Syphilis (Or Was It Facebook?) Blamed For People Not Understanding That Correlation Does Not Mean Causation

from the not-without-a-chi-square! dept

I really really really wasn't going to write this post, but so many people kept submitting it, I figured it needed to be done. The Telegraph has some ridiculous story claiming, without any actual evidence, that Facebook is "linked to the rise in syphilis." Quite a claim. The evidence? Oh, that's not included. There's just some public health guy claiming that there's evidence -- without presenting any. About the only thing in the article is that (a) more people in this particular area of the UK seem to be reporting that they got syphilis (b) people in that area are also (marginally) more likely than in other areas to use social networking (c) at least some of the people who got syphilis mentioned that they have met sexual partners via Facebook.

So, yes, you have a bit of weak correlation combined with self-selected anecdotal bias. And that proves what? Uh, absolutely nothing. So, please, for the sake of the sanity of statisticians everywhere, please learn to practice safe statistics, where before you claim something is linked to something else, you actually use "protection" in the form of some real data.

Filed Under: causation, correlation, social networks, statistics, syphilis, uk