Grade predictions for school-leaving examinations: do they predict anything?

The attributes of 721 medical students admitted to the United Medical and Dental Schools (UMDS) of Guy's and St. Thomas' Hospitals, London, UK between the years 1991 and 1994 were examined to determine the relationship between A-level grade predictions (as completed on each student's UCAS form by their school/college at the time of their application to the school) and subsequent assessments of their academic ability and performance. Predicted A-level grades were found to be significantly, if weakly, correlated with the rating of academic ability made at interview by the UMDS interviewing panel. They were not however, related to the grades obtained by students in the A-level examination itself. Further, while success at pre-clinical examinations was predicted by obtained A-level grades, it showed no relationship with the predicted grades. In contrast, the interview-based rating of the applicant's academic potential was significantly predictive of subsequent A level and, to a lesser extent, pre-clinical examination performance. It is concluded that predicted A-level grades may not offer a valid method of assessing the academic potential of applicants for medical school.