Monday, 30 November 2015

Varsity magazine

Bathsheba Wells-Dion wrote an interesting article in Varsity magazine about the rising number of Cambridge undergraduates from state schools. The article is interesting not because it is correct but because it presents the same flawed statistical analysis that has been quoted to me over and over again since I was at Cambridge 20 years ago. Not just quoted to me verbally but even quoted in written bigoted hate speech posted up around college and pushed under our room doors. This misinformation feeds hate and Varsity are broadcasting it.

Specifically, the article says:

"However, for Cambridge as well as Oxford, private school students remain dramatically over represented in admissions statistics. Only seven per cent of children in the UK go to private schools but 37.8 per cent of students admitted to Cambridge in 2014 were from private schools."

Firstly, the claim that 7% of children go to private schools is true for primary schools but Cambridge University candidates aren't applying from primary school. They are applying from secondary school where (according to the ISC), over 18% of pupils over 16 attend private schools and they are the ones applying to Cambridge. So even if we assume the student populations at private and state schools are equivalent (which they aren't) the discrepancy is more like 2x rather than the fantastical 5x reported in this article.

Secondly, grammar school students are also "dramatically overrepresented" for the same reason as private secondary schools: they are highly selective. In other words, only the best candidates get into the school in the first place. Consequently, one cannot assume that private school students are representative of the general public as Bathsheba Wells-Dion has done and, therefore, the conclusion that "private school students remain dramatically over represented" is not justified by the numbers quoted.

So the statistics presented, while fascinating, do not allow us to draw any strong conclusions at all and certainly do not justify the tabloid-like conclusion presented in the article. The real question is how much of the discrepancy can be fairly attributed to selection bias and how much (if any) is difference in quality of education (e.g. smaller class sizes).

The article goes on to tug at the heart strings by bringing up "imbalance in university life" despite the fact that it is presenting cherry picked misinformation designed to support the conclusion the author set out to draw.

Let us just consider how this alleged problem might be addressed. The simplest solution is surely just to remove information about public vs private school background from the admissions procedure. That would surely be preferable to publishing bad science in Varsity in order to support a hate movement that is already discriminating against kids based upon their background. If the objective is for admission to be based upon candidates "demonstrating their potential to excel at Cambridge" then why not get all the candidates in a room, teach them something randomly chosen from their course and examine them at the end?