Today we publish a new report authored by Dr Gill Wyness of the UCL Institute of Education on the inaccurate predictions of students’ A-levels results. The report must be the catalyst for a change to a better and fairer admissions system. No other major countries employ a system that is based on flawed guesswork and the time has come for us to get rid of this outdated practice.
We need a system where students apply to university after they have received their results.
The report shows that just 16% of A-level predictions made for 1.3m students during 2013-2015 proved correct. This shows just how impossible a task it is for hard-working teachers to make accurate grade predictions, in some cases eleven months ahead of when students will actually receive their results. The report shows the failings of a broken system, and not the hard-working teachers tasked with the impossible job of trying to make predictions.
The report looks at the top three A-level results from all participants who sat A-levels in 2013, 2014 and 2015 went on to higher education through the UCAS system. This involved 1,356,055 young people (roughly 452,000 per year). The overwhelming majority (75%) of students were over-predicted. You could argue that this simply allows students to make ambitious application decisions. However, over-predicted students were more likely (14%) to go through clearing than their correctly (9%) or under predicted (5%) counterparts.
This matters because decision-making during clearing can be haphazard; it is time-pressured and constrained by the availability of places. We don’t want students turning to clearing having previously rejected an appropriate university place because of overinflated and incorrect predictions of likely success. It is noteworthy that there are no national statistics on the drop-out rate for students who go through clearing.
Around one in 10 (9%) of applicants were under-predicted. While this figure could be declared ‘small’, further exploration of the figure reveals that there are problems around fair access. Under-prediction is more likely to affect high-achieving students. A fifth of students (21%) achieving grades AAB at A-level were likely to be under-predicted, compared to 9% of the overall cohort.
However, accuracy varies dramatically according to how well students do at A-level. It is obviously much easier to over-predict students who end up getting low A-level results and much harder to over-predict the more successful candidates coming in with A*s. Crucially, the research shows that students from the most disadvantaged backgrounds who get the higher marks are slightly more likely to be under-predicted than their more advantaged peers.
So while disadvantaged students who get the lowest grades at A-level are the most likely to have been over-predicted, often by quite a margin, at the top of the scale amongst students getting the highest marks, disadvantaged students are actually slightly more likely to be under-predicted. A quarter (24%) of AAB applicants from lower income backgrounds are under-predicted compared to a fifth (20%) of AAB applicants from the highest income backgrounds. This is important because under-predicted candidates are also more likely to apply to, and be accepted at, a university which they are overqualified for.
Students should make their application decisions alongside good information, advice and guidance. However, poor predictions cannot offer helpful advice or guidance and could artificially reduce student choice. This compounds the fact that students from the most disadvantaged backgrounds are already less likely to apply to Oxbridge and other high tariff institutions because of reasons such as few family or friends considering it as an option, or little history of their school sending pupils there.
The admissions process is one area where the sector could work together to deliver improvements that could support fairer outcomes relatively quickly. The use of predicted grades as a proxy for actual achievement is flawed. Moving to a system which relies on actual achievement rather than predicted results would allow students and staff to make better and more appropriate decisions.
A recent report revealed that seven out of every ten staff involved with university applications back the move to a post-qualifications admission system. Both the previous and current heads of OFFA, Sir Martin Harris and Les Ebdon, support the introduction of a post-qualification admissions system.
A post-qualifications admission system is doable. It would increase transparency, eliminate the impossible task of ‘guestimating’ grades and support students to make more informed application decisions for themselves.