University of Cape Town

Proposal for modifying UCT's undergraduate admissions policy

Executive summary

The goal of the admissions policy is - and remains - to transform the student body into one that is more diverse and representative of the population while recruiting the best students available. Twenty years on from 1994 some progress has been made but transformation is incomplete. The proposed modified policy is a recalibration, with the same goals in mind, taking account of the changing realities of race and class in South Africa since 1994.

In brief, whereas in the past, almost all black applicants to UCT were uncompetitive in terms of their school leaving results because overwhelmingly they came from poor schools and disadvantaged backgrounds, now many come from good schools and are admitted on a competitive basis in terms of school leaving results. Others may be less competitive on this basis because there is still educational disadvantage through their school or home backgrounds - but the playing fields can be levelled by taking these backgrounds into account - without reference to race.

However, while these selection procedures can successfully recruit many black students, there are still so many more strong white applicants than there are strong black applicants that we would not achieve adequate transformation of the class without a further explicit racially-based selection mechanism.

Thus the proposed new admissions policy is a hybrid procedure using three mechanisms for selection: one part of the class selected just on marks; a second component selected based on their performance and ability which takes account of their school and home background; and a third component which is driven by achieving demographic targets based on an applicant's race and performance. (Note no students would be offered a place who does not exceed a minimum threshold that predicts a high probability of success.)

Given the current profile of our applicants and the competitiveness of UCT admissions, depending on the faculty, on average 75% of students could be selected without reference to race, while 25% would be race based. The proposed hybrid procedure will still enable UCT to increase the number of both black and disadvantaged students.

We believe this proposed change to our policy is consistent with our long term commitment to a non-racial society, endorsed by the constitution. In making this commitment we do not seek to take away people's right to racial identity: people should be free to identify themselves however they wish. But we are committed to the development of a society which does not distribute resources and opportunities based on one's membership of or classification into race groups.

Outline

To make sense of the proposed changes to UCT's admissions policy, it is necessary to explain briefly the current policy, and comment on its strengths as well as the factors that have prompted the review of the policy and proposed changes. The proposed hybrid policy is explained, and finally some examples of the modelling of admissions outcomes using the proposed new policy are given.

Current admissions policy

For over two decades, UCT has had in place admissions policies that explicitly attempt to increase the numbers of African, Coloured, Indian and Chinese students (hereinafter referred to as black)[1]. This has been necessary because for many courses where selection is competitive, if we were to rank applicants simply using their National Senior Certificate (NSC) marks, there would have been very few African students selected. For example, in the case of the medical class of 200 first years, there would have been about 10 African students. There are several reasons why this is of concern to us and to society more generally. Here are the primary ones.

We want to promote fairness and social justice. It is clear that most of the variation between students' NSC marks is much more directly related to the schools they went to and other socio-economic factors, than to their ability or motivation or hard work. So clearly we cannot only use marks because it is unfair to those students from less privileged backgrounds in further denying them crucial life opportunities.

We are concerned with social justice as it affects both the individual and the community from which he or she comes. We, therefore, need to have an approach which will acknowledge those circumstances which may impede opportunities for the individual and those which operate at a broader social level and which are responsible for either discriminating against a group of people or advantaging them. Our redress policies must, therefore, be sensitive to both the individual and group experience.

We want to attract the most talented students. Because many such students will have gone to poor schools and have lower marks they would not be selected if the selection process only considers marks. Yet we know that, given the opportunity, they will become top performers.

We are concerned to produce a new generation of professionals, leaders, intellectuals, political actors and analysts who are more demographically representative of the population. It would be very dangerous for society if, when the medical or law class graduates in 2019, 25 years after the first democratic elections, a small minority were African. It is also an important change strategy that there should be role models for students from historically excluded groups to emulate.

We think that the education of all students benefits from having diversity within the classroom, since the different histories, cultures, languages, ways of seeing and being contribute both to the analysis and understanding of subject material, and to developing the social competencies of students.

The current admissions policy has two elements. The first is the setting of targets for each population group in each programme. Targets are adjusted each year based on three considerations: (i) our aspiration increasingly to reflect the demography of the region and country; (ii) our determination of the minimum threshold mark for each programme below which, experience tells us, students are unlikely to succeed (we do not accept students below that threshold); and (iii) the likely pool of applicants with good enough results in the appropriate subjects for the particular programme. The new proposals do not make any change to this component of the policy.

The second component of the admissions policy describes the process for selecting students and rationing scarce places to meet the targets. The principle underlying the selection mechanism has been that talent is randomly distributed across population groups, and that the top few per cent of African applicants are likely to be as talented and motivated as the top few per cent of white students even though they may have different marks.

The current policy places black South African students in separate baskets according to their legacy apartheid population group (race)[2], where they have declared this, and places all undeclared applicants, all white applicants and all international applicants in an ‘open category basket, ranks them academically within their own group, and selects the top few per cent from each list depending on how many places we are targeting for that group.

This achieves the goals of selecting the top students and also ensures that the playing fields are more level, i.e. that applicants are not excluded simply because of the legacy of apartheid schooling and other disadvantages. This also addresses the diversity and redress goals. This explains why in most programmes, the cut-off mark for white students is higher than the cut-off marks for African and Coloured students.[3]

They are in different baskets and not competing against each other, and the concentration of white students in good schools means that marks in the white basket will be much higher than within the black baskets. This difference is even greater if the competition for places in the white basket is high i.e. the number of eligible applicants greatly exceeds the number of places for that group.

The system has served us very well for over a decade. It has achieved remarkable diversity as anyone attending our graduation ceremonies will attest, with a majority of students now being black (i.e. African, Coloured or Indian).

So why change what appears to be working? And in particular, why change if we have not yet reached our targets for African and Coloured students? Because circumstances have changed, and because we can do better.

Why modify the admissions policy?

The motivation for adapting the policy has three drivers: first, an ideological commitment to non-racialism in the long term; second, a change in the educational preparation of many black applicants to UCT resulting in them no longer being significantly disadvantaged and no longer needing affirmative action interventions; third, a recognition that as the profile of applicants has become more middle-class, the current race-basket approach fails in the selection of socio-economically and educationally disadvantaged students. There is also an additional, technical factor, regarding the problems of doing race classification.

Commitment to non-racialism

Apartheid racial constructs were used to distribute power, to create divisions in society, to signal superiority and inferiority and to promote ethnic loyalties. One of the main goals of the post-1994 South Africa is to transform the society into one which does not privilege people or deny them opportunities on the basis of race. This is how we understand the constitutional commitment to non-racialism.

There is, however, general support for the view that the path to that goal requires an interim period of redress, of conscious structuring of opportunities to undo apartheid's legacy of racial inequality. What is contested, and thrown into sharp relief when considering admissions policies, is whether redress policies should be structured on the apartheid system of racial classification or on the intermediate determinants of inequality, i.e. on how racism and racial oppression work to create and perpetuate inequality.  

One school of thought argues that since redress is about countering the effects of discrimination against people classified as African, Coloured, Indian or Chinese, the interventions should simply focus on those who were, or would have been so classified under apartheid.

The other school argues that one must examine how race discrimination operated under apartheid and the ongoing effects of race-structured inequality, and target affirmative interventions on the basis of those factors directly (for example to those who have been denied access to good schools or adequate income).

This lobby argues that any form of racially based preference firstly requires a system of race classification which is both legally and morally problematic, and secondly, entrenches a view of the world which links entitlements and access to resources only to one's colour regardless of one's actual degree of privilege or status in society.

Thus, if we can achieve the racial diversity we aspire to for UCT, while moving away from a dependence on using race classification to do so, we believe this would be a positive contribution towards non-racialism. And it seems we can move in this direction because of the progressive realignment of race and class in SA over the past two decades.

Black applicants have had better schooling

Historically, there has been a very close correlation between the categories of race and disadvantage. Most African and Coloured applicants were poor, most of their parents poorly educated, and most attend township or rural schools. Race has been an excellent proxy for disadvantage.

Furthermore, even for students who are middle class and have gone to good schools, educational performance is heavily affected by one's cultural capital - a determinant of performance that crosses generations and therefore should be expected to reflect the circumstances of the parents and grandparents of our current applicants. It is the race classification of the parents, whose life opportunities were structured by apartheid, which significantly affects the degree of educational disadvantage of the children.

Over the last few years, our research shows that the circumstances of black applicants have changed in various ways. Increasing numbers of our black applicants are coming out of excellent schools with very good NSC results, often from wealthy families. They can get into UCT in the general open competitive pool.

They do not have to be in a different basket competing only with others of the same race.[4] While there remain differences in cultural capital which may affect performance, these can be compensated for through weighting the marks using direct measures of parental disadvantage. We can achieve substantial diversity without needing to select students on the basis of race.

Not only is affirmative action unnecessary for these black students, but there is a legitimate question about whether well-performing black students at previously white schools remain disadvantaged and if so to what extent, and whether it is fair to white classmates from the same schools, that the black students should get in at the expense of white students who may even be less privileged.[5]

Realignment of Race and Class - the need to recruit disadvantaged students

While we affirm the need for racial diversity at UCT, we have always emphasised the need to take socio-economic disadvantage into account. Over the last 20 years, the old colonial and apartheid correspondence of race and class has been shifting. About a half of black students at UCT are now middle class. No longer can we assume that all black students are economically disadvantaged.

We aspire to greater socio-economic diversity - in the interests of fairness, and equal opportunities. We would be doing better if we could recruit the most talented poor black and white students whose marks, we know, will not be competitive with students from good schools.

Some difficulties with Race Classification

A relatively minor concern is the current dependence on students self-identifying their ‘race'. This leads to a variety of problems. Some students, including disadvantaged students of colour, on principle do not want to declare their ‘race', which they disavow. This leads to them not benefiting from the redress policy even though they may still be suffering the legacy of educational disadvantage.

Other students wilfully misclassify themselves in relation to the old categories -particularly Whites and Indians claiming to be Coloured. Since there is no legislated way of classifying people, this puts UCT admissions officers in the untenable position of having to decide how such applicants should really be classified. This we refuse to do. There are also significant numbers of applicants who may legitimately classify themselves as black but should not be beneficiaries of redress policies at the expense of others (black or white).[6]

In summary, UCT's commitment to non-racialism is not a denial of race or that race matters. It is a commitment to becoming an institution and promoting a society which does not judge someone's merit on the basis of their race, nor grant access to educational opportunities based on race.

Fortunately circumstances have changed in a way that allows us to select students through a variety of approaches that supplement the singular consideration of race with other more direct measures of disadvantage that are the consequence of a racially structured society.

What's being proposed in the new admissions policy?

The following principles were developed to guide the new policy. These are:

1. The goals of redress and diversity in the to-be-constituted first-year class should not be relaxed, and should in fact be advanced, i.e. a change to UCT's admissions system should continue to increase the number and proportion of African and Coloured students compared with the current situation.[7] It is thus a race conscious policy.

2. A portion of the class selected should be the very highest performing students nationally, regardless of their race or degree of disadvantage.

3.  The to-be-constituted first-year class should reflect a wider diversity of social class background by increasing the proportions of first-generation university students, students from poor schools, and from poorer households. Such diversity should span all "races".

4. The admissions policy should identify applicants with the highest potential, recognising that they may not have achieved the highest marks in the NSC or National Benchmark Tests (NBTs) due to the quality of the school they attended, or other obstacles or lack of enabling environments. Marks should thus be adjusted to take account of disadvantage.

5. Students selected should have a high probability of graduating, and therefore students will be given the necessary support and extended programmes, as required.

Since the desire for greater racial diversity, redress and fairness, remains unchanged, we propose to strengthen the first component of the admissions policy, i.e. the annual setting of race/population group targets for admission and the progressive stretching of these targets to achieve greater representivity in the student body. In time, we should look at setting additional targets for students from poor backgrounds or from rural communities, recognising that simply setting targets for black students will increasingly fail to achieve socio-economic diversity.

To reduce our dependence on race as a proxy for disadvantage, we have, over the last few years, been researching more direct measures of disadvantage. Those we have found to be useful include the quality of school attended, the education levels of the parents and grandparents, the dependence on social grants, and the language spoken by the parents where this is different from the medium of school education (English or Afrikaans).

Our application forms now request information about all these aspects of our applicants' backgrounds.[8] From this information, which notably does not include any reference to race, we propose to compute a disadvantage weight. Not surprisingly, most of the people who would be identified as disadvantaged through the use of these criteria are in fact Coloured or African.

This means that using such measures in a selection process still achieves racial diversity but in addition it promotes some increase in socio-economic diversity within all population groups. Moreover, it enables us to look at and quantify the experience of the individual rather than assuming that all members of a population group have had a similar experience in terms of home background and schooling.

The proposed policy is a hybrid model which has three selection bands each with its own criteria to constitute the class in the desired way. In the example below, the percentages are for illustrative purposes only.

*Note:

1. X+Y+Z must = 100

2.How X, Y and Z are determined may vary per qualification or per set of qualifications within a faculty.

To determine X,Y and Z the faculty must consider the enrolment and equity targets set for each qualification and the historic data about applicants. Allowance may be made for X or Y or Z to vary within a range as long as the total of X+Y+Z is 100.

3. In the pie chart above, X=60%, Y =15% and Z =25%

Band A: Merit measured by weighted FPS

The first band selects X per cent of the class (X having been determined by the Faculty based on historic applicant data and targets, especially equity targets, set for the cycle). In this band we factor in the disadvantage weighting to the Faculty Points Score (FPS) of the applicant. The FPS is either the National Senior Certificate score, or a combination of NSC and the National Benchmark Test scores, depending on the faculty in question.

For applicants in this band, the disadvantage weighting will add a specified amount to their scores. Thus if two applicants have the same FPS, the one who has achieved this in spite of a more disadvantaged school and home background is considered the stronger candidate and will be ranked higher. (See appendix 1 for more detailed explanation of the calculations.)

Band B: Based only on FPS

In the second band, Y per cent of a class (Y having been determined by the faculty) will be selected based only on marks (either NSC, or NSC and NBT combined) regardless of race or disadvantage.  This is because UCT wants to attract the top achieving students in the country and internationally; therefore we do not want to turn any such students away because they did not meet certain demographic or diversity criteria (see principle 2 above).

In some highly competitive programmes, this group is at present primarily White and Indian from relatively affluent families. In these programmes, therefore, we remain committed to applying policies of redress based on race, or a proxy for race, in respect of the large majority of our admissions. In such programmes the proportion in B and A will be small. In many less highly competitive programmes, band B by itself yields a significant proportion of the demographic representivity that UCT seeks. It can therefore be used to make offers to a much larger proportion of the class.

Band C: Target-based allocation using redress category as proxy for race

The third band makes up the remaining Z per cent of the class. This band is selected using the basket system. Our simulations have shown us that were we to rely on the bands A and B alone as the basis of redress in our admissions, the proportion of black students admitted to UCT would drop - this is because band A, while favouring African and Coloured students also brings in disadvantaged White students, and even with the disadvantage weighting, there may be a large number of White students with still higher scores.

To achieve the target numbers of black students, as per our existing admissions policy, we will select in band C, students in the interests of racial diversity per se. (One positive consequence of this approach is that it offers flexibility - performance in the NSC is changing rapidly -as we see in our annual analyses - and the magnitude of the third band will be re-evaluated each year and will probably diminish with time.)

The following table highlights how the size of the bands varies in different programmes using actual 2013 applicant data, and achieving demographic targets that are a few percent better than those achieved in 2013.

See appendix 2 for illustrative simulation of the outcome of the proposed policy in terms of different programmes.

Conclusion

This proposed hybrid, race-conscious admissions policy achieves a substantial move away from a reliance on race-classification in that for most programmes, about 75% of the class will be selected without race being taken into account. Most black students admitted will be admitted in open competition, recognising the obstacles they have overcome to achieve the marks they have.

Racial stereotyping will be reduced, since most black students will not be admitted because of their race. And it moves us towards the long term goal of not needing race classification to distribute opportunities.

It recognises that redress and social justice are promoted not through privileging people just because they are black, but because of how legislated race discrimination impacted and still impacts on their lives - their home backgrounds, their parents' education and cultural capital, the quality of their schools, and the impact of not studying in one's first language.

Yet we have found it necessary to keep ‘race' in the policy for three reasons:

First, the goal of greater demographic representivity remains and we still have a way to go to achieve it. Consequently, the first component of the admissions policy, viz. setting targets using race, and monitoring outcomes using race remains unchanged.

Secondly, because at a university like UCT, where competition for places is high, and the numbers of very high performing white applicants is so much higher than the number of black applicants even when weighted for disadvantage, it is necessary to select students for diversity per se - i.e. because they are black.

Thirdly, retaining race as a criterion for selecting a portion of the class is a recognition of the fact that race still matters because there is still racism, racial discrimination, stereotyping of expectations by race, all of which actually affect the performance of black students even at advantaged schools. Removing race altogether would suggest that race no longer matters.

As stated earlier, our ideal is that one day it should not; but for now, race as cultural capital, as identity, as a basis of current discrimination, as legacy of past exclusion, as sense of integration or alienation in institutions such as private schools and universities - that race is real. The debates around it must be vibrant and tough. Erasing race from admissions policy would do a disservice to achieving our ideals.

Dr Max Price

Vice-Chancellor

23 May 2014

Appendix 1. The redress weighting and how the model was tested.

We have identified the following indicators which have statistical significance and explain 56% of the variance in educational performance (measured in this particular model by performance on the NBT score for Academic and Quantitative Literacy).

Quality of school (ultimately the measure we found most reliable, and allows annual adjustment, is a ranking according to the average NSC score of all grade 12s of a school over a rolling five-year period, combined with some quintile and historic school authority information).

Education of an applicant's mother, father, grandparents (as three separate variables)

Language spoken by mother at home if Afrikaans or an indigenous SA language other than Afrikaans (in the final weighting developed we do not give points for Afrikaans)

Financial indicators, such as whether the applicant's family receives a social pension or child support grant.

The weight is a value from 0 to 10, and is either the home disadvantage score or the school disadvantage score, whichever is the greater. We do not add the two together since this would amount to double counting - almost all students in disadvantaged schools would also score high on the home disadvantage score, but not vice versa. Thus an applicant with a score of 5 for home disadvantage and 8 for school disadvantage will receive a score of 8.

The school disadvantage score is based on ranking schools based on their average NSC score for all students writing NSC at that school over a three year period. The schools are then grouped into the top 10% (score = 0), percentiles 86-90 (score = 6), percentiles 81 to 85 (score = 7), percentiles 76 to 80 (score = 8), percentiles 71 to 75 (score = 9) and bottom 70% (score = 10). Some schools do not fit this schema, and are individually placed.

The home disadvantage score is made up of the sum of points for parents and grandparents lacking a university degree (up to 3 points), indigenous African home language (6 points), and reliance on a child support grant or social pension (1 point).

Once the score is derived this is applied to the Faculty Point Score (FPS = the faculty's specific way of combining the NSC and NBT scores), to increase the FPS score by up to 10% as follows:

Final weighted score = FPS score x (1+ disadvantage score/100)

Example: FPS score = 70%     Disadvantage weight = 8

Final weighted score = 70 * (1+8/100) = 70 * 1.08 = 75.6

In the case of Health Sciences, because the gap between the top students (in the 90s and 80s) and disadvantaged students is so great (the latter in the 70s), a weighting of 20% is used. Example from health sciences: FPS score = 70%, Disadvantage score = 8: Final weighted score = 70 * (1+2*8/100) = 70 * 1.16 = 81.2%

Appendix 2. Modelling the proposed admissions policy

The model was tested for each faculty and programme, (in the example below, Mechanical Engineering) by using all the applicants for that programme who applied as first time entrants (i.e. school leavers) in 2012 for admission in 2013 and in 2013 for 2014. Those who were not eligible because of their subject choice, or marks below the faculty published minima or had not written the NBTs were excluded (i.e. the same as was done in reality for those applicants when they were considered in 2012).

The applicants were first all ranked in terms of their Faculty Point Score, and the top few selected to meet an arbitrary target for band B in the region of 15%. The rest of the applicants below this cut-off score then have their disadvantage-weighted scores calculated and ranked, and the next on the list are selected until the target of 60% of the class has been reached (i.e. 75% of the class in total).

The race profile of the class is then compared to the original race profile of the actual offers made for admission in 2013 and to the increased equity targets set for 2014. If the profile matches or exceeds the target, then more can be selected from bands A and B. But if the profile shows proportionately less black candidates, from the actual offers, then the remaining applicants are put into race baskets, and the offers are topped up from the ranked lists within each basket to achieve the desired race profile of the class, provided minima have been met.

Note: the targets can only be achieved by adjusting the proportions selected in Bands A and B. Hence the size of the bands and the cut-off scores for each band differ across qualifications.

The following table demonstrates the outcomes for Mechanical Engineering comparing the actual offers made in 2013 (2^nd row) to the offers that would have been made if the proposed 3-band hybrid admissions policy had been used (Bottom row).

The percentages in the right hand column indicate that 22% of the offers would have been made based only on marks (FPS); 53% on disadvantaged-weighted FPS; and 25% using race. Seventy-five percent of offers would have been race-blind.

The following example is a modelling of admissions in the Science Faculty. (Note the different faculties use different FPS systems so in Science the score is out of a maximum of 600, whereas in Mechanical engineering it is out of 100. In the Faculty of Health Sciences (table 3) the maximum FPS achievable is 900).

In Science, 95% of the offers are race-blind, and 5% are race based.

In Medicine (MBChB), 58% of the offers need to be race-based to achieve the equity targets. This is because of the very competitive nature of admissions which pushes the cut off both for open offers (band B) and even for weighted offers with a 20% weighting( band A) to such a high level that there are relatively few black applicants who are competitive at that level. Bands A and B must therefore be much smaller so that the diversity targets can be met from band C.

Appendix 3. Example of some of the complexities of applying the disadvantage weighting and why each faculty and programme has unique needs.

Faculties, and even programmes within faculties, differ with respect to their specific selection criteria, minimum scores in specific subjects, whether or not National Benchmark Test scores are included, whether Maths is required, and the different approaches used to admit students into extended programmes. This has meant that we need different models for each faculty, and different models for extended (ADP) and regular programmes.

Just one example can illustrate this: since regular and extended (ADP) programmes for the same degree may have different minimum Faculty Points Score (FPS) requirements, adding points for disadvantage to someone whose FPS is only sufficient to get into the extended programme may raise an applicant's position up the rankings to make him or her a preferred admission applicant in the regular programme, yet the applicant will not be able to cope in the regular programme. In this case there is no point in displacing someone with higher points who would cope in the regular programme.

The points for disadvantage should thus only be used to choose between students eligible for admission to the regular programme on the basis of their unmodified FPS. The points for disadvantage could also be used to choose between students in the extended programme but only amongst those above the minimum FPS for that programme. The model has to be applied differently to these two groups.

Footnotes:

[1] These race categories are as defined under the Apartheid Population Registration Act. ‘Race' in most places in this paper is used synonymously with population group as defined in that Act. Reference to African students is to black South African students. The term ‘black' is used for African, Coloured and Indian students collectively.

[2] For 2015 admissions this is on the basis of the race-classification of the applicant's parents under apartheid.

[3] While the very top marks of both groups may be similar, the NSC marks of most of the top decile of black applicants will stretch lower than those of the white top decile because very few black students are in advantaged schools and most are in lesser quality schools and environments, even if they are in the top decile.

[4] We also get feedback from these black students that they resent the assumption that they wouldn't make it on their own merit, and the attendant racial stereotyping - itself an apartheid legacy - that views all black students as suffering from an academic deficit.

[5] We get feedback from school principals about the tensions this causes when, after 12 years of schooling in which all learners are treated in a race-blind manner, and have been taught to aspire to an egalitarian, equal opportunity, non-racial society, they find on entering grade 12 and applying to UCT, that they have to identify their race, knowing that this will substantially change their prospects of being offered a place.

[6] If one parent would have been classified white and the other Indian, the children might legitimately classify themselves as Coloured, but in reality suffer none of the legacy of apartheid disadvantage deserving of redress. Similarly children of immigrants from other African countries who have SA citizenship classify themselves as African South Africans or as black, and benefit from affirmative action at the expense of other black and white South Africans.

[7] Indian and Chinese students in most programmes already exceed the targets for redress.

[8] This approach applies only to South African citizens or permanent residents.

Source: The University of Cape Town.

Click here to sign up to receive our free daily headline email newsletter