ability of refl ection-in-practice (Eva and Regehr, 2005). Work done: UCL general practice trainees were required to participate in experiments run in 2005 and 2006. Participants sat a validated MCQ test on four domains. Self-assessment was measured before and after the test using a 7-point Likert scale for each domain (predictive and summative modes). Concurrent self-assessment was evaluated by spectral analysis of degree of certainty scores for each item. Results: 127 GP trainees (of 168) took part in 2005 and 120 (of 147) took part in 2006. Predictive and summative self-assessments were poorly correlated to MCQ score. Spectral analysis will be discussed further in the presentation. Conclusions: Our study confi rms previous research showing it to be very diffi cult for trainees to self-assess. Spectral analysis may prove a useful tool for concurrent self-assessment measurement and feedback