Paired Comparison Model

Thurstone's comparative judgement
The job of an exam is to judge the students using the right evidence

The notion that we need numbers in exams was developed in the University of Cambridge in the late 18th century, as a tool to avoid unfairness: but it is not necessary. The requirement is that we find some way to judge the students' performances in order to create the scale we need, and marking items to add up their scores is just the way we seem to have chosen to do this. Louis Thurstone proposed, in 1927, a method for constructing a scale based on direct comparisons of pairs of objects.

We introduced this method to educational assessment in 1993, and it has since become the standard experimental method for comparing standards in England.

We are now piloting the method for examinations, in collaboration with Goldsmith's College, University of London. Early pilots are very promising, and an internet-based system has been developed to collect the judgements from judges. Because judges compare students' work directly in terms of quality, we believe the method may prove more valid than marks-based assessment

Further information

For more information, download these papers from Our Publications:

Title Authors Date Publication
Let's stop marking exams Alastair Pollitt with assistance from Gill Elliott and Ayesha Ahmed June 2004 IAEA Conference, Philadelphia
What raters really pay attention to Author Author Author 1993 Location

See also: Bramley, T(2007) Paired comparison methods. In Newton, P, Baird, J, Patrick, H, Goldstein, H, Timms, P and Wood, A (Eds) Techniques for monitoring the comparability of examination standards. London, QCA.

Home | Our Models | Our Research | Our Training | Our Publications | About Us | Contact Us