Differences in Marks Awarded as a Result of Moderation: some findings from a teacher assessed oral examination in French

Abstract

Identical standards and criteria are unlikely to be applied by all teachers when marking school‐based components of examinations, so moderation procedures are required to check and, if necessary, adjust their marks. This paper considers differences between the marks awarded by teachers and moderators in a French oral examination when the moderators re‐marked the work without knowledge of the teachers’ marks. It finds that although teachers were generally more lenient than moderators, their rank orders of the candidates were similar. A general statistical method of transforming the marks to a common mark scale, when there is little overlap between the assessment objectives that are tested in the internal and external components, is outlined. The implications of the differences in the judgements of teachers and moderators for the statistical procedures are considered. It is suggested that a structural regression method should generally be used to transform the marks, and that all candidates should be awarded the teacher's mark adjusted in this way, rather than the moderator's mark or the raw teacher's mark.