Core/Margin: HEFCE publish some more data

HEFCE have now published some more core/margin data here. This is an annex to the circular they had already published here, and which I have already discussed in too many places, but principally here. This post builds on David Kernohan's post here, now that we have some more data.

I'm going to take SOAS as a case study. SOAS is a 1994 Group institution, if perhaps not a very typical one, but it is also small, making the data rather more tractable. From the Unistats data you can see that the breakdown of students' qualifications across different subjects is like this:

Subject	Median tariff	% Below 359 points
(L3.53) Politics	410	10%
(L3.60) Law	400	15%
(L3.52) Economics	420	19%
(L3.56) Anthropology	375	25%
(L3.87) Asian studies	400	29%
(L3.57) Others in Social studies	395	30%
(L3.88) African and Modern Middle Eastern studies	370	43%
(L3.90) History	350	53%
(L3.94) Theology and Religious studies	330	60%

The median is the tariff value of the student in the middle of the distribution, the % below 359 shows (roughly!) how many students are below that AAB threshold. You can see that the median is not precisely correlated with the length of the tail - economics has a high median but also a fairly long tail.

The Heidi data don't give an exact match, but you can see where most of the enrolled students are sitting:

Subject	Enrolled first year Undergrads
Social studies	240
Law	85
Business & administrative studies	45
Languages	275
Historical & philosophical studies	155
Creative arts & design	15

Now, looking at that HEFCE data, we can see that SOAS should expect to loose a fair chunk of their quota places. I think it looks like this:

Current Places	714
Expected Places	269

Go back to the Unistats data and let's look at the position in Economics across all of London:

Institution	Median tariff
LONDON SCHOOL OF ECONOMICS & POLITICAL SCIENCE	540
UNIVERSITY COLLEGE LONDON	480
SCHOOL OF ORIENTAL AND AFRICAN STUDIES	420
ROYAL HOLLOWAY, UNIVERSITY OF LONDON	380
CITY UNIVERSITY	340
BRUNEL UNIVERSITY	300
KINGSTON UNIVERSITY	200
UNIVERSITY OF EAST LONDON	180
MIDDLESEX UNIVERSITY	170

SOAS seems to sit in a comfortably high position, but consider the tail again. Nineteen per cent of the enrolled students are below 359 UCAS tariff points, so almost certainly not AAB, given that they will mostly have other tariff-bearing qualifications than their three best A-levels.

So suppose the LSE decides to take 10% more economists now that their AAB numbers are uncapped. UCL might have to respond by taking a few students whom they previously would have rejected, and so on down the line. SOAS might have done the same. But now SOAS only has 269 places available for students below AAB. The majority of the students even at City and Brunel would seem to be below that line, based on these tariff data. So the pool for SOAS to fish in is small. If the LSE decides to expand Economics, SOAS may have to take capped places from History or Theology to sustain their own economics programme - essentially cannibalising their 'weaker' programmes to sustain the 'stronger' ones. I suspect this will be controversial with the staff concerned...

Consider the alternative case of Kingston. Just 6% (203) of their UK-domiciled entrants in 2009-10 were AAB and above. They are therefore largely immune to poaching by the institutions above them, and can be sure of getting most of their numbers. Even if demand collapses, they can still take places from UEL or Middlesex, because all the students in question are below AAB, so the quota is still there to recruit them. Since I think it is unlikely that overall demand will collapse, it might be better to be sat in Kingston, UEL or Middlesex than SOAS come September 2012...

So my lessons for today are:
• The AAB policy locates risk very sharply in certain institutions which recruit many of their students at about this boundary, these are not the institutions we conventionally call the 'squeezed middle', but actually rather prestigious places;
• Within those institutions, different patterns of recruitment in different subjects will distribute the risk differently. It may actually be the 'strongest' programmes in certain institutions which are most at risk;

My lesson for tomorrow (it may not be tomorrow when I post it, of course) is how poor the data quality is. The proportion of entrants with known data in the HEFCE spreadsheet ranges from 0 to 100%, and at the median institution, it is about 50%. More on that later.