Post-qualification offers would improve fairness in UK admissions

By many measures, the system of admissions to UK universities is very good. Most applicants are able to attend their university of choice. The numbers of those failing to complete their degrees are low by international standards. And there are excellent employment rates for graduates.

Yet we still need to ask whether the system is truly fair. Is it fair to the applicants themselves? To the universities seeking to recruit students who will benefit from their courses? To employers looking for graduates with the right set of knowledge, skills and behaviours? And indeed to society as a whole, which invests considerable amounts of money in higher education and wants a reasonable return, both in terms of tax revenue and wider non-financial benefits?

There are obvious reasons why this matters. The graduate premium in lifetime earnings amounts to about £170,000 for men and £250,000 for women and has remained relatively constant over the past 20 years in real terms. There is therefore a genuine economic advantage to getting into university. Graduates also benefit in other ways, since they tend to drink less, exercise more, suffer less depression and live longer.

Perhaps unsurprisingly, therefore, there are still many more applications to universities than acceptances, despite significant increase in capacity over the years. In 2020, about 570,000 of around 730,000 applicants were accepted. Furthermore, there are a number of universities for which application numbers greatly exceed their capacity. Their courses therefore represent a limited resource in high demand, so finding a fair way of selecting students for admission is essential. It is surely a matter of natural justice to create a system that is seen to be as fair as possible to the greatest number of applicants.

So what do we mean by “fair” – and how can we create a system of admissions that would be considered fair by most people? Here I want to offer some general thoughts, drawing on Isabel Nisbet and Stuart Shaw’s 2020 book Is Assessment Fair? before going on to more concrete suggestions.

In considering assessment, Nisbet and Shaw argue that a fair process should be valid and reliable. It should meet legitimate expectations. It should be impartial, and it should consider the context of the candidate (for instance, previous educational experience). I will examine each of these criteria in turn.

Looking at the validity of applications processes, it is first important to decide what universities are selecting candidates for. Is success measured just by completing the course or by gaining a “good” degree (a 2:1 or a first)? Are we selecting for students who might become the most accomplished scientists, mathematicians or historians, or the most likely to progress to postgraduate study? Or are we looking at future professional achievement, those who might become the best teachers, engineers or doctors?

Once we have decided on our goals, we also need to look at the validity of different methods. Are A levels and other terminal summative examinations as valid as qualifications that consist of assignments and continual assessment, such as BTECs? Would it be more appropriate to have standardised admissions tests, such as the SATs (Scholastic Aptitude Tests) widely used in US college admissions? The International Baccalaureate (IB) also has benefits in breadth of study and mixed modes of assessment, as well as more international recognition.

Certain assessment methods are commonly used to test aptitude for particular vocations. Familiar examples include portfolios of material to determine artistic merit, presentations in the performing arts and interviews to assess empathy and other appropriate characteristics in the caring professions. Interviews more generally raise a number of issues. Their reliability can be enhanced by embracing a standard structure, although this can suppress the spontaneity that often differentiates candidates. They are also criticised for selecting people like the interviewers rather than those with different but equally useful attributes. (They are therefore unlikely to enhance diversity, unless some sort of affirmative action is taken in the selection of interviewers.) Another problem is that calling applicants to interviews can put up barriers to those from lower-income households, who may not be able to afford the transport costs or have the technology required for a virtual interview.

Similar factors also apply to personal statements, since more disadvantaged applicants are likely to receive less support from tutors, career advisers and parents, and to lack the social capital conferred by visits to potential workplaces and supporting extracurricular activities.

The next criterion, reliability, can be broken down into precision and repeatability.

So how precise is the method of assessment? Is a candidate who scored 65 per cent distinctly better than one who scored 64 per cent, or does the imprecision of the method mean that a grade boundary of, say, 5 per cent (between 62.5 and 67.5) is a more appropriate level of granularity for selection?

It is also important that repeated assessments give the same results. Their fairness is likely to be questioned if a single piece of work gets very different marks on different occasions or when considered by different assessors. Some universities address this through double-marking and accepting an average mark – and even, when necessary, by bringing in a third marker. (It is also possible to derive some confidence in the reliability of a test if a statistically valid number of cases are double-marked, and any differences are found to be acceptably low.)

We must now look at one of the distinctive features of UK higher education. More than 97 per cent of 18-year-olds who apply to university do so before their assessment results are known – on the basis of their teachers’ predictions of what grades they will get. Yet the accuracy of such predictions is low, with only about 15 per cent to 25 per cent of predicted and actual grades matching exactly. Teachers are actually highly consistent in making predictions – it is just that they consistently over-predict by about 2.1 points (where an A* is 6 points, an A is 5 points, a B is 4 points and so on). This is hardly surprising, given that judgement of their teaching quality may depend upon examination outcomes!

There is a view, particularly in the US, that black students are hampered by the low expectations of their (white) teachers. It is also claimed that basing offers on predicted grades could further harm already disadvantaged students, presumably on the assumption that those in advantaged areas or schools are likely to be given relatively higher predictions. Yet university admissions systems have become highly capable of “compensating” for the more common forms of over-prediction. And the data do not indicate that utilising predicted grades harms equality generally. Indeed, for most disadvantaged groups, it has a positive impact.

So it is likely that the use of predictions is less unfair than might be imagined. Nevertheless, it is self-evident that the use of actual results is more accurate, and provides a more transparent platform for the legitimate expectations of applicants and, more broadly, society. 

Legitimate expectation represents the second pillar of fairness set out by Nisbet and Shaw.

The legitimate expectation of applicants to university has changed according to government policy and the expansion of higher education. It is also influenced by published information. If a university sets out entry requirements for a particular course as AAB and an applicant achieves AAB, then they will have a legitimate expectation of recruitment. Yet such information has also become important from an institutional perspective, since entry grades offer a proxy for quality – with many universities inflating their advertised minimum entry requirements to attract a larger number of applicants, but then accepting students with grades below those advertised. There is a strong argument, from a fairness and transparency perspective, that institutions should publish the actual grades at which they recruit, accepting that these may have to be retrospective.

There are of course legitimate expectations, from employers and society, that graduates have acquired the knowledge, skills and behaviours essential to their jobs. So we could perhaps legitimately expect higher entry grades for more academically demanding careers, such as medicine. Yet it is important to balance this against the unnecessary grade expectations of some professional, statutory and regulatory bodies, which can be perceived as creating “closed shops” to encourage demand-driven salary premiums for those already “in post”.

The third pillar of fairness relates to impartiality – or ensuring that similar cases are treated alike.

UK universities now pride themselves on their impartiality, although this has not always been the case, either in the UK or elsewhere. In the US, some universities have historically embraced explicit donor preference schemes, whereby the children of large donors (as well as children of alumni) are given preferential access. For privately funded institutions this may be acceptable, if unpalatable. Where higher education is supported by the state, as in the UK, it is considered inequitable to give advantage based on anything beyond the merit of the candidate. This does not mean that applicants will be treated in exactly the same way by every university. Each institution has its own character, prestige and capacity. The important thing is that the universities of Oxford and Hertfordshire will each treat all applicants to their universities in the same way.

Concerns are often raised about the relatively low number of recruits to Oxford and Cambridge, as well as to other pre-1992 universities, from ethnically diverse backgrounds and poorly performing state schools. So might impartiality be enhanced if applications were anonymised and information such as their school or college removed?

There are mixed views on the benefit of anonymised recruitment, reflected in the mixed outcomes of various specific efforts to improve gender balance in particular professions. Gender balance at the Boston Symphony Orchestra was improved when auditions were held behind a screen. However, ascribing a woman’s name to a name-blind CV increased the chance of the applicant making it to interview in the Australian civil service by 2.9 per cent. Anonymisation could support impartiality if it is combined with overt contextualisation. Otherwise, it could have the opposite effect, since it is likely that contextualisation on the basis of race or school is often already implicitly taken into account.

The fourth major pillar of fairness relates to context. Where university admissions are concerned, this refers to previous educational experience. It is all very well to be completely impartial on the basis of A-level results, but that is not fair to a student who carries a legacy of prior educational disadvantage. In this regard, it is useful to consider issues of equality, deservedness and merit.

From an equality perspective, a compelling argument can be made to compensate individuals who have been historically disadvantaged, since the qualifications that they bring to the admissions process reflect their previous educational (and, more broadly, social) experience, and not their future potential. It is known that a male student from a state school is 6.5 per cent more likely to get a good degree (2:1 or first) than a student from an independent school who had equivalent A-level grades at entry to university.

Research indicates that the effect of social background on attainment begins at the age of two – and a child from the most deprived cohort is sometimes said to be likely to have heard 30 million fewer words than one from the most advantaged by the age of three. The question then arises how best to determine the level of disadvantage to the individual and how much to compensate for this.

There is no right answer and, in an ideal world, each applicant would be considered as an individual, with a battery of metrics ranging from parental income, family circumstances, location of home and historical achievement of their school, among many others. Ideally, a similar range of metrics would be embraced by all universities, such that there was equal compensation. This is well beyond the current admissions systems in subtlety and complexity.

Nevertheless, in Scotland, universities have embraced a robust and workable system, using the Scottish Index of Multiple Deprivation (SIMD). This is based on seven domains: income, health, employment, education, housing, crime and environment. Universities use the lowest one or two quintiles of deprivation as priority cohorts for contextualisation. Each university sets a minimum entry requirement for every programme, which should allow successful recruits to complete their course. Applicants from these quintiles who achieve the minimum requirement are then accepted on to the course outside the general competitive entry process.

Is it possible to use measures of deservedness to assist in creating a fair admissions system?

One of the tragedies of life is that diligent, hard-working individuals, who deserve success, are often usurped by opportunistic chancers who have the inherent ability to succeed with less effort. For a hard-working individual of modest natural ability, a university admissions system based on a terminal summative assessment process, such as A levels, could appear unfair. It could also be argued that, from an employer’s perspective, someone who is diligent and hard-working is often a greater asset than their lazy but more able counterpart.

In most US schools and universities and some UK universities, the traditional UK degree categories have been replaced or supplemented by a grade point average (GPA). This gives an outcome based on assessment throughout the course, encourages steady effort and offers a future employer a measure of both effort and ability. At the point of entry to university, it is also likely that the BTEC and IB systems, which include assignments and coursework, support the more diligent applicants. It is interesting to note that graduates who gained admission to university with a BTEC have better employability outcomes than graduates who gained admissions through A levels.

Source: 

Gary Waters/Getty/iStock montage

Finally, there is the question of merit: is someone worthy of admission to a university?

Whether assessed over a period of time or by summative examination, it is likely that candidates who are both diligent and able will score well and will be considered worthy of admission. The idea of a meritocratic society, in which those who are worthy succeed, is rather more universally accepted now than it was when first described by Alan Fox in a 1956 article in Socialist Commentary as “a society in which the gifted, the smart, the energetic, the ambitious and the ruthless are carefully sifted towards their desired positions of dominance”. Nisbet and Shaw devote considerable effort to debating the value of merit from the perspective of fairness in assessment but eventually reject it on the basis that it does not account for social context and is likely to be greatly influenced by the merit of an individual’s parents.

For university admissions, however, merit potentially brings together many of the elements of fairness already discussed. It conforms to many principles of natural justice. It can be used to discriminate within a competitive process or to determine adequacy for a particular course. It is fair, in the sense of impartial, to the individual, the provider, employers and society. Used together with contextualisation (or compensation) for previous educational disadvantage to offer a form of restorative justice, merit emerges as the most appropriate measure for admission to university. It is also likely that a system of admission based on merit will select students able to succeed in their courses. And the validity and reliability of any measure of merit can be tested by retrospective analysis of success rates at university.

It remains very unfortunate, therefore, that limited government enthusiasm for radically changing school or college assessment methods, not to mention the fiercely autonomous institutional character of UK universities, makes it unlikely that standardised measures of merit will be created or accepted.

How might the basic elements of fair admissions be translated into a practically workable system?

First of all, it would be fairer if everyone were subjected to the same entrance process – at least for particular courses or groups of courses. There is also a strong argument for standardised testing, although sometimes using the specific tests (such as the BioMedical Admissions Test) relevant for particular courses. A general test for all university applicants might pick up on universally desirable attributes, such as critical thinking, but would potentially fail the validity test. (Could it be equally valid at determining aptitude for chemistry and for modern languages?)

What about common interviews? It seems unnecessary and expensive for a candidate applying to five different nursing schools to have to do five separate interviews. On the other hand, universities are very diverse institutions, so an interview for a veterinary place at Glasgow, with strength in cattle medicine and pathological research, might with good reason be different from an interview for the Royal Veterinary College in London, with strength in small animal medicine and surgery. There is also a strong argument that the interview process gives the candidate a chance to assess the institution and determine if it would fit with their character, values and expectations.

I have already considered the issue of predicted grades and argued that actual results would be more appropriate. This, however, presents a major practical challenge. The results of A levels (and other level 3 assessments) are made available to candidates and universities in the middle of August. The time required to complete the recruitment process – for students to get their results, apply to up to five universities, perhaps undergo an interview or several interviews, receive and decide between any offers – would make it virtually impossible to start a course at the beginning of October.

Although it would theoretically be possible to change the school or university timetable, neither sector seems enthusiastic about doing this. It has been proposed that the university start date be moved to the January following the A-level results day, but this would mean candidates spending about six months in limbo – a time too short to get useful employment or have a worthwhile gap year, but ideal for allowing 18-year-olds to engage in antisocial behaviour or sink into idle despair.

A more workable proposal was made by the Universities UK group that reviewed fair access to universities. According to this, candidates would put in their applications any time from the January before the September/October start date. When their assessment results are released, universities would make offers to suitable candidates, who then choose which to accept. (There would also be a clearing system, much as now, to pick up candidates whose results fall below expectations and do not receive an offer from any of their chosen universities.)

The whole process has been modelled by the Universities and Colleges Admissions Service (Ucas) and would allow time for everything to be completed by the current university start dates. One concern is that students might not receive appropriate information and guidance in the “teacher light” period at the end of the summer holidays. This could be overcome by the Department for Education providing high-quality careers advice at that time of year. Furthermore, applicants should already have been receiving advice from their school career advisers for the six months prior to results day.

In the current “predicted results” system, more than 70 per cent of students are accepted by their first-choice university. There is no reason to anticipate that this would change negatively if actual results were used.

From a standpoint of fairness, a post-qualification offer (PQO) system is the compelling option. The idea of offers being made on actual rather than predicted grades is not a new one. It was recommended for development by Steven Schwartz and his colleagues when they reviewed fair admissions in 2004 (and is, of course, common in many other countries). That it has not happened is down to self-interest and inflexibility on the part of the actors involved and a lack of commitment or courage by successive education secretaries.

The second radical reform required to enhance fairness in the admissions process is a more reliable and uniformly applied system of contextualisation, to compensate applicants for previous educational disadvantage.

The Scottish system described earlier is sufficiently pragmatic to be widely adopted, yet sufficiently flexible to give confidence that institutional autonomy is not compromised. A system of contextualisation utilising the Index of Multiple Deprivation (IMD), which ranks population areas of about 1,500 inhabitants, should be adopted in England, too, and should embrace at least the lowest two deciles of disadvantaged applicants.

One option would be to allow each higher education provider to select its own minimum entry grade at which applicants in the lower IMD deciles would be accepted. Alternatively, a universally agreed tariff uplift could be applied to such applicants – perhaps graded across several of the lower deciles. The IMD is sufficiently granular to give a reasonable picture of likely deprivation. Other measures, such as whether applicants received free school meals (or whether they were in care during their school education), are also useful, but the data on these are not universally available, and subject to data protection legislation.

It is unlikely that an admissions system could ever be created that would appear fair to all applicants, parents, universities, future employers and society. The system we have in the UK is by many measures good, but could be improved from the perspective of reliability, legitimate expectation and restorative justice.

A post-qualification offers system, with more uniform contextualised adjustments for previous educational disadvantage, would be fairer, more transparent and more just. We owe it to future generations to make it work.

Quintin McKellar is vice-chancellor of the University of Hertfordshire. He is grateful to Aoife Keenan (senior policy officer) and Sally Mapstone (principal) at the University of St Andrews for help in understanding the Scottish system.