A report by the Sutton Trust for the Times (£) has been making waves due to its claim that "only ten schools produced 12 per cent of the country’s most senior businessmen, politicians, diplomats and leaders of the professions".
The breakdown varies significantly between professions. So, for example, the report claims that, of leading journalists, 25 per cent were educated at a grammar school and 52 per cent at independent schools (the rest were educated at comprehensive, secondary modern, direct grant, or other state schools, some of which may also be selective), while 37 per cent of leading politicians were independently educated and 27 per cent at grammar schools.
In some of the reporting that followed, the "leading" aspect of the report managed to fall away, making it seem like three quarters of all journalists were educated selectively. But it is also important to look at the methodology of the Sutton Trust report (pdf):
The study is based on 7637 people educated at secondary school in the UK, whose names appeared in the birthday lists of The Times, The Sunday Times, The Independent or The Independent on Sunday during 2011. These lists of names provided a snapshot of the country’s leading people across a range of sectors.
So the report lives or dies on how accurately the names on the birthday lists of four newspapers represent a "snapshot of the country's leading people". If there are any longstanding biases – if, for instance, journalists working online are less likely to be mentioned than journalists working in print, or if religious leaders from some religions are more likely to be mentioned than others – then that could severely reduce the usefulness of the report.
For instance, there's a very important point which nobody has made about this data set. It shows that the majority of leading journalists are independently educated – using a set of names put together by journalists. If those journalists are more likely to include people like themselves on the birthday lists – even unconsciously – then that could skew this report considerably.
Similarly, almost an eighth of the data set was cast out because schooling information was not available. Again, unless we are sure that that information being unavailable is uncorrelated with where someone went to school, a significant bias is introduced. It is reasonable to suggest, for instance, that Eton College keeps far better lists of alumni than most comprehensives do – so nearly everyone on the list who went to Eton would be included, while a number of state educated people may slip through the cracks.
None of which means that the conclusion of the report is necessarily untrue. In fact, given what else we know about the concentration of power in Britain, its broad claims are very likely to be correct. But it is an instructive example of how important methodology sections of studies like this are – and anyone quoting the actual figures ought to be aware that they need to be served with a hefty grain of salt.