Lies, Damned Lies and Religious Statistics
[He] uses statistics as a drunken man uses lamp-posts – for support rather than for illumination
– Andrew Lang
If there’s one thing I dislike about Twitter, it’s the way some people seem to repeat the same handful of slogans in support of some cause or another over and over, with occasional variations here and there. I usually tune them out, but there was one that did the rounds recently which caught my attention:
If every #atheist left the USA, it would lose 93% of the National Academy of Sciences but less than 1% of the prison population.
That sounded like quite an interesting statistic, but it also sounded a little bit suspicious, so I thought I’d do some digging. It turns out that this claim, or similar ones, can be found all over the place, but – surprise surprise! – the full story is a bit more complicated than that.
First of all, the figures for the National Academy of Sciences (NAS) and prison inmates come from different (and slightly dated) sources. This appears to be the survey of NAS beliefs, dating from 1998, and this looks like the source of the prison stats, as at 1997. That may not sound like a big deal, and it isn’t really, but it’s not trivial, either. If nothing else, it means that the statistics were collected around 15 years ago, at different times with different methodologies, which makes any comparison tricky.
The NAS survey asks whether the scientists believe in a personal God, allowing the three options of belief, disbelief, or doubt. The prison statistics cover religious affiliation, and exclude those giving no response, so immediately, there’s a clear difference. Affiliation is related to belief, but it isn’t the same thing. It’s possible to feel a connection to a religion – an affiliation – whatever your beliefs. There’s also the question of whether an atheist or agnostic would typically respond at all – they don’t have a religious affiliation – so they may well be over-represented in the excluded “no answer” group.
My next quibble is that agnostics are being treated differently in the two datasets – they’re counted as honorary atheists to reach the 93% figure for scientists, but within the prison population, where agnosticism wouldn’t qualify as an affiliation and augmentation of the atheist/agnostic figure would be undesirable, they’re absent completely, most likely either excluded in the “no answer” group or allocated to whichever church or religion they feel some sort of connection with. In any case, given that they’re counted as effective atheists in the 93% quoted from the NAS stats, the absence of an “agnostic” category for prison inmates should ring alarm bells.
In addition, I have doubts over whether the prison statistics accurately reflect reality. It’s commonly understood that a profession of some sort of faith is helpful in prison, thanks particularly to the religious prejudices that can be found on parole boards. This shouldn’t be news to atheists, as it’s the sort of institutional religious discrimination they often complain about, with good reason. Nevertheless, despite the obvious risk of underestimating the atheist prison population, these figures have been taken as the final word – dare I say “gospel” – on the number of atheists in prison.
And of course, even if it’s clear that there’s a significant difference in the theological positions of scientists and prison inmates (which I believe there is, despite the methodological flaws), that proves nothing about the truth or otherwise of those positions. To claim that it does is to engage in an argument from authority (or possibly argumentum ad populum) on a massive scale. Of course, some people who quote this statistic are happy to acknowledge that they’re just playing around, but most of the time, there’s no indication that correlation may not indicate causation.
I understand that there are reasons why atheists (especially in the USA) feel a need to justify their existence, to give reasons why they’re valuable to the country, and not the divisive, unpatriotic troublemakers they’re often portrayed as. I understand that this sort of statistical jiggerypokery is more often defensive than offensive. But this sort of weak and easily-countered argument doesn’t help to make the case that atheists are rational in their approach.
I’m especially pleased to see that Hemant Mehta has picked up on a recent Pew Forum survey which (while potentially just as flawed in its own way as the 1997 figures) suggests that the true proportion of prisoners who are atheists may be significantly higher than 1%, giving a figure of 10%. It would be nice if that was reflected in future claims.