A true test of talent

John Stein turned down Laura Spence. He is developing a test to identify smart students for Oxford

Laura Spence has done brilliantly to get five As, despite the furore that followed Gordon Brown's attack on our decision not to give her a place to read medicine at Magdalen College, Oxford.

I am glad to say that the students we did give places to also did well. However, Mr Brown's criticism did publicise the efforts many of us at Oxbridge have been making to increase the number of able students we admit from underprivileged backgrounds. We have tried to boost our intake of state school students from its present 53 per cent up to 60 per cent because 60 per cent of those who get three As at A level are in state schools. But a high proportion of these state school students have also been quite privileged.

We would like to increase our proportion of highly able pupils from poorer backgrounds. But, unfortunately, too few apply. Unless talented students from inner city comprehensives can see that they have as good a chance to win a place as those from Eton they will not apply.

Improving our ability to detect talent in applicants who have not had the benefits of a private education is unfortunately not just a matter of counting exam grades. Exams are not particularly good predictors of success at university and they are not good enough discriminators. About 20,000 pupils a year get three As at A level, of whom half apply to Oxbridge. But we only have room for about one-third of them, so we need extra discrimination.

Independent schools are better at getting their pupils As in exams: they enter only 15 per cent of the UK pupils taking three A levels yet achieve 40 per cent of those scoring triple As. Comprehensives enter 30 per cent of the pupils yet achieve 25 per cent of the triple As. In fact, Peter Lampl's Sutton Trust recently discovered that a working class pupil has a less than 1 per cent chance of getting into a top university, whereas an independent school pupil has a 25 per cent chance of doing so.

What we need are techniques for identifying high ability and motivation in students independent of the quality of their teaching. This would enable us to separate the really talented three-As students from the merely well-taught products of independent schools. This would also help us to detect high ability in the underprivileged who are not going to get three As because their teaching has been indifferent.

Two years ago, Jane Mellanby and I began to think about how to design tests to detect high ability and motivation in students who have not gained the knowledge base and self-confidence that independent schools seem better at giving their pupils. We wanted to devise a test that measured problem solving ability (fluid intelligence) and motivation, without requiring a large knowledge base (crystallised intelligence).

Jane had found that IQ tests were not suitable because many of the subtests depended on factual knowledge and, anyway, they were not very good predictors of university success.

In recent years there has been much research on how student learning styles affect progress. "Deep learners", motivated by a genuine interest in the subject, tend to be successful in the long term, though not necessarily at the multiple-choice exams that favour the "surface learners" of the "facts".

We have found that simple questionnaires can distinguish deep from surface learners.

Other recent research has attempted to assess ability to think beyond given facts by asking students to comment on a passage of text about the dangers of overreliance on chemical fertilisers. Answers did not depend on prior biological or chemical knowledge.

Last December, with financial assistance from the Sutton Trust, we piloted this test on 150 volunteers, all applicants for university. State school pupils performed no differently from independent school pupils, although their GCSEs and A-level predictions were lower. So our test probably was less dependent on the knowledge base that independent schools seem better at imparting.

Eventually we need to see how well our test predicts performance in whatever universities the students go to. The first step was to see how well it predicted those who gained offers at Oxford. Deep learning together with the text commentary turned out to be an extremely good predictor, which suggests that it identifies the same kinds of things that Oxbridge admissions tutors are looking for, but without "culture bias".

If subsequent comparison with other university achievement milestones, together with a planned larger study, confirms these preliminary results, then we will incorporate it into our admissions process. This should help us attract more bright students from inner-city comprehensives because they will know that their application will be considered impartially.

John Stein is senior medical tutor at Magdalen College, Oxford, and was part of the admissions panel that rejected Laura Spence.

* Do universities need to develop special selection tests?

Email us at soapbox@thes.co.uk.