This week, we have published research to show that the use of predicted A-level grades in the university application process is deeply flawed.
We commissioned Dr Gill Wyness at the UCL Institute of Education to delve into the accuracy and impact of predicted grades. The data, based on predictions and outcomes for 1.3million students between 2013-2015 is conclusive: just one in six (16%) university applicants achieve the exam grade points that they were predicted to achieve by teachers or lecturers, based on their best three A-level results. Indeed, 75% of applicants were over-predicted and in nearly one in 10 cases (9%), students achieved higher grades than they had been predicted. This is problematic because predicted grades are the main basis for student and institution decision-making.
I want to be clear that this is not an attack on teachers who do a great job in often very difficult circumstances. The reality is that predicting grades is an impossibly difficult task. It is very hard to predict student behaviour - whether a student will produce a final sprint close to exams, or suffer a personal event that knocks them off-kilter. Teachers are expected to make predictions nearly a year in advance of when they actually sit their exams. The problem is that these predictions can be life changing! Surely it would be better to have a system which allows teachers to focus on teaching and advising their pupils.
The research published this week supports the argument for a move to a system which relies on actual achievement rather than predicted achievement. Doing so would support staff and students to make better and more appropriate decisions. It would be fairer and more transparent.
Overall, the report found that students from disadvantaged backgrounds and state schools are most likely to be over-predicted whilst students attending independent schools are more likely to receive accurate predictions. The data shows that the accuracy of predictions varies according to how well students do in their final examinations. This makes sense. It is clearly easier to over-predict a student with lower A-level grades because it is difficult or impossible to over-predict a student who achieves As or A*s. Also, there are typically more higher achieving students in independent schools.
Crucially, the research shows that students from the most disadvantaged backgrounds who achieve marks at the top end of the scale are slightly more likely to be under-predicted than their more advantaged peers. This is important because under-predicted candidates are also more likely to apply to and be accepted to a university which they are overqualified for. This 'mismatching' has implications for social stratification across universities and of course, wider social mobility.
It would also be apt to respond to Mary Curnock-Cook's response to our report 'Don't change the UK university application system - it works just fine'. This is not what UCAS said in 2011 when it consulted on moving to a process where students apply to higher education after they receive their results. At that time, UCAS said:
"... there are aspects that could be improved and many applicants are asked to make choices about courses and HEIs before they are ready. The cumulative effect of predicted grades, insurance choices and Clearing have led to a system that is complex, is thought to lack transparency for many applicants and is inefficient and cumbersome for HEIs. Only the best informed applicants and advisers are able to optimise UCAS applications and there is an undesirable divide between those applicants who receive effective advice and those who do not... the Clearing system is inefficient, stressful and confusing for applicants".
Curnock-Cook argues that nearly three-quarters of students are placed in their first choice institution. We are arguing that students' choices might have been different if they were able to make a more informed decision: under this system they have been mismatched with institutions.
Furthermore, clearing being available to students who "change their minds, or prefer to apply after their results" is not a fair resolution to the complex clearing system. Rather it risks creating a two-tier system because on lots of courses, student numbers are fixed, so places may already have gone. Our research also found that over-predicted students are more likely to go through clearing. Another example of how the application system might skew the decision-making process.
The decision to apply to university is a high-stakes one, with lifelong implications for students. A quick glance at student forums and Twitter conversations between students about predicted grades shows rejection of Mary Curnock-Cook's idea that students should "get their university applications out of the way early".
This week's research provides further evidence about the flaws in the university application and admission process. In a report we published last year, 70% of staff with a role in admissions backed a move to a post-qualification system. The higher education sector is not famous for warmly embracing change, but the case for change, fairness and transparency is only getting stronger.
Sally Hunt is the general secretary of the University and College Union