State educated students at the University of Cambridge make up 59.3%, the independent sector comprising only 7% of UK secondary school pupils and 13% of those taking A-levels. This 59.3% is unevenly distributed across UK schools, a large proportion represented by selective grammar schools. In 2011 the Sutton Trust published a report showing that between 2007-2009 the combined total of pupils sent to Oxbridge by some 2000 UK schools and colleges was smaller than that shared between Westminster, Eton, St Paul’s, St Paul’s Girls’ and Hills Road sixth-form college (a state 6th form in Cambridge).
The university commits increasingly to Access through outreach programmes, but with statistics like these Cambridge rightly faces questions as to whether they are fairly admitting the best students the UK has to offer, or whether their admissions process favours students from the private/independent sector.
Recently government-appointed director of the Office for Fair Access, Professor Les Ebdon, supports the use of “contextual data” in university admissions. Applicant information is set into socio-economic or educational context, using information within or in addition to a UCAS application. High quality contextual data would be rich, reliable and verifiable. In an interview with BBC Radio Four’s Moral Maze, Dr Geoff Parks, Head of Admissions for the University of Cambridge said;“The limited contextual data we do receive is based on Censuses and other demographic studies which are at least a decade old”.
The extent to which contextualisation is based on a large amount of information specific to an individual currently depends on how much money, time and resources an institution can put into discovering this information. Cambridge does this more than other universities by interviewing more candidates and having additional application forms to probe background. Perhaps Les Ebdon could find a way of allowing universities to gain enough reliable individual data in order to increase the quality of contextualisation. This is obvious. What is more difficult is that all the elements of this contextualised data can themselves be put into context by looking at trends in degree performance and outcome. This is likely to vary between institutions and may affect the weight which is given to different aspects of candidate profile.
Educational disadvantage may, to varying degrees, stem from underperforming state schools which require improvement. Whilst universities do face the challenge of trying to correct for this, the role they can play in promoting social mobility is important. If enough contextual data tilts the balance in favour of a student from a disadvantaged background, whether this is “social engineering” or not doesn’t matter to the university: they are admitting the “better” student which is advantageous to the institution.
Use of contextual data is supposed to allow candidates to be selected on ability, on merit of their achievements relative to circumstance. One of the problems with representing ability by A-level performance is discrepancy in teaching methodology. A top independent school student may have been highly coached on exam-technique and spoon-fed to pass the exam, without actually having a higher aptitude for the subject or the independent research skills to embark on a challenging degree. How, therefore, do universities select based on ability? In Cambridge interviews, candidates are provided with problems which expose how they naturally think and approach difficult intellectual concepts. This not only measures ability but whether the institution is suitable to develop the level of potential demonstrated by a candidate. Cambridge (and Oxford) can do this because they have the resources to interview a huge proportion of applicants, but very few other universities do. So contextualising grades as much as possible is important, but if we can’t rely on A-levels as a true indicator of ability and the magic word is “potential”, other universities face a huge challenge. How can they measure potential, let alone put it into context?