Some observations on the problems of grading examinations with several components: a reply to P. J. Squire

By Roger MurphyRobert Adams

Abstract

In a recent article in [Educational Studies], Squire (1978) describes a phenomenon well known to those concerned with examining; namely that when candidates' final marks are obtained by summing a number of examination components, with or without differential weighting, the dispersion in the final marks tends to be less than that in the component marks. This he ascribes wrongly to "regression to the mean", an effect which is commonly misunderstood, and which in any case preserves dispersion in the regressing variable. (For a fuller description of "regression to the mean" see Nunnally, 1964, pp. 92-96). The phenomenon described by Squire is in fact a consequence of the fact that scores obtained on the different components of examinations are never perfectly correlated.

How to cite

Murphy, R. and Adams, R. (1979). Some observations on the problems of grading examinations with several components: a reply to P. J. Squire, Educational Studies, Vol. 5, Iss. 3.

Keywords