David Miller and Ricarda Zoellner (Letters, November 24) have fallen into the trap they set for themselves by putting numbers (not "percentages") on students' work and not using explicit criteria of quality. In my letter (November 17), I used the 15 grades A+ to F- (no Es).
Long before benchmarking, we avoided grade drift by agreed definitions of level of performance for each letter grading.
The only way to agree a single mark for degree classification is to take a weighted mean of numerical equivalents of the letter grades.
We corrected for the tradition of making the borderline 70 a first and 40 a third, with its huge ranges for first-class and failing work, and for regression to the mean, by "stretching" the numerical equivalents of the letter grades above and below B-/C+. Unfortunately, the institution has banned the best of this triumph of rationality and transparency.
David Booth.
Faculty of psychology Birmingham University