Abstract

Self-assessment ability in medical students and practising physicians is generally poor, yet essential for academic progress and professional development. The aim of this study was to determine undergraduate medical students' ability to self-assess their exam performance accurately in a real-world, high-stakes exam setting, something not previously investigated. Year 1 and Year 2 medical students (n = 74) participated in a self-assessment exercise. Students predicted their exam grade (%) on the anatomy practical exam. This exercise was completed online immediately after the exam. Students' predicted exam grades were correlated with their actual attained exam grades using a Pearson's correlation. Demographic data were analysed using an independent t-test. A negative correlation was found between students' overall predicted and attained exam grades (P < 0.0001). There was a significant difference between the students' predicted grades and actual grades in the bottom, 3rd and top (P < 0.0001), but not 2nd quartiles of participants. There was no relationship between the students' entry status into medical school and self-assessment ability (Year 1: P = 0.112; Year 2: P = 0.236) or between males and females (Year 1: P = 0.174). However, a relationship was determined for these variables in Year 2 (P = 0.022). The number of hours of additional self-directed learning undertaken did not influence students' self-assessment in both years. Our results demonstrate the 'unskilled and unaware' phenomenon in a real-world, high-stakes and practice-related setting. Students in all quartiles were unable to self-assess their exam performance, except for a group of mid-range students in the 2nd quartile. Poor performers were shown to overestimate their ability and, conversely, high achievers to underestimate their performance. We present evidence of a strong, significant linear relationship between medical students' ability to self-assess their performance in an anatomy practical exam, and their actual performance; in a real world setting. Despite the limited ability to self-assess reported in the literature, our results may inform approaches to revalidation, which currently frequently rely on an ability to self-assess.