There are three major statistical dimensions of equality in 18-year-old entry to higher education: where you live, men/women, and ethnic group.
Some of these dimensions get more attention than others, but in terms of people they each reflect broadly similar equality deficits. There are in the territory of 40,000-60,000 of each of the under-represented group(s) ‘missing’ from HE, compared to if they had the same entry rate as other group(s). If you’re interested in equality, and are data-led, you’ll give roughly equal attention to each of these. If you’re really interested, you’ll use multidimensional measures like the UCAS multiple equality measure (MEM), which account for the complex ways these dimensions can interact.
But the MEM data sets have not been published yet, so we’ve used the (excellent) UCAS ‘day 28’ data sets to look at each dimension individually. We’ve taken 18-year-olds, and used POLAR in England for “where you live”, and UK-wide measures for sex and ethnic group. We’ve considered entry to UCAS-recorded HE as whole, the data on entry to different types of university is not available yet.
Where you live
POLAR groups people according to the entry rate of their immediate neighbourhood. It is mostly driven by a rich/poor wealth dimension and works because the UK has a highly stratified residential structure. We’ve used POLAR3 here for technical reasons (its definition window interferes less with the period we are most interested in) but the results are much the same with the updated POLAR4.
Increases in the entry rate of the lowest entry rate group (Q1) since 2012 has been very strong – but this has changed in 2018. Typically, young people in these areas have been around 4% to 8% (proportionally) more likely to go to university every year. This year that growth in university entry chances has collapsed to almost zero. Only the fee-perturbed year of 2012 was worse.
The importance of where you live to your relative chances of entering university has been declining rapidly since 2012, mostly driven by this strong growth from lower entry rate areas. With that growth gone, the trend of rapidly decreasing inequality ratios has gone too. The Q5:Q1 ratio has typically reduced by 5 to 15 ‘ratio points’ each year. In 2018 it has barely moved, reducing just a single point – only the supply-squeezed 2011 cycle has a comparably small reduction. More comprehensive measures (comparing upper and lower halves of the population) give a similar pattern.
Back in the summer of 2015, after two strong years of Q1 entry rate growth, the Government set (and still observes) a Q1 entry rate target of 27% by 2020. Reaching it no doubt seemed like a fairly safe bet at the time. Not anymore. At 2018’s rate of growth (0.1 percentage points) that 2020 target won’t be hit until 2085, a substantial 65 years late.
As a triple (low growth, low equality reduction, high target deviation), this is arguably the worst set of POLAR data on record. There would be another 57,000 18-year-olds from England (65,000 at the UK level) starting at university this Autumn if young people in lower entry rate neighbourhoods (quintiles 1, 2, and 3) went to university at the same rate as their peers living elsewhere.
Inequality between men and women
In 2018 the UK entry rate for 18-year-old men at day 28 was 27.8%, up just 0.1 percentage points. For women 38.1%, also up by not much (+0.4 percentage points), but still four times more than for men. It seems that the slowing overall entry rate has hit men harder, pushing the university gap between men and women to a new record. Young women are now a startling 37% more likely to enter HE than men. The absolute percentage point gap between men and women has gone past 10 percentage points for the first time at 10.3 percentage points.
These differences equate to 38,000 men not starting at university this autumn compared to what we would see if men had equal entry rates with women. Measured this way, the men/women equality deficit is about 60% of the size of the rich/poor one, but seems to receive a somewhat lower share than that of attention. Of course, ‘men’ are a large population group, and sub-groups within them have some high entry rates. But this is the case with any grouping, and it doesn’t seem to inhibit targets or interventions elsewhere. The UCAS MEM approach is designed to allow for just this high/low subgroup concern. These data show that men make up 74% of the most under-represented fifth of population micro-groups.
With this weight of results, if we are to be guided in access equality by what the data actually says, surely this means having entry differences between men and women in a top three priority list. Yet there remains no explicit HE target to galvanise responses. OfS’ recent access and progression proposals seem set to continue this convention, with men inexplicably omitted from their target list of ‘under-represented’ groups. This apparent inaction in the face of highly compelling data on the scale and direction of this equality deficit is perplexing, and perhaps a factor in why it continues to get worse.
Differences by ethnic groups
There is a range of 18-year-old entry rates and trends across ethnic groups but for some time, the entry rate of the white group (30.8%) has been the lowest of the major ethnic groups that UCAS calculate rates for, and the only one to be under-represented against the national average (32.8%), though the mixed group is close (33.1%).
The 2018 data shows equality between ethnic groups receding further. The entry rate for the white group has fallen (slightly) whilst entry rates for all other ethnic groups are all increasing, by (proportionally) 5% or more for the asian and black ethnic groups.
With entry rates going in opposite directions, the differences between the below-average ethnic group (white) and ethnic groups with above average entry rates (asian, black, mixed and other) is increasing. In 2018 there were around 50,000 fewer 18-year-olds from the white ethnic group starting university compared to if they had the same average entry rate (39.6%) as young people from ethnic groups with entry rates higher than average.
Entry rates by ethnic group are one of the most dynamic areas of equality (and difficult to measure, as National Pupil Dataset based work gives slightly different results). As recently as 2009 the entry rate for 18-year-old in the black and white ethnic groups were almost equal at 26%. Today the black ethnic group has the highest entry rate (42.1%) and the white group the lowest (30.8%). It is possible that policy in this area is being outpaced by these fast-changing figures.
The Government’s old access and participation by ethnic group goal (again, still in operation according to DfE) is to have a 20% increase in numbers of higher education students from the black, asian, mixed and other groups between 2014 and 2020. Whilst increasing numbers of students is clearly a laudable target, this target doesn’t really have much to do with an entry rate perspective on equality. This is because it is expressed in terms of numbers (rather than more suitable rates) and because it only relates to a subset of ethnic groups (those that have entry rates above average). Meanwhile, like sex, ethnic group is a major dimension where equality is worsening year by year: things were more equal in the past than they are now.
Should have done better – time is ticking
The scene was set for a great year for equality. Falling populations in key age groups, no number controls, and (by DfE’s reckoning) £1 billion per year being spent on ‘widening access and successful participation’. But it hasn’t happened. Across the three major dimensions in the UCAS statistics, we’ve seen arguably the weakest year on record.
What has changed in 2018? We don’t know yet. More – probably a lot more – data is needed. It would need to cover interactions of equality characteristics, together with their patterns of qualifications, grades and geography. Universities, often deeply committed to equality in my experience, will already be worrying about their own 2018 figures and targets. A richer national data set to calibrate against would help them. It is probably inevitable that Governments seem to demonstrate the most interest in talking about the national data on university equality when things are going well, and the least interest when they are not. But more consistent attention would be helpful to the sector. As would reassessing whether current policies and targets – which powerfully direct what universities do – still align with where the data says the major equality deficits lie.
The key message in these 2018 figures is the need for a greater sense of urgency. The current low demand / high supply environment is only going to last a few years. After that demand will likely rise rapidly, probably outpacing supply. Large numbers squeezing to get in acts as a strong headwind for equality. If more equal entry is not embedded in universities before the early 2020s then it is quite possible that little progress will be made for a generation.
Let’s just say that each school can send a fixed number of students to each of Oxford, Cambridge, Bristol, … so children from every school get to to go every university as of right.
It seems that the claim made by government and many in the sector that increasing supply would solve the problem of inequality has been proved false. Given the slowing of Q1:Q5 ratios from 2014 when the cap came off student numbers, it seems the primary gains of more places have been made my privileged students. I imagine this would be significantly worse if/when Corver’s team analyse more traditionally (but often no longer) highly selective universities.
We baulked when the Tories initially suggested folks could buy an additional place at a HEI if they had the right grades. Now it seems they are getting in to those institutions through clearing and other methods without any penalty (and often lower than advertised tariff points). Why this isn’t causing more outrage I dont know.
Very interesting article Mark. Perhaps the changes to the BTEC qualifications framework (from QCF to RQF) may be influencing the entry rates of disadvantaged (and male) students. From data I have seen, the likelihood of students achieving high grades such as Distinction or Distinction* under the new RGF framework (with an examination element now embedded) is considerably lower (and I mean considerably) than the old QCF. This means that 2018 applicants under the new framework would have been at a distinct disadvantage in the admissions process to their counterparts (in the same subject) from the old framework. This is important because WP and male students are disproportionately more likely to enter HE via the BTEC route. So if they achieving lower tariffs through the new framework, then it follows that there will be a smaller pool of applicants upon which to draw. If, (and it is still an if at this stage) this BTEC change is influencing the results, it is set to get worse in terms of disadvantaged student access. A majority of 2018 BTEC entrants were still on the old QCF framework (especially the Extended Diploma), but we can expect considerably more to come through the new route in 2019 and beyond. A by-product of these changes too (although this is based on anecdotal evidence rather than hard numbers), is that students at the lower end of the GCSE performance ladder are deciding to take the A-Level route rather than the BTEC route. As they see it, if they have to do exams, they may as well do them in A-Levels. This could have implications for A Level average tariffs too. All speculation on my part (and my argument does fall down a little because BME students disproportionately study the BTEC qualification, but your article shows that the participation gap between BME and white students in increasing, not decreasing), but I think it’s worthy of consideration, as all HEPs have their own access targets to meet. Although perhaps access losses will be offset by student success gains, as we are all aware that, on average, BTEC entrants (under the old framework) are less likely to succeed in HE…….
It’s such a shame that there are few headline indicators of social class left other than POLAR.
I’m not sure you can assess social mobility trends on the basis of a metric that ignores 90 out of 326 local authority district areas containing over a fifth of the population and excludes virtually the whole of London, where only 2% live in POLAR Q1. For comparison, 20% of all secondary school children claiming FSM live in London.
It also gives massively perverse incentives for universities (as documented by e.g. UWE researchers) who are encouraged to completely ignore disadvantaged children who do not live in POLAR Q1 wards in their WP activity and in their attempts to make access fairer at the most selective institutions and instead search out advantaged kids who happen to live in POLAR Q1 instead.
There is undoubtedly indirect racial discrimination involved in the focus on POLAR Q1 given the exclusion of London where – looking at secondary school children – 57% of Black children, 31% of Asian children (including over half of Bangladeshi children) and 29% of mixed race children live in London. Overlay on that the fact that high ethnic minority wards will not be in POLAR Q1 given higher participation rates, that starts to look like a bit of a scandal to me…
Add to that analysis from HEFCE that tells us that over half of young people from POLAR3 Q1 who entered HE and had a known social background came from managerial or professional backgrounds (NS-SEC 1-3) and the urgency of sorting this out and ensuring that OfS develop some far better individual-level measures of social class is apparent (and this needs to be broader than just FSM to capture those from working-class backgrounds – the former is really a measure of worklessness).
A quick bit of work on Excel stacks up that indirect discrimination point:
1. Only 10% of people from ethnic minorities live in POLAR Q1 versus 19% of the white population.
2. Only 8% of people living in POLAR Q1 and 11% in POLAR Q2 are from ethnic minorities compared to 15% Q3, 19% Q4 and 17% Q5.
This is despite the fact that ethnic minorities tend to live in more deprived areas and are also significantly more likely to have a low income than the white population.
I agree with this whole-heartedly. The use of POLAR and FSM are very limited both geographically and sociologically. As you say the POLAR measure has big problems with the geography of HE participation, which underlines the fact that the current measures cannot account for long-term trends. Increasing HE participation will likely mean inequalities between *which* universities parents have attended and what they have studied. POLAR won’t capture any of that detail.
More strictly focussing on neighbourhood measures, why isn’t something like MOSAIC/ACORN/OAC being used? This would at least provide more detailed geodemographic background with much greater data on cultural preferences and economic status along multiple measures. There would still be the issue of the ecological fallacy but it would at least be a much more detailed geographical measure of how cultural, economic and social inequality. Looking at higher education background alone at neighbourhood level is unnecessarily simple to the point of being reductive.
FSM, and income for that matter, are fine if what we are interested in is deprivation alone but that clearly is not the case. You only need to read a small number of qualitative analyses of widening participation and access to know that class inequalities are much more complex in how they play out during admissions and whilst at university. Occupational background is at least a more subtle measure of an individual’s background. I think we can and should have a more detailed analysis of class than the NS-SEC but as a backstop it at least provides a more granular and subtle understanding of inequality than the blunt measures of FSM or income.
“What has changed in 2018? We don’t know yet. More – probably a lot more – data is needed.”
Assuming that the differences are in fact significant, one thing that changed was the A-level examinations and course work. Another change was a rise in unconditional offers (see previous discussion on WONKHE over whether UOs favour students with affluent postcodes. There are, however, many other possible explanations and factors to consider before concluding that any light is flashing whatever colour.
The data does, however, imply that any university achieving an improvement (however small) is potentially bucking the trend.
On Q1 growth and BTECs points:
There have been some very strong increases in English Q1 entry rates post 2012. Up by a third between 2012 and 2017 (arguably the sector was effectively uncapped even before the cap was removed due to the demand/supply position in those years). It is the sudden reduction in the Q1 rate of growth this year that is unusual.
As Mike notes, BTECs have been hugely important to Q1 entry rate increases over this period (accounting for a least a third of the total growth 2012-2015). Any reduction in participation on BTECs at level 3 would hit Q1 entry (proportionally) more than other groups. But no data to know if that is the case yet.
On what POLAR is measuring/missing:
POLAR measures what POLAR measures: differences in HE entry. It isn’t attempting to proxy low income or specific occupational backgrounds – those are much better served by other measures. But in measuring or targeting HE entry rate equality, the POLAR approach of ranking areas by the HE entry rate itself does give a strong identification of high / low entry rate inequality, compared to grouping the same areas by other statistics.
Any single dimension of equality, including POLAR, will conceal variations on other dimensions. Using the MEM approach, which explicitly considers the interactions between dimensions, avoids some of this and so is better (when released). UCAS MEM data from previous years does confirm that the population distribution association between ethnic group and POLAR is what you would expect to see given the pattern of entry rates by ethnic group.
I appreciate that Mark but the issue is that these nuances are completely lost in terms of how POLAR data is increasingly used as the sole measure of success for widening access activity and to target funding and access activity.
For example, the three Office for Students Key Performance Indicators on Participation (Access) are all based on POLAR and the proposal is that POLAR-based measures will form the headline targets for institutions in Access and Participation Plans.
This will cause – and already is causing – access funding from the Office for Students and access activity by institutions to be focused only on POLAR Q1 areas. This systematically excludes working-class ethnic minorities from benefitting from such funding or from counting as “success” by the most selective institutions in terms of their widening participation outcomes – young people from ethnic minorities are half as likely as white British young people to live in POLAR Q1 areas. This is despite very few people from (most?) ethnic minorities accessing the most selective institutions that have the best labour market outcomes. The impact is indirect discrimination which is surely not the policy intent.
The lack of individual-level social mobility data is also leading to all widening participation activity to be targeted on a small sub-set of areas when we know that it is someone’s characteristics rather than where they live that is the key driver of whether they will go to university and, if so, whether they are able to access the most selective institutions which have the best labour market outcomes. Working-class in an affluent area or an area where there are a high proportion of ethnic minorities? Tough luck – there’s not going to be any help for you to access university, at least not from universities or the Government.
Thank you Mark, this report provides some useful insights in terms of groups that require more attention in WP policy and outreach. Also there are some good comments about POLAR and the over reliance and validity in terms of targeting WP cohorts. These are large areas, where populations are not homogeneous. If postcode measures are to be used, they need to be contextualised with other data (e.g. FSM6, ethnicity, gender parental HE background and prior attainment). This will ensure that targeting is not directed at the ‘usual suspects’.
I have published a report on Linkedin that you may be interested in that shows how WP pupils intentions to progress to HE have decreased since the fee increase. This research involved measuring the HE intentions/aspirations of over 14,000 pupils, pre and post Browne review and the implementation of the new tuition fees:
https://www.linkedin.com/pulse/white-elephant-room-have-tuition-fees-decreased-pupils-matthew-horton/