Causality and statistics

One of the first things that people learn when they start taking statistics seriously is that most of the statistical studies out there are completely meaningless (or as some people might put it, 87.2% +- 0.5% of them - sorry, couldn't resist). The reason they are meaningless is that they try to conclude causality without actually having any methodology to do so. The example that triggered me to think about it was on a Brazilian newspaper:

Ficar muito tempo na frente da TV e do computador aumenta depressão - Staying for too long in front of the TV or computer increases depression. The problem with this story is that there is certainly a causality that goes on the other way: people that are depressed have the tendency of isolating themselves from social contact and end up staying in front of their TV and computers more often.

Then they try to prove causality by using time as a causal agent: they look at adolescents that watch too much TV or stay too much in front of their computers and then check how many of them end up more depressed. Then they start trying to explain it by saying that if they interact more with people it makes them less likely to be depressed. However, if they are already spending too much time in front of computers and TV when they are young, maybe it's because they don't have engaged parents that provide them with more things to do. And that could be very well be a much more important source for depression (lack of stability at home) and watching too much TV and staying too long at the computer are just a measurable side-effect.

Actually, those types of news are what make me depressed. And I read them because I read things on the internet. If I didn't spend this much time on the internet I wouldn't be depressed. So there you have the causal connection!