Posh? Exceeded your A level expectations? You are less likely to go to university
David Kernohan is Deputy Editor of Wonkhe
Tags
What with all the charts in my main piece on the UCAS End of Cycle sector level data, there wasn’t space to get into the perennial Wonkhe fascination of predicted and actual A level grades.
The classic “galaxy brain” take is to conclude that because predictions are generally higher than actual grades the predictions are (a) useless and (b) should be scrapped because they make university applications unfair – which in most cases leads to advocacy for some form of post qualification admissions.
PQA is a bit like seasonal flu – it turns up around the same time every year to give me a banging headache – but every few cycles we get a really bad dose that lasts for a lot longer.
Even in low risk years it’s worth taking basic precautions, and my recommended prescription is the UCAS End of Cycle data, which lets you look at actual points, predicted points, and the difference between the two.#
I’ve mentioned points and UCAS, so you are probably thinking that I mean tariff points. Not so – for this release, we use an alternative system where A* gets six points, A gets five points and so on. And only your top three A level results count.
Our first plot shows actual and predicted points by POLAR4 quintile. As a reminder quintile 5 are 18 year olds in areas where 18 year olds are most likely to progress to higher education, and quintile 1 are from areas where 18 year olds are least likely to progress to higher education. POLAR4 is looking rather long in the tooth now (it refers to people who entered higher education between 2009-10 and 2013-14) – I guess eventually UCAS will move to using the newer TUNDRA (updated methodology, 2012-16) quintiles at some point.
The orange lines are predicted grades, the blue lines actual grades and the red bars the difference between the number of accepted 18 year old applicants predicted that grade, and the number that actually got that grade (with a positive value meaning more people were predicted the grade than got it). You can filter by acceptance route (firm, insurance, clearing, and so on.
The first thing that sticks out to me is that 18 year olds from quintile five are more likely to be predicted three A* grades than anything less. This is a newish (post-pandemic) trend – and is at significant odds with what quintile five students are most likely to achieve, which is somewhere between three A grades and three B grades.
On the face of it, that is a slam dunk for team PQA. You can almost imagine those sharp-elbowed middle-class parents pushing their kids’ teachers to predict generously. But this is too hasty a reading. Teachers of less advantaged children are also likely to over predict their A level performance, and rather than muddying up the top of the chart this pushes the top end of this part of the cohort over the bar that gets them into what we used to call the selective end of the sector.
And that’s the thing. There’s a bigger difference between quintiles for actual grades than there are for predictive grades. The “sharp-elbow” story is not as compelling as the one where the ability to do well in your A levels is linked to your postcode (and thus your level of structural socio-economic disadvantage).
This is a bit clearer if we plot the differences between predicted and actual grades, as proportions. Each coloured bar here is 100 per cent of accepted applicants from that POLAR quintile for that year. You’ll see that quintile five are marginally more likely to be underpredicted (the green bits – this has become less prominent over the years), and that quintile one are substantially more likely to be overpredicted (the red bits).
So to remove the predicted grades would disadvantage a group that are less likely to go to university, and would give a consequent advantage to a group that are more likely to go. Which strikes me as adding unfairness to the system.
Many providers already use “contextual admissions” (lower required grades for people from disadvantaged backgrounds), and it could be argued that the system of predicted grades as a basis for offer making decisions is a sector wide version of this. Contextual admissions are an admission that it is simply harder to get good A level grades if you are from a disadvantaged background, no matter how “bright” you may be – and for people who aren’t on the Toby Young “IQ eugenics” train it is pretty clear that this is going to be down to structural disadvantage (less good schools, less time/space to learn and revise, less extra support or tutoring, more likely to be working or a carer) than because poor kids are thick.
If you wish to argue that the the differences between proportions is actually quite low (around 80 per cent of POLAR quintile 1 were underpredicted, compared around 70 per cent of quintile 5) you are on slightly less dubious ground. But this argument is the reason I have added the thin bars (which represent numbers of applicants) to the chart above. There are masses more accepted applicants in quintile 5 than quintile 1 – which is what you would expect, because that’s what POLAR is.
Why is this a factor? Well, there’s a strangeness relating to how likely people are to actually go to university after applying – based on their POLAR quintile and the difference between their predicted and actual grade. Basically the most advantaged group (quintile five) are far less likely to go to university at all if they do better than their prediction than their less advantaged (quintile one) peers.
I cannot overstate how counterintuitive this is. If you do better than your predictions and you are from a quintile one background your likelihood of going to university if you applied is just under 91 per cent. If you are underpredicted and you are in quintile four or five it is nearer 87 per cent. Posh bright kids are less likely to go to university if they apply and smash their A levels. And that’s less likely than all applicants who slightly underperform.
I have no explanation for this phenomenon. If you are wondering about a celebratory gap year, recall that this is the top fifth of 18 year olds, socio-economically – hardly the super-rich. There’s ample opportunity for these young people to trade up (via “decline my place”) and everyone other than Oxford and Cambridge is in clearing anyway.
But across all the quintiles other than (just about) quintile 1, 18 year olds are less likely to attend university if they over perform. Are they all in a huff with their teachers? Seems unlikely. Comments welcome.
Looking outside the U.K.?
One possibility is re-applying the following year for a higher-tariff university rather than “decline my place”, but you stated, “Basically the most advantaged group (quintile five) are far less likely to go to university at all if they do better than their prediction than their less advantaged (quintile one) peers.”, the crucial bit is “at all”, which is ambiguous in meaning. Is it to do with the how the five choices are selected? Following the advice of their teachers, the POLAR 5 quintile over-constrain their list of choices, so should aim higher throughout the list for each of the five choices than advised by their teachers. The teachers are likely to be using their own expectations of the students performance rather raw predictions, so tend to be cautious when providing advice about the selection of the five universities. Huff could be the explanation.
Though most universities go into clearing, not all courses do. For example, Medicine, Dentistry and Veterinary Sciences have capped numbers, often aren’t in clearing, and may have extra requirements like having to have already taken tests earlier in the cycle. Students often take a year out to reapply if they got the grades but don’t have offers.
Mixture of these students likely hoovering up the high profile degree apprenticeship places across the country and probably (usually independent schools) getting advice around exploring overseas (usually US) opportunities.
Be interesting to see if that phenomenon affects certain providers more than others…
Our top students rank US unis higher than British. So I agree with the last comment. I think these students are going to uni abroad.
I take your final point about the “celebratory gap year” being unlikely to explain away all of that gap, but I would definitely be more convinced if these stats took into account HE admission within two years of results rather than just the immediate cycle. If this is a persistent trend, would tracking this with HESA data be possible do you think or is that not really viable given a lack of access to the underlying datasets? I’d also argue that whilst you’re right that a lot of universities are in clearing nowadays, many aren’t in clearing across the board and whilst the pool of “highly selective” courses is diminishing, it’s certainly not vanished altogether.
Providers in the EU who teach in English are seeing higher enrolments from UK students than they ever have before. I can’t remember who told me this – I possibly have it from ENQA – but it’s certainly a thing I learned recently. So that lends some weight to the ‘going overseas’ argument. I would, right now…
Trading up to a higher ranked uni as a Q5 is harder than as a Q1, as the higher ranked unis will be desperately trying to hit their APP targets.
As with others comments, would be interesting to see whether ‘at all’ means ‘that year’.
Yes, the obvious candidate explanations are (i) going overseas to study or (ii) taking a year out (perhaps intending to work, but with economic support from parents as a backstop) to make an application to a higher tariff provider, either because the right course at the latter is not available in clearing or because they want to take the time to make a considered decision rather than rushing into something unplanned.
Maybe they need a break after overachieving….