Wednesday, November 10, 2010

Correlation vs Causation?

I never really did enough political science to get a proper understanding of maths and stats, but the one handy question I learned (which can be asked of a lot of presentations) is whether the findings show causation (i.e. A actually results in B) or merely correlation (i.e. the two merely happen to occur together, perhaps because both are results of C).

Unfortunately, I see no evidence that anyone asked that question before this scaremongering claim that texting leads to underage drinking and sex:
Parents have been warned to watch out for signs of excessive texting in their children, amid concerns it poses a new health risk.
Teenagers sending 120 text messages a day are more likely to drink, smoke and have sex, claims a US doctor.

That teenagers who text heavily are more likely to have done these things does not mean that texting causes them to engage in such activities. The causation could possibly run the other way round or, perhaps more plausibly, it could be that both behaviours are explained by some third factor, such as high disposable income or lack of parental supervision.

1 comment:

  1. a) 120 texts a day? 120 texts a day? WTF? That's about 8 an hour. It's a wonder they find time to drink, smoke and have sex.

    b) What exactly is the issue with teenagers drinking and having sex? Neither of these things are intrinsically harmful, and both of them are usually enjoyable. As long as the sex is consensual and the drinking isn't resulting in them doing anything too stupid, I'm really not sure what the problem is.