A method of moderation of school-based assessments: some statistical considerations
By Frances Good
Most GCSE examinations will include a school-based component marked by the teacher; adjustments to these marks may be required to give a common measurement scale for all centres.
Variations on a statistical method of transforming the marks are considered, together with the assumptions that are inherent in each. Formulae by which the accuracy of the resulting marks can be assessed are given. The meaning of the adjusted mark and the implications of using the moderator's or teacher's mark instead are considered.
A small data set is analysed to illustrate the suggested procedures and the differences in the marks awarded with each. It is concluded that a structural regression line should be used to calculate the adjusted marks; and that these marks should generally be used in preference to either the teacher's or moderator's marks.