Friday, 11 March 2011

On reporting statistics

I've read a number of articles this morning that all take the form of "doing X increases / decreases your change of getting Y by Z%".  Not one of them state what the baseline chance of getting or avoiding the condition actually is.  If your 25% more likely to get liver disease by drinking more than 1 glass of wine a day, but the chances of getting it when drinking nothing are 0.001% then that increase is insignificant.  If it's 10% then it makes quite a difference.  My basic suspicion is that most of the baselines are closer to the former than the latter which is why they go unreported, they'd fail to have the "scare" impact if people looked at the figures and realised that 90% of fuck all is still fuck all.

No comments:

Post a Comment