An Educated Population

Of perennial concern to the public, particularly in times of cuts to welfare and services, is the government’s wisdom in spending over £10bn per year on higher education. Advocates might point out that pretty much all affluent countries spend large sums in this area – check out the table below (we’re spoiled with an abundance of columns, but I reckon the third from the right (27) is most informative).

Table A1.3a from OECD report Education at a Glance (2015)

Whether these countries are affluent because they are educated or the other way around (or both) doesn’t really matter right now: the point is they’re prepared to pour between 1% and 2% of their GDP into putting a significant proportion of their youth through university at least once.

So who benefits from this? Potentially everyone. Businesses get to hire skilled workers, which feeds back into creating additional employment opportunities. The economy grows, incomes grow, tax revenues grow. Young people can revel in the development of their passions, and the exploration of new ones. The societal elite can revel in an institutionalised population which is, at least in the UK and USA, so saddled with debt that they’re guaranteed to be working productively, desperately, for decades to come (more on this later).

Unequal Access

If we accept that an individual benefits both personally and professionally from holding a degree, we should also take the (banal) position that access shouldn’t be predicated on some accident of birth – socio-economic class, sex, gender, race. In the grand words of the UN’s International Covenant on Economic, Social and Cultural Rights (Article 13.1),

The Parties … recognize the right of everyone to education. They agree that education shall be directed to the full development of the human personality and the sense of its dignity, and shall strengthen the respect for human rights and fundamental freedoms. They further agree that education shall enable all persons to participate effectively in a free society, promote understanding, tolerance, friendship and peace among all nations and all racial, ethnic or religious groups.

Though many would argue that countries like the UK have made a good deal of progress toward the ideals of social equality and inclusivity, our collective failings remain plain and indicting. This post is motivated by the question, “Does the disenfranchisement of certain racial or socio-economic groups extend to discrimination in higher education?”. Unsurprisingly, the answer is complicated: the academic literature festers with conflicting evidence and shrewdly-spun statistics from all camps. Still, it’s only by trying to understand what the problems are that we can begin to address them. Tally ho.

Racial and Ethnic Minorities

An obvious way to expose potential discrimination against ethnic minorities is simply to look at the composition of the student body. Finding the proportion of minorities at university to be much lower (or indeed much higher) than in the general population would suggest that something is amiss.

On this premise, something is amiss, as some minority groups are acutely under-represented in higher education, both as students and as teachers. This can be partially attributed to exam results – white students in the UK tend to fare better on standardised tests than their Caribbean or Pakistani classmates. That said, there are some encouraging indications that existing gaps are closing, sometimes rapidly (e.g. in the case of Bangladeshi pupils). Indeed, it would appear that whites now score below the national average.

Bias in Admissions

The issue of why our education system might fail some groups more than others is one that I intend to skirt. Let me instead look at the slightly simpler problem of institutional bias within universities – I mean bias predicated not on indicators of merit like exam results, but on racism (presumably unconcious) within the admissions procedure.

What we need is a study which “controls” for exam grades by comparing candidates with similar A-level results across racial groupings. LSE’s 2014 report Black and Minority Ethnic Access to Higher Education provides a recent example. It found, according to co-author Dr M. Shiner,

Even when we take account of A-level grades, candidates’ chances of receiving an offer vary according to their ethnicity, the type of school they attend and their family background. For some candidates these factors combine to create quite marked differences.

This is damning indeed. For instance, the authors calculate that that while over 70% of white university applicants were offered a place, equally qualified Bangladeshi and Pakistani students had an offer rate close to 50%.

Behavioural Factors

Yet any study which relies on population data runs the risk of missing a key factor. The Russell Group (a consortium of the UK’s most prestigious universities) and the Universities and Colleges Admissions Service (UCAS) claim that two important behavioural effects have been overlooked in conventional analyses such as the one quoted above.

The first factor is that some of the alleged bias in offer rates is down to subject choices at secondary school – students from minority backgrounds are slightly more likely to have attained their good marks in less impressive / academic subjects: tourism rather than maths, say.

Perhaps paradoxically, the second factor highlighted by the Russel Group and UCAS is that minority students are more likely to choose oversubscribed and competitive courses, and hence are less likely to be offered a place. According to UCAS,

The expected offer rate to Asian, Black, Mixed and Other applicants is lower than to White applicants. This reflects the fact that (for the same strength of predicted grades) Asian, Black, Mixed and Other applicants are more likely to apply to [more prestigious] universities and courses that have lower offer rates. Most of the difference in offer rates between ethnic groups is accounted for by these different application patterns.

This point is not inconsistent with two other pieces of evidence (stated very crudely I’m afraid): that educational ambition appears to be higher among non-whites; and that non-whites tend to achieve worse degree classifications – according to the Higher Education Funding Council (2014), 72% of white students with BBB A-levels gained a first or upper second, compared with 56% for Asian students and 53% for black students) – since the student who chooses a highly taxing course is more likely to struggle.

Aside: Historical Precedent

It isn’t the first time such an effect has been witnessed. In the early 1970s, the University of California in Berkeley was taken to court on the gounds of discrimination against female applicants for postgraduate courses: admissions data showed that Berkely offered places to 44% of men but only 35% of women. However, when statisticians Bickell, Hammell and O’Connell carefully analysed the data, they came to a diffrent conclusions, echoed today by UCAS. From the abstract of their paper Sex Bias in Graduate Admissions: Data from Berkeley (1975) in the journal Science:

Data on graduate admissions to the University of California, Berkeley, for fall 1973 shows a pattern of bias against female applicants. However, examination of the disaggregated data reveals few university departments show statistically significant departures from expected frequencies of female admissions, and about as many appear to favor women as to favor men. If the data are properly pooled, taking into account the tendency of women to apply to graduate departments that are more difficult to enter, there is a small but statistically significant bias in favor of women. Thus, the apparent bias favoring men stems not from any pattern of discrimination on the part of admissions committees, which seem quite fair on the whole, but apparently from prior screening at earlier levels of the educational system.

Unresolved Discrepancies

Taking all this at face value, we’d conclude that universities themselves are off the hook. If we’re determined to resolve the discrepancies between between ethnic groups, or men and women for that matter, we’re back to grappling with the underlying cultural and historical biases that I took such pains to avoid.

But the story continues to wend its tortuous way, because subsequent analysis suggests that even after we account for all the behavioural effects I’ve noted so far, there still appears to be statistically significant discrimination. Dr V. Boliver from Durham University in the UK is an outspoken researcher of university admissions policy, and has ‘found that offer rates were between seven and 16 percentage points lower for applicants from British ethnic minority backgrounds than for white British applicants, after taking into account their A-level grades and the popularity of the courses applied for’. From Boliver’s 2015 Sociology paper:

Contrary to received opinion, the greater tendency of ethnic minorities to choose highly competitive degree subjects only partially accounts for their lower offer rates from Russell Group universities relative to white applicants with the same grades and ‘facilitating subjects’ at A-level.

I understand this to be a fairly robust result, and I’m going to treat it as the last word on the matter. So let’s take a moment to recap. For me, the main conclusions so far are: (1) Despite recent positive trends in school and university attainment, there remain significant differences between certain ethnic groups, with Caribbean males and Roma of both sexes being in greatest need of support. (2) University admissions statistics (particularly those of the Russel Group) demonstrate favouritism towards white British students. (3) Low ambition among certain groups (including whites) is a crucial factor in their under-representation at university.

Unambiguous Discrimination

Okay, so if we’ve learned nothing else from the foregoing discussion, it’s that studies on population data are inevitably fraught with oversights. Thankfully, in the case of ethnicity-sensitive research, there’s a cleaner (albeit more limited) research method available: you can simply submit identical university or job applications under different names and see which one gets accepted.

A seminal work along these lines was the 2004 paper “Are Emily and Greg More Employable than Lakisha and Jamal?” by M. Bertrand (University of Chicago) and S. Mullainathan (MIT). A more recent study by K. Milkman (University of Pennsylvania), D. Chugh (New York University) and M. Akinola (Columbia) (I didn’t come across any UK studies) focussed specifically on university departments. From the abstract of their 2015 paper

In our experiment, 6500 professors were contacted by fictional prospective students seeking to discuss research opportunities prior to applying to a doctoral program. Names of students were randomly assigned to signal gender and race (White, Black, Hispanic, Indian, Chinese), but messages were otherwise identical. We found that faculty were significantly more responsive to White males than to all other categories of students, particularly in higher-paying disciplines and private institutions.

Their analysis was quite detailed, but here’s an easily digestible graphic from their paper, charting response rates in different disciplines with white males as the baseline for comparison (beware easily-digestible anything – it’s likely to be misleading. Really one should look at their data tables where they cross-correlate response rate with the ethnicity and gender of the professors contacted, and of the departments as a whole, and a whole bunch of other stuff):

Figure 1b from A Field Experiment Exploring How Pay and Representation Differentially Shape Bias on the Pathway Into Organizations (2015). (I'd like to see some confidence intervals to demonstrate statistical significance, but maybe that's too demanding.)

This study, and others like it, support Dr Boliver’s findings quoted above: discrepancies in attitudes towards different ethnicities are significant, even when controlling for other factors. And in they tend to favour whites.

One straightforward intervention, which should have been obvious from the beginning, would be to require all university applications, job applications, submissions to academic journals, legal documents and the like to be anonymised, making it impossible to directly ascertain the person’s ethnicity or gender. Who knows, it might have an appreciable effect.

Socio-economic Background

Barring a priceless moniker like Fauntleroy Barclay-Windsor II Esquire, I think you’ll agree that it’s harder to apply the above methodology to socio-economic class. That’s unfortunate, because the representation of poorer students at university is a highly contentious issue, especially regarding the Russel Group. For instance, just over 7% of British children are privately educated; yet these constitute over 40% of Oxbridge undergraduates. Again, the factors contributing to this imbalance encompass far more than bias of university admissions boards. Indeed, they mirror to a large extent the considerations discussed above.

For instance, state schools generally provide a lower standard of education than private schools, a fact reflected in exam scores, and which can be at least partially attributed to levels of funding (by my calculations, the government spends £7,000 per year per pupil in the state sector, whereas private schools charge on average almost double that).

And, just as in the previous discussion, more sophisticated treatments of the data which attempt to account for exam qualifications still find inherent biases. Some of these biases originate in applicants’ behaviour: according to an article published in 2013 by our friend Dr Boliver,

Applicants from lower class backgrounds and from state schools remain much less likely to apply to Russell Group universities than their comparably qualified counterparts from higher class backgrounds and private schools.

And even those who do apply to the Russel Group are more likely to apply for career-focussed and competitive classes, rather than more rarefied subjects like classics. This is the very same issue that skewed admission statistics for ethnic minorities and women in 1970s California.

Self Confidence

You can speculate for yourselves about the underlying reasons for these disparities. One thing I will explicitly mention is that an important and increasingly acknowledged barrier to ambition and attainment for disadvantaged groups (at university and beyond) is self confidence – a resource more plentiful in private schools and among the wealthy. Lack of confidence reduces the inclination of poorer students to apply for a top university; it reduces the likelihood of them being offered a place (because they perform badly in interviews); and crucially, even if they clear these hurdles and are accepted to the course, the prevailing environment makes it harder for them to thrive socially or academically. It’s no coincidence that economically underprivileged and minority students at Russel Group universities report the lowest levels of degree satisfaction and personal support.

But let’s return to the university administration, and ask a more manageable question: do university and governmental policies add to the (already considerable) disadvantage poorer students face? Well, there is one policy that has come under considerable fire in the press and popular media.

Tuition fees

1998 saw the demise of “free” tertiary education in the UK – in that year Labour (in)famously introduced a £1,000 annual tuition fee. In England the figure has since risen and now stands at £9,000 (Scotland, Wales and Northern Ireland have different arrangements). Costs are typically met through a government-sponsored loan (for domestic and EU students), with repayments indexed to income above £10,000 – the idea being that a graduate only has to pay when they can reasonably afford it.

Nevertheless, it is highly plausible that the prospect of £27,000 debt (plus living costs) coupled with three years’ delay in obtaining meaningful employment will strongly deter those with less financial support from home. If so, then each rise in tuition fees represents a government-initiated (and university-endorsed) blow to the disadvantaged.

This opinion has many adherents – including the UN, whose aforequoted International Covenant on Economic, Social and Cultural Rights (Article 13.1.c) demands of its signatories that

higher education shall be made equally accessible to all, on the basis of capacity, by the progressive introduction of free education

And yet there is an another line of reasoning which arrives at the opposite conclusion: equal access is enhanced by higher fees, as universities are required to put a portion of the money towards extending capacity for bursaries, outreach, and student support. Recent studies suggest that not only did the rise in tuition fees have little impact on disadvantaged students’ application numbers (especially after a year or two), but also that it is far more effective to target spending on primary and secondary education, since that’s where divergences in ability and ambition first set in.

Moreover, if it’s true that university students are typically from wealthier backgrounds, then it makes little sense for the taxpayer (who is not) to pay for their education through increased taxation and/or cuts to public services. The realities of national budgets have in some cases caused both the quality of education and the representation of disadvantaged students to drop after the introduction of free tuition. Indeed, LSE’s Professor Barr praises the UK’s education policy for being among the more effectivly progressive.

But if you thought I was going to leave the issue resolved on this perfect swell of consistency, you clearly haven’t grasped how this blog works. The last argument might have seemed compelling, but we have to remember that the first argument (of tuition fees’ deterrent effects) was also pretty compelling. Taking a closer look at the data, Professor W. Hutton (of Oxford University, no less) found many important things. Here are two of them: 1. the reported increase in less privileged student numbers masks enormous losses to part-time courses, and to applications to the Russel Group; and 2. the long-lived effects of debt are eating away at national finances to such an extent that it will be a principal contributor to the next credit crisis. This line from Hutton’s article is particularly chlling

Britain is in the process of creating the most stratified, least socially mobile, cruelly unfair society in its treatment of the young in the advanced world.

(though it only just beats the line about us doing worse than the USA). Perhaps you don’t buy every part of Hutton’s argument; but we must not ignore its core message, lest we jeopardise the prospects and prosperity of a generation.

Final Remark and Part Two

Though it was undeniably a morass of divided opinion and contradictory evidence, I hope this post has illuminated something of the ongoing conversation about representation at university. While researching it, I was at times pleasantly surprised. At other times I was not. But even in the preponderant times of confusion I at least gained some perspective on what to be confused about.

Of course there’s so much more to learn about this subject, and I haven’t even touched on the fact that the education system seems to be failing boys at a significantly higher rate than girls, from secondary school onward (whether this is cultural or institutional I don’t know for sure, though I’d be surprised if it wasn’t both). I simply have to move on or I’ll go mad.

“But this was only part one!” I hear you cry. Right. In part two, I intend to talk about the pressures on universities to measure cleverness, and the ethical gauntlet we run when we try to use discrimination as a tool to redress social iniquities. All in good time.


Joy 10 February 2016, 12.19

Like you said, besides racial discrimination there is also something really amiss in terms of diversifying according to socio-economic background. For instance, the undergraduates at Oxford are, as a group, a lot whiter than the postgraduates; but although the postgraduates are more diverse ethnically, it seems that there is very little diversity in terms of socio-economic background. These are just anecdotes, but for instance, a girl from India told me that all Indians at Oxford she knows are from the 0.1% richest group in the country. And a friend of mine who’s a Rhodes scholar said that basically everyone who has a Rhodes scholarship is from the most privileged groups in their own countries, too (which usually means they could afford to be at Oxford even without the scholarship). I know it’s a self-perpetuating thing, that privileged people with better access to education will have better chances of getting into Oxford etc. But because the postgraduate community „looks diverse“, and because, on paper, it is very international, these issues are swept under the carpet and addressed much less than the problem that privately educated students are over-represented at the undergraduate level.