Combining Grades from Different Assessments: how reliable is the result?

Abstract

Assessment usually involves combining results from a number of components. This has traditionally been done by adding marks and the issues raised are discussed in most books on assessment. Increasingly, however, there is a need to consider ways of providing an overall assessment by combining grades from component assessments. This approach has been little discussed in the literature. One feature of it, the likelihood that the overall assessment will be less reliable than one based upon the addition of marks, is explored in depth in this paper. The reliability of the overall assessment is shown, other things being equal, to depend upon the number of grades used to report achievement on the components. It is concluded that the overall assessment will be satisfactorily reliable only if the number of grades used to report component achievements is equal to, or preferably greater than, the number used to report overall achievement.