Exploring the relationship between examiners' memories for performances, domain separation and score variability

Background: OSCE examiners' scores are variable and may discriminate domains of performance poorly. Examiners must hold their observations of OSCE performances in "episodic memory" until performances end. We investigated whether examiners vary in their recollection of performances; and whether this relates to their score variability or ability to separate disparate performance domains. Methods: Secondary analysis was performed on data where examiners had: 1/scored videos of OSCE performances showing disparate student ability in different domains; and 2/performed a measure of recollection for an OSCE performance. We calculated measures of "overall-score variance" (the degree individual examiners' overall scores varied from the group mean) and "domain separation" (the degree to which examiners separated different performance domains). We related these variables to the measure of examiners' recollection. Results: Examiners varied considerably in their recollection accuracy (recognition beyond chance -5% to +75% for different examiners). Examiners' recollection accuracy was weakly inversely related to their overall score accuracy (R = -0.17, p < 0.001) and related to their ability to separate domains of performance (R = 0.25, p < 0.001). Conclusions: Examiners vary substantially in their memories for students' performances which may offer a useful point of difference to study processing and integration phases of judgment. Findings could have implication for the utility of feedback.