How will flaws in predicted grades affect this year’s students?

Lee Elliot Major is Professor of Social Mobility at the University of Exeter.

A version of this blog was published by the Daily Telegraph.

It’s one of those puzzling aspects of our university system that has always caused controversy. In the UK, unlike any other country in the world, students apply for university degrees not with their actual grades, but with grades predicted by their teachers. Universities give conditional offers on the basis of these predicted grades. Offers, under normal circumstances, are then confirmed when students get their actual A-level grades.

The most obvious flaw of the system is that pupils’ predicted grades are wrong most of the time. The most comprehensive analysis to date finds that 84 per cent of applicants didn’t achieve their predicted grades in their best three A-levels. Three quarters of applicants were over-predicted: their grades were predicted to be higher than they actually achieved. Just eight per cent of applicants were under-predicted.

Some six per cent of candidates are over-predicted by as much as five grades in their three A-levels. That means a candidate who was predicted two grade Cs and a grade B for example actually ended up getting two grade Es and a grade C. On average among the 280,000 applicants, the mismatch between predicted and actual grades was 1.7 grades for the three best A-levels. Lower attaining applicants are more likely to have their grades over-predicted, whilst high attaining students are more likely to be accurately predicted by their teachers.

These figures, for students entering universities between 2013 and 2015, come from a working paper being published by two leading experts, Gill Wyness at University College London and Richard Murphy University of Texas at Austin. They confirm the conclusions of previous studies. The paper is being released by UCL’s Centre for Education Policy and Equalising Opportunities.

The study reveals differences in predictions by student and school background among higher academic achievers – those applicants achieving at least two As and a B in their three A-levels. High achieving students from disadvantaged backgrounds are more likely to be under-predicted (by around half a grade) than their more privileged high achieving counterparts. Meanwhile high achieving students from state schools are more likely to be underpredicted than high achievers from private schools.

Many want to abolish the admissions system based on predicted A-level grades because it’s bad for social mobility. Getting into the most selective degree courses is associated with higher future earnings in life. If students from disadvantaged backgrounds are ruling themselves out of applying to elite universities due to low predictions even though they achieve the actual A-level grades to get in, then they may be missing out. It may be that teachers provide accurate predictions for a given level of achievement to date, but do not anticipate late growth surges in student attainment among poorer students.

What we also know is that teachers are becoming less accurate in their academic forecasting with each academic year. In 2015, just over half of all English 18 year old applicants missed their predictions by two or more grades; an increase of 34 per cent since 2010.

What does all this mean for the coronavirus cohort of university applicants, without the luxury of actual A-level grades?

Most could be winners in the academic race. Universities may have to enrol far greater numbers of students on predicted grades this year because they won’t have the actual A-level grades to turn down their conditional offers. This may be a blessing in disguise. Many institutions are concerned about declines in numbers of international students in the wake of the coronavirus pandemic.

High achieving students from disadvantaged backgrounds meanwhile could lose out again.
Universities may want to consider lower grade offers for candidates who have shown true academic potential at school. Admissions officers could use the GCSE grades of this year’s candidates alongside predicted A-levels, as well as the personal statements and bespoke admissions tests where available. Exam regulators may be asked to review the consistency of predictions across different schools.

Meanwhile we don’t have any such comprehensive analysis of GCSE predictions, partly because they are only produced within schools. But if A-levels are anything to go by then we should also worry about the reliability of forecasts here. A difference in just one GCSE grade can have a life defining impact as well.


Leave a Reply