Is the Catholic Education Service statistically illiterate?

The Catholic Education Service (CES) has a long history of abusing statistics, both to justify its schools’ discriminatory admissions policies in general, and to force the removal of the 50% cap on religious selection in particular.

It has long claimed that its schools are socio-economically inclusive, by comparing how poor the areas its schools are located in to the national average. Catholic schools, the CES says, are disproportionately located in poorer areas. This shows that they are disproportionately inclusive of children from poorer backgrounds, right? Wrong. All this tells us is that its schools are in inner cities. It doesn’t tell us anything about the pupils themselves. In fact, when you compare the intakes of Catholic schools to the children living near those schools, you see that Catholic schools actually take  pupils who are much better-off than the local average. Moreover, when you do the same comparison with the national average, Catholic schools still take a disproportionate number of well-off pupils.

The CES has also long claimed that its schools are much more ethnically diverse than the national average. But its schools are only diverse to the extent that Catholics themselves are diverse. This means its schools take lots of white and black pupils. But in terms of taking pupils from ethnic backgrounds not associated with Catholicism (i.e. Asian minorities), its schools are massively exclusive. This is true in comparison to both the national and local averages, meaning they are still not as diverse as schools of no religious character.

More recently the CES has started to boast about the number of Muslim pupils in its schools. In fact, its schools take about 2-3 times fewer Muslim – and 10 times fewer non-religious – pupils than there are in Britain as a whole. While 15 of its schools have Muslim-majority intakes, around 73 are located in areas where most pupils are from Muslim families.

Yesterday the CES added another line to this sort of faulty argument, when it argued that a 2017 poll showing rank-and-file Catholics are against discrimination in admissions is unsound:

‘A source within the CES… told The Tablet that the results of the survey quoted in the letter were misleading. The letter cited 80 per cent of the public – 67 per cent of which they said were Catholic – opposing the government’s consultation on the policy.

The survey, conducted by research and strategy consultancy Populus in May 2017, asked 2,033 people, 129 of whom identified as Catholic, their thoughts on admissions policies for state-funded faith schools.

This is a very small sample size of Catholics compared to the thousands of Catholics who have petitioned the Education Secretary to lift the cap, the source said. At the last count 15,000 faithful had signed the petition.’

Let’s explain how the statistics here work. First, let’s work out how many Catholics there are. The latest British Social Attitudes Survey suggests that about 8.6% of British adults are Catholics. The ONS says the British adult population is just over 50 million. So this comes to 4.3 million Catholics.

A self-selecting sample of 15,000 out of a population of 4.3 million tells us absolutely nothing about the views of the 4.3 million as a whole – not least given that the 15,000 were responding to a call to action promoted by Catholic dioceses.

There were in fact 149 Catholics in the Populus poll, not 129, of whom 67% said they’re against the dropping of the faith school admissions cap. This may sound like a small number. But we have to remember that, unlike the 15,000, those 149 were selected to be representative of the population as a whole. That scientific approach is why people pay polling companies to do research, instead of just asking the same number of their friends and supporters what they think, or asking people who visit their organisation’s website (which is basically what a petition does). And, indeed, if you stick a 4.3 million population figure and a 149 sample figure through a margin of error calculator, it spits out a margin of error of almost precisely +/-8%.

8% is not a small margin of error. But it is easily small enough to mean that that 67% undoubtedly represents a majority of Catholics either way.

So, yes, CES, 149 representatively selected Catholics is a significant number. Whereas 15,000 self-selected Catholics is not.

FSA team