Wednesday, 2 February 2011

Chaos, correlation and cause

I am regularly amazed by the way in which I can read the same research as Oliver James and come to the opposite conclusion. For example, a couple of weeks ago, he wrote this post arguing that we ought to clean our houses more. It's based on this paper, which shows that levels of chaos in a household predict children's behaviour even after accounting for parenting practices. Household chaos was measured by a short questionnaire including statements such as "you can't hear yourself think in this house". A cross-sectional study. Now, surely, it is at least possible that these households are more chaotic BECAUSE they include children with behavioural difficulties, rather than the reverse? This is statistics 101 isn't it? Correlation does not equal causation? Or it might equal reverse causation?

In any case, I don't think its strong enough evidence to make me clean my house.

No comments:

Post a Comment