Using the Rasch Model to Investigate Inter-board Comparability of Examination Standards in GCSE.

Abstract

By treating each examination as a polytomous item and a grade that a student achieved in the exam as a score on the item, the partial credit model (PCM) has been used to analyse data from examinations in 16 GCSE subjects taken by 16-year olds in England. These examinations are provided by four different exam boards. By further treating students taking the exams testing the same subject but provided by different exam boards as different subgroups, differential category functioning (DCF) analysis was used to investigate the comparability of standards at specific grades in the examinations between the exam boards. It was found that for most of the grades across the examinations, the magnitude of the DCF effect with respect to exam boards for the majority of the subjects studied is small, with the differences between grade difficulties for individual exam boards and the all-board difficulty in the unit of grade being less than one fifth of a grade. The effect of DCF varies between subjects and between grades within the same subject, with higher grades shown to be generally more comparable in standards than the lower grades between the exam boards.