Crap Car Crash Correlation

One of the reasons I like to read the Inquirer is because they're so god damned sarcastic! I wasn't going to write about a new piece of research which claims that players of video-game driving simulations take more risks when driving, simply because it's a bit of a stupid conclusion. But when I saw the Inq's take, I thought I'd share:

The latest in a long – and frankly tedious – line of shock horror research aimed at maximum publicity in the media and thus generating more cash for further pointless research, reveals that people playing driving games are more likely to take risks on the road in real life.

...Let's examine some of the issues here. The drivers found most likely to take risks after playing video games were males aged under 24. This maps precisely onto the demographic for drivers most likely to take risks anyway, a fact reflected in their insurance premiums. And staying up all night playing driving games is hardly likely to reduce the tiredness and increased reaction times that lead to many accidents.

And so you see this is one of the disadvantages of a correlation. You can't say that driving games make people bad drivers, because, as the Inquirer points out, the bad drivers could be those people who are drawn to play video-games. It is the same with most correlations you see in the press: 'video-games increase violence', 'pornography leads to rape', 'cannabis leads to schizophrenia' - but when you begin to understand the limitations of the correlation, you begin to get very wary of these kinds of big claims. Yes, some correlations do have an obvious direction (the further you fall, the more it hurts), but when you're dealing with complex human behaviour, the answers are never so simple. (Wikipedia has some more on this, under correlation does not imply causation).