The study analysed the finals results of all Cambridge undergraduates in 1997 and 1998. These revealed that not just gender but also a student's ethnic origin and school background have an influence on the marks they achieve. It also emerged that the connection between achievement and social factors varies widely in different disciplines.
For example, although men outperform women in mathematics, the clear high performers in maths are students from ethnic minorities: 40 per cent of students from all ethnic minorities were awarded a mathematics first compared to 33 per cent of white students. However, students from ethnic minorities do not perform as well in history, where they are awarded 12 per cent fewer firsts, 10 per cent fewer 2(1)s and 20 per cent more 2(2)s than their white contemporaries.
One particular ethnic group, black students, is performing very poorly across the university. Only 3 per cent of black students achieved a first class in the two years studied; while 16 per cent were awarded a third class (compared to a Cambridge average of 2.6 per cent).
The analysis also threw up some seemingly odd patterns. For example, students who described their ethnic background as Indian performed strongly compared to other ethnic groups. But this was only true of Indian students from independent schools - more than 30 per cent of whom were awarded firsts. Indian students from state schools performed very poorly and only 5 per cent were awarded first-class marks.
What is so striking about these results is that previous educational achievement - typically GCSE and A-level scores - is pretty much "fixed" at Cambridge. Most students have been the highest achievers at school, and so it is tempting to ascribe the differences at finals to something within the Cambridge system. Perhaps a stuffy, male-dominated, public school atmosphere within many colleges militates against the success of certain students?
However, social class was unrelated to marks in finals. And, contrary to suspicions, students coming to Cambridge from the independent sector do not outperform their contemporaries from state schools across the university.
But a student's school background does seem to have an effect in somesubjects. For example, while women and men studying law at Cambridge performed equally well, those from comprehensive and grammar schools achieved significantly more first-class marks (17 per cent) than those from independent schools (11 per cent). In history, on the other hand, 24 per cent of students from independent schools were awarded firsts compared to 8 per cent from comprehensive schools.
A popular explanation for sex differences in performance at undergraduate level has been that men and women have different written styles in exams. Men, as nature's "hunters", are more inclined to take risks and hence display more of the originality required for first-class marks. Women, as "nurturers", play their final exams safer - producing competent but not exceptional, second-class work.
But much of this explanation goes out of the window in light of results from the Cambridge study. Not only are there a number of instances of women out-performing men in some subjects (the biological sciences, for example). The differences in performance of students from different ethnic groups, and from different school backgrounds, suggest an alternative explanation.
The subject a student studies is clearly a very important factor in terms of academic performance at Cambridge. For a start, different disciplines award wildly different proportions of firsts, 2(1)s etc. So even though women get far fewer firsts in maths than men, and men and women get the same number of firsts in law, a woman mathematician still stands almost twice as much chance of getting a first than her lawyer friend.
But what also emerges from the Cambridge results is the rather particular way in which different social factors (gender, ethnicity and school background) have influences on performance in different disciplines. It is almost as if each subject has its own profile in terms of which students - male or female, black or white, and so on - will succeed.
One explanation for these subject-area differences might be that the social expectations of a student's teachers and peers might have an influence on individual confidence levels. For example, if there is a general expectation that a brilliant young physicist will more likely be male, a brilliant young woman might find it more difficult to convince her teachers (and college peers) of her ability. Similarly, if there is a prevailing belief that the skills required for excellence at history are better honed at an independent school, a student from a comprehensive may find it difficult to fit the mould, and he may find less expectation of success from his teachers and peers.
Clearly, a next step is to discover how far the results at Cambridge extend across British universities. Previous studies of gender and performance at undergraduate level have indicated that, nationally, men and women can anticipate very different rates of success in different disciplines. If the same is true for undergraduates from different ethnic groups and social backgrounds, then it would appear that the subject a student studies is perhaps the most crucial factor in predicting eventual academic success.
The writer lectures in psychology at Goldsmiths College, University of London. The results of his research are reported in `The Curriculum Journal'