Can Australian medical students’ predictions of peers’ responses assist with gaining reliable results on course evaluations?

Abstract

Introduction: Student feedback is integral to continuous improvement of medical programmes. A key challenge with student course evaluations is gaining large enough response rates for results to be reliable. This study investigated whether student predictions of peer, rather than personal, responses could address this challenge.Method: An anonymous paper-based student experience of learning and teaching (SELT) survey was distributed to the Year 1-3 medical student cohorts. Students responded to 20 survey statements, using a 6-option Likert-type scale. Ten statements evaluated students' personal perspectives of the course, while the other 10 statements asked students to predict the most common response by their year cohort. Mean scores between the individual opinion-based and prediction-based statements were compared. An iterative process involving random subsampling was conducted to enable calculation of the minimum required number of responses for a stable outcome for each statement.Results: Two hundred and fifty-nine students participated (response rate 81.7%). Three out of the 10 paired statements in the prediction-based survey accurately predicted the group opinion-based mean. For the remaining seven statement pairs, there were statistically significant (although small) differences in mean. The calculation of mean number of responses required for a stable outcome found that the prediction-based SELT required significantly fewer (189) responses than the opinion-based SELT (215) (95% CI 15.3-35.7, p < 0.001).