There’s so much in the A&P data that it’s difficult to fit it all into a single article. Here’s some interesting findings derived from yesterday’s dashboards.
I’m focusing particularly on attainment and access. The data for progression is the same DLHE data that has been around for ages, and the most recent continuation data is from 2018-19 – interesting in itself but in many ways data from another world. For all these examples I’ve picked out larger universities and, where possible, statistically significant findings.
It’s attainment that is likely to get most interest – we saw (as I noted elsewhere) clear evidence that “no detriment” approaches were particularly helpful in removing or lessening attainment gaps, a finding that should prompt the whole sector to consider what it is about traditional end of course assessments that means that the achievement of non-traditional students is harder to recognise.
The analysis of this data from OfS very much stayed out of this issue. Chris Millward said
While there is evidence that the gap in attainment for black students compared to their white peers is closing, it remains far too high.
He’s right – the gap (frankly, any gap) is far too high. But I am surprised he displayed less interest on why the gap may have shrunk this year, and what the sector could learn from it.
UCL had the smallest gap between the attainment of Black students and white students. The difference between the proportions of each achieving a first or 2:1 was just 3 percentage points for all, a quarter of what it was five years ago. At the University of Buckingham, the gap is 45 percentage points, and Buckingham was one of the few places where the gap grew between this year and last.
Across the whole sector, students from IMD quintiles 1 and 2 (the most disadvantaged backgrounds) don’t do as well as their more advantaged peers. The gap for 2019-20 was 10.1 percentage points, and has remained fairly constant over five years. At Hartpury University – there is no gap, and at UCL the gap is just 2 per cent. In contrast, the University College Birmingham reported a gap of 12 percentage points. Over the sector, 81.2 per cent of white working-class (IMD1-2) students got a first or 2:1 – including 98 per cent at the University of Bristol but just 55 per cent at Buckinghamshire New University.
At most providers in most years a greater proportion of women than men get a first or 2:1 – for the sector as a whole the gap is 2.5 percentage points. But there were a few places where men did better last year. At Arden University the gap was 10 percentage points in favour of men, and at Coventry University it was 3 percentage points. This represents the first time in five years that the sex gap at Coventry has been in favour of men.
At Buckinghamshire New University students with cognitive and learning disabilities were 16 percentage points more likely to get a first or a 2:1 than their peers without a disability. This contrasts with a 9 percentage point gap in the other direction at City, University of London. Ninety-eight percent of students at UCL with a reported mental health disability did well at their degree, compared to 70 percent at Bedfordshire. At LSE, 97 per cent of all students with a disability got a first or a 2:1, compared to 68 per cent at Bolton.
Students who start their studies under the age of 21 were on average 9.6 percentage points more likely to get a first or 2:1 than their older peers. At the Manchester Metropolitan University mature students were 5 percentage points more likely to get a good degree than their younger counterparts, a reversal from previous years. Younger students saw a particular advantage at the University of Warwick, with a 20 percentage point gap in their favour.
Students from POLAR4 Quintile 5 (those most likely to have a background where it was common to attend university) were 6.3 percentage points more likely than other POLAR4 quintiles to get a first or a 2:1. This gap was widest at the Royal Veterinary college (15 percentage points) But at Falmouth University, that group was 4 percentage points less likely to do well.
Last cycle was a strange moment in time to be accepting a university place. If you think back to the summer you’ll recall that there was much concern that numbers would be down (students put off by the prospect of online provision) and that application patterns would shift (pandemic experiences cause a rethink in life plans). In actual fact it was a strong year for applications to HE, and the examishambles helped more people get the place they wanted.
The difference between application rates from POLAR4Q1 and POLAR4Q5 was 17.7 percentage points – continuing an ongoing trend and achieving significance over 5 years. We can’t talk about POLAR4 without talking about Teesside University – the gap there is now 21.6 percentage points in the other direction! POLAR4Q1 is far more likely to be the background of Teesside students, this year saw another four percentage points added to this gap. As you would expect, Oxford and Cambridge are at the other end of the scale – for both there is a 50+ per cent gap between POLAR34Q1 and POLAR4Q5 participation, though this is declining over time.
Coventry University leads the way on access for Black young people – with a 25 percentage point gap between the proportion of 18 year old students admitted and the proportion of 18 year olds in the population. For the sector as a whole the gap is 3.5 per cent, but at Teesside (who recruit locally in a deprived area with few Black students) there’s a gap of 2.3 percentage points in the other direction.
Sector wide attainment
Which subject should you be doing if you want a first or a 2:1? We get a set of broad subject areas for the whole sector only – and medicine and dentistry is the answer (93.2 per cent of students get a first or a 2:1). Conversely education courses must be more difficult – 79.4 of qualifiers got the “good” degree.
There’s also a sector wide attainment breakdown by entry qualifications – wherein we learn that more than 90 per cent of each A level grade cohort above BCC got a first or a 2:1 – and that 78 percent of students who got DDD or below at A level achieved the “good” degree, an achievement rate above any group of BTEC grades.
There’s much more in this data – and luckily for you I’ve built a pair of dashboards to help you explore.
Although there is data on access, continuation, attainment, and progression only access and attainment includes data from last year. In particular, progression links to the last year of DLHE (still!) so there’s nothing new in there since the original dashboards. Adding Graduate Outcomes data was on the roadmap, but the sheer amount of other things OfS are doing – and the difficulties inherent in developing a sensible five year time series from one year of data – seem to have stalled this.
The A&P data is a tremendously valuable dataset that offers a nuanced view of the student lifecycle, and how parts of it may be disadvantaging non-traditional students. The sheer breadth of the data makes it complex to work with, and presentational choices do not (for me, at least) help. But it is absolutely worth persevering with.
Change is coming to the data: 2021 will see data on household residual income and care experience added to the dashboard at a sector level, along with indicators linked to the OfS’ experimental Associations Between Characteristics (ABCs) work. There’s also a promise of continued work on making the data accessible and usable.
Unfortunately, the plans linked to this data are very much a blunt instrument. An evaluative report from Nous notes that smaller providers lack resources to engage with the data-intensive plan development and approval process, and that providers of all types had seen the OfS approach to target negotiation move from helpfully challenging to being unrealistic and divorced from a provider’s own plans and strengths in their relentless pursuit of sector-level KPMs. A concerning 68 per cent of providers were unable to agree that the plan assessment process was fair and reasonable.
OfS’ response to this report doesn’t exactly address the central issue of trust and applicability of communications. The closest we get is a line that:
we believe the relationship should entail mutual respect and, ultimately, work towards shared objectives.
The idea of a “degree of distance from the sector” and a “dispassionate” (really?) focus on the interest of students does not and should not make for a needlessly antagonistic ongoing relationship. The “principles-based” regulatory reset is the only action that the regulator feels is needed.
5 responses to “The new narrative in data on access and participation”
Thanks for a fascinating article as ever! So much to explore Quick question on where I can find more info about the planned additions of household income & care experienced status and what underlying data will be used? Thanks
Hey Wonkhe this article is very helpful and informative, but whenever you refer to a gap please can you explain exactly what you mean (it seems to mean different things at different times, sometimes you explain it, sometimes not). I realise this might upset the flow of the article so perhaps you can include these explanations as clickable links?
Has OfS carried out any analysis of the “unwarranted” gap in degree classification for black students/students from POLAR Q1 etc after controlling for A-level attainment, degree subject and other demographic characteristics similar to what they’ve done for grade inflation?
The Access data covers 2015/16 to 2019/20, e.g. the most recent cohort mostly entered in Sept 2019, before the pandemic. The sector hasn’t submitted 2020/21 HESA data yet, so the strange admissions cycle of 2020 will only be included in this time series next year, despite what the article implies by starting that section the way it does. I agree that the attainment data, which does reflect the impact of the pandemic, holds some potentially valuable lessons for the sector.
” But I am surprised he displayed less interest on why the gap may have shrunk this year, and what the sector could learn from it.”
And there, in a nutshell, is the problem with the whole approach. Lots of numbers, no insight