Aims To explore whether child marking of student performance is valid and has similar statistical characteristics to other simulated patient (SP) scores.
Methods 28 primary school children aged 8–10 years participated in a summative year 4 medical student OSCE for 262 students, covering all year 4 specialities. The children underwent cranial nerve examination, and were asked to assign a mark out of 10 in response to the question “If you had to see a doctor again, how happy would you be to see this one?”. The paediatrician examiners were asked to independently predict the child's score. Data were analysed using SPSS with Pearson's correlations. Child score (CS) was correlated against: examiner prediction of child score (ES); subscores for organisation and communication (SSOC) within that station; and examiner global rating (EGR) (which contributes to standard setting). CS was then also correlated against summated SSOCs in all other stations, and with summated SP scores for the 10 stations with an adult SP, whose scores were analysed similarly.
Results CS and ES were highly negatively skewed, but the examiners' scores were less skewed (skewness −1.22 and −0.67, respectively). The Pearson correlation between CS and ES was 0.40 (n=217, p <0.001); therefore, ES accounted for 16% of variation in CS. Within the station, there was little evidence of correlation between CS and SSOC, or with EGR. This is in contrast to other stations, where SP scores were highly correlated with these measures. Across the whole OSCE, CS did significantly correlate (0.20–0.26) with summated SSOC, and summated EGRs. However, there was no significant correlation with summated SP scores.
Conclusions There was some correlation between child score and examiner prediction of child score, although children were more likely to assign higher ratings. Paediatricians could not accurately predict child score. Within the station, child score did not significantly correlate with subscores for organisation and communication, in contrast to adult SP scores. However, child score did correlate with these subscores across the whole exam. This indicates either that the child score is a less reliable indicator of student performance, or measures a different facet of performance.