An investigation into the accuracy of A-level predicted grades

This paper presents the results of an investigation into the relationship between teachers’ predictions of A‐level grades and the actual grades obtained by a sample of applicants for university places. In total, over 9,000 predicted grades from approximately 3,000 applicants were considered. The data are analysed according to gender, type of education centre attended, ethnic origin, age of applicant, examining board and subject. Summary data for each category are presented. A two level analysis of the data, with applicants nested within subject, was performed. The results of this analysis enabled interaction effects between the categories to be considered, thereby allowing some explanation of the picture presented by the summary information. The largest contribution to the optimism or pessimism of the predicted grade is the actual grade awarded, indicating that differences between grade distributions must be taken into consideration before offering any substantive interpretation of results for two sub‐categories, e.g. male and female. A slight gender effect was observed, with predictions for males being slightly more optimistic than those for females. A slight age effect is also apparent, with predictions for applicants aged 19 being slightly less optimistic than those for applicants aged 18. The mean difference between predicted and actual grade was not influenced by the ethnic origin of the applicant.

Share this page