I love analysing social science statistics, especially when they’ve been in the hands of PR professionals and journalists. I think it’s probably because errors are easily disputed and easily traceable, so we can see how social knowledge is created and transmitted: tracking back sources is usually easier as one can search for specific ‘facts’, and where those ‘facts’ feel wrong one can search out the original data. Here I’ll discuss two of this weeks stories, one on burglary and one on university entrance.
The ‘burglary’ story caught my eye because the hotspots seemed unlikely (Chorlton-cum-Hardy, Blackheath) and there seemed to be some slippage in what the story was about. The story is originally from moneysupermarket.com (so it’s a PR story) who analysed their insurance inquiries. So the first important observation is that the data is from a) people using the internet, to b) renew their insurance, and c) doing some shopping around. This is a biased sample in lots of ways: obviously they are people who buy online and have insurance (around 10% of people aren’t covered, presumably the poorest). That they are shopping around suggests that they have made a claim recently too. It’s not hard to imagine that this biases the sample towards 25-34 year old professionals.
Furthermore, the original press release gives away the slippage. Although it has ‘home theft hotspot’ in the title, and spokesperson quotes like ”Home is where the heart is and there’s no denying that having it burgled is an emotional and frightening experience”, the data refers to ‘claim[s] for theft on home insurance’. So this includes muggings, thefts from the beach, bikes being stolen, lost wallets on the bus and so on. Again, this is probably biased to young professionals: the kind of people who get their phones nicked in the pub, or have their bike nicked from outside.
So to education.
‘Just 1% of poorest students go to Oxbridge’ cries the Guardian. ‘In contrast, 10,827 students attending Liverpool John Moores University and the University of East London claimed a full bursary – 4.7% of the total for the whole country’
This is statistical nonsense. Without knowing how many students in total go to each university we can’t know if these figures are better or worse than average, or bang on. Perhaps 10% of all students go to Liverpool John Moores or UEL, so they aren’t attracting as many poor students as they should. We also don’t know if there are other of the poorest students who are attending and not counted as they get other forms of support or none at all, but I’ll let that go.
The original report has the more important figure, ‘the proportion of full fee-paying students this number [those receiving bursaries in the lowest income group] represents’. This in itself is problematic as bursaries are only for UK students, and I don’t know if the ‘full fee-paying students’ include EU and overseas students: I suspect not, but if so then those with more non-UK students will have a lower percentage.
Anyhow, the figure for Cambridge of full-bursary over students is 11.1%. This is low, but some are lower: Grennwich – 9.1%, Guildhall – 11%, Leeds Met 0.4% (very strange, I need to look into this). UEL is high at 61% but LJM isn’t particularly high… it’s just a big university.