Art of course assessment

Degree standards have been maintained by the switch in emphasis from exams to course work. We have been living through a decade and more of a declining unit of resource in British universities while expansion of student numbers has been enormous. In 1996 universities are more crowded and cheaper than ever before.

When one in three students now go to university whereas not long ago it was nearer one in ten, and when the institutions in which they study are characteristically ex-polytechnics bursting at the seams, it is hard to resist the view that what is being offered is a poorer quality product. Students do not see their tutors as much as before, classes are much bigger, and library spending (and space) has failed to match the increased demand.

In all logic, it seems that standards must have fallen. This is an enormously serious charge, not least because it appears to be a widespread opinion among employers, who tend to favour recruits from the elite universities.

Yet degree results continue to improve. How can those who insist that more means worse respond to the evidence that the proportion of students getting a good degree has risen from about 30 per cent in the 1970s to almost 50 per cent today?

Of course, the cynical reply is that standards in universities have fallen, that since students can enter higher education without an A level, then we ought not to be surprised that getting a degree is easier than before.

I am concerned about how today's graduates compare with those I encountered in the 1970s, I have been collaborating with Lisa Lucas and Graham Gibbs at the Oxford Centre for Staff Development to test the evidence on changes in academic standards.

Aggregated data might tell us that degree standards are rising, but records from Oxford Brookes' modular degree programme allow us to examine performance at the local level. Sociology has been part of the modular degree at Brookes since 1978 and we have accumulated a database. It was not difficult to extract data on sociology students' performance by module registration, mode of assessment and method of teaching between the years 1984 and 1994. At Brookes all modules are of an equal size in terms of learning hours and module grades receive equal weighting for degree results, so one can compare like with like across the years.

We decided to focus on the relation between module size and student performance, seeking to test the proposition that the larger the classes the less well students perform. We also looked at the relationships between patterns of teaching and of assessment and student attainment. What we discovered was rather surprising.

We did confirm the general finding that students do less well on big modules (more than 20 registrations) than they do on small ones (less than 20). It would seem to follow, therefore, that as classes have generally increased in size as student:staff ratios have expanded from about 13:1 to 22:1, then students will do less well and thereby the commonsense view, that standards are declining, will be confirmed.

However the records show that, in spite of a 70 per cent increase in enrolment, average performance of students at 58 per cent has been maintained and even improved over the decade.

All this is counter-intuitive. Students do less well in bigger classes, and classes are getting bigger, but then why are their overall scores being maintained and, if anything, improving? The answer is to be found in the connection between teaching and assessment and student performance.

What we did find was axiomatic to maintaining standards was the very significant change in methods of assessment introduced since 1984. Bluntly, course work as a proportion of total assessment has shifted from just 16 per cent of assessment in 1984 to almost three-quarters in 1994. Furthermore, there is a very positive correlation between course work and student performance, marks being significantly better on course work than in examinations.

This seems to be the key factor in accounting for poorer student performance in large modules. These are prohibitively expensive to run if course work is included, so the trend is towards having large modules assessed by examination, and it is with this means of assessment that students do particularly badly. Our suspicion is that, if a large module is assessed by course work, then standards will be maintained. As it is, however, it is smaller modules on which course work predominates, and where students do relatively well.

Some will argue that we have even relaxed our marking standards on course work since the 1980s. We took ten undergraduate dissertations from 1984 and another ten from 1994 and had them marked blind by four assessors. The experiment was less than perfect but it gave no support to the proposition that marking standards have eased over the years.

It seems to us that it is the prevalence of course work assessment that is the key to the maintenance of graduate standards. Since our students have a similar socio-economic profile, and since they come with much the same A-level scores as before, then it is hard to resist the view that course work is the primary reason for good performance. It is certainly a moot point whether degree standards would be the same in 1996 if we had retained the norm of three-hour unseen examination papers.

This does not mean we now provide degrees of a lower standard. A good case can be made for the appropriateness of course work assessment over examinations. Nevertheless, course work is damnably expensive in terms of staff time and energy. As cuts inexorably continue, then there must be others who are tempted to return to examination-only assessment since it is so cheap to offer. Were that to happen, then our fear is that a real decline in student performance will be registered.

Frank Webster is professor of sociology, Oxford Brookes University.