All together now: Correlation does not imply causation

I’ve moaned about this before, and I agree with D’Arcy Norman who rants that it should be taught to every child. So when on the radio yesterday I heard a BBC reporter talking about a new report that looks at marriage, cohabiting and single parents in the UK, it inevitably got another airing. One of the findings was that children of married couples were more likely to stay on in education. ‘So,’ the reporter concluded, ‘it seems that being married has benefits for the children over just cohabiting.’

No, no, no – there will undoubtedly be a number of other factors here. For example, maybe married couples are more conformist and thus their children more likely to succeed or enjoy education. I’m not suggesting this is a factor, who knows? There could be a thousand such factors at work (I think they controlled for socio-economic factors, so we could rule the obvious ones out). But it would be unlikely that marriage causes children to stay on in higher education. The danger of such lazy reporting and interpretation is that daft politicians (stand up David Cameron) start making policy on the back of such ‘evidence’.

I’m married so this isn’t a personal defence of cohabitation, just another example of the correlation error, which seems particularly prevalent in research around families. Nearly all of the guides to good parenting are based on this error – ‘Parents who are X with their children have children who turn out to be X in later life. Therefore being X causes all children to be this way.’ Where X is any attribute or habit, good or bad (gentle, loud, aggressive, kind, patient, etc). This becomes advice on how to parent, ignoring the very obvious genetic influence (‘Parents who are X will tend to have children who are X also’).

I think I’ll have it on my tombstone ‘Correlation does not imply causation – except in this case.’

Leave a Reply

Your email address will not be published. Required fields are marked *