The Comparison and Scaling of Student Assessment Marks in Several Subjects

SUMMARY This paper is concerned with the comparison and scaling of assessment marks when these are available for several subjects for a group of students. It is assumed that it is important for the marks in different subjects to be in reasonable agreement because, for example, average marks have to be used to grade the students. A method of comparing subjects based on cumulative sums (CUSUMS) of differences between subject marks and student mean marks is proposed. A randomization test can be used to see whether observed CUSUMS are within the bounds expected if subjects do have comparable marks. If scaling is needed to bring subjects into line, then methods for doing this are suggested.