Transparency revolution: Is there bias in university admissions? Part II

Image: IKON

As we noted last week, full analysis of UCAS’s new release of institutional admissions data broken down by socio-economic background, ethnicity and sex is be a vast undertaking. Over 200,000 individual lines of data were released by UCAS, and it will take months of work by social scientists to get a full picture of what it tells us.

Having had some more time and space to take another look at the spreadsheets, there are even more intriguing findings to discuss. Responses to our findings have also suggested a range of hypotheses on what might be the cause of ‘gaps’ in offer rates for different ethnicities and how they might be explained (or in some cases, explained away). Getting to the bottom of this will take considerable time, effort and diligence in universities up and down the country: we are shining light on a pencilled sketch, not a finished painting.

Asian students also have offer-gaps

From our analysis of the 2013, 2014 and 2015 datasets, the following institutions have a significant gap in offer-rate between Asian applicants and the average application rate in two of the last three years:

University of Aberdeen
Aston University
University of Birmingham
Birmingham City University
University of Chester
University of Dundee
University of Derby
Edinburgh Napier University
Edge Hill University
University of Huddersfield
King’s College London
Kingston University
University of Leeds
Leeds Beckett University
Leeds Trinity University
University of Liverpool
University of Manchester
Manchester Metropolitan University
Nottingham Trent University
Oxford University
University of Reading
University of Sheffield
Sheffield Hallam University
University of Strathclyde
Teesside University

Similar to our previous analysis of offer-rates for black students, this is quite a considerable list – nearly one-fifth of UK HE providers. ‘Asian’ encompasses a broad range of ethnicities with very different levels of participation and success in higher education, including Chinese, Indian, Pakistani and Bangladeshi applicants. The list is also diverse – no single mission group or level of tariff dominates. Some of the institutions accept the largest number of Asian students of any university, including Aston, Manchester Metropolitan, Manchester and Birmingham. As with our list of offer-gaps for black students, it appears that despite this, Asian students are still not quite receiving the number of offers that might be expected even with the very high level number of applicants and acceptances.

White students’ offer advantages?

From our analysis of the 2013, 2014 and 2015 datasets, the following institutions have a significant positive gap in offer-rate between the White applicants and the average application rate in two of the last three years:

Anglia Ruskin University
Aston University
University of Bedfordshire
University of Birmingham
Birmingham City University
Brunel University London
Coventry University
Imperial College London
University of Kent
King’s College London
Kingston University
Leeds Beckett University
University of Leicester
London South Bank University
Manchester Metropolitan University
Nottingham Trent University
Oxford University

Given the number of institutions that have a negative gap in offers towards black and Asian applicants, it is little surprise that a number of institutions show a mirror image advantage towards white applicants. Most of these institutions appear in either the above list or our lists from our previous analysis. In these institutions, white students appear to be more likely to get an offer than non-white students with similar grades at A level or equivalent.

Mind the gap

A considerable degree of caution must be given before the above lists are used to directly accuse the above institutions of systematic racial bias, but I believe a healthy degree of scepticism also needs to be levelled at some of the excuses given for the above gaps in offer rates.

The UCAS statistical controls for comparing a group of applicants’ offer rate with their ‘expected’ offer rate only accounts for a small number of factors that admissions’ departments take into account. As UCAS put it, “a difference simply means that the offer rate is higher or lower than it is for all applicants who are similar in terms of the subject applied for and a summary measure of their predicted grades”. Many have been hasty to point out that there are other factors that influence students’ chances of success: personal statements, interviews, aptitude tests (in subjects like medicine and law), and portfolios of work (in the creative arts).

These explanations are purely speculative and more qualitative and quantitative investigation is required to confirm them. Furthermore, if black and Asian students are indeed receiving fewer offers due to less success with personal statements and interviews, the sector should be asking itself some hard questions about how fair and ‘race-neutral’ these forms of application are. If success in the personal statement is to some extent reliant on a form of cultural capital that gives white students an advantage, this should be accepted as an indicator of bias. Similarly, if a deficit of cultural and economic capital to produce portfolio’s for creative arts applicants is the cause of offer rates at specialist institutions, this should be given a priority focus in outreach and widening access work.

A more sympathetic explanation for these offer-gaps is that the UCAS controlling method is unable to account for the combination of subjects studied at A level or equivalent. I am not currently aware if there is data publicly available on subject combinations broken down by ethnicity or social class. This would be an interesting future release for UCAS to get working on. Are black, Asian and POLAR 1 students simply less likely to choose “facilitating subjects” (defined in their broadest sense, not just the Russell Group’s definition) than white students?

One further unfortunate limitation of the data release is the inability to do analysis by multiple measures of disadvantage. We can look at POLAR 1 students or black students in isolation, but not combine the two to look at black POLAR 1 students. To get into this level of analysis will require social scientists to do extensive work with the individual-level data that UCAS is planning to release to the ESRC in due course.

Moving forward

It is important to emphasise that getting to the bottom of the issues highlighted by our analysis will require institutional level – and in some cases subject level – solutions. Universities take vastly diverse approaches to managing their admissions, and there are divergent approaches within many universities who have kept to process relatively decentralised.

It can only be hoped that the release of this data is indeed a catalyst for focused, sensible, evidence-based change. There is significant anxiety within the sector that the way this data can be spun or misunderstood, particularly in the national press, will only harm efforts to widen access. This is not a sufficient argument for not making more of these releases available. There is substantial capacity with the HE sector, the research community, think tanks and the government to come to solid and useful conclusions that can instigate change. In Offa, the sector also has an organisation with the clout and expertise to improve policy and practice. The HE sector is capable of rising to these challenges; it now needs to get to it.

Find the first part of this analysis here.

This article was amended on July 10th to remove Loughborough University form the list of institutions with significant positive offer gap for white students. This was due to a rounding error in the initial analysis. Loughborough falls just out of the bounds of statistical significance on this measure.

1 thoughts on “Transparency revolution: Is there bias in university admissions? Part II”

  1. DC says:

    Interesting stuff. I’d also be very interested to know if there’s any noticeable bias regards ‘socio-economic background’ across all ethnicities. Thanks

Leave a Reply

Comments 1 See all

  1. DC View