"When studying statistics in the 70s one of the best examples cited was the strong correlation in the increase in lung cancer and the rise in membership of the Labour Party, a good reminder that correlation between two sets of data does not prove that one causes the other."For some reason I was reminded of this.
Thursday, February 15, 2007
Labour causes lung cancer
Ok, clearly a joke title, but in my post this morning where I mentioned the problem of mistaking correlation for cause, someone added the following comment
Subscribe to:
Post Comments (Atom)
3 comments:
On the causation/correlation question, it seems to me that we only ever know of causation through correlation. We can do controlled experiments to hone on the causative factor, but that's only going to work if we understand everything sufficiently well; we could come to the conclusion, for example, that A causes B through controlled study, but be completely unaware of C, which actually causes both A and B. So, what we need is a good model to eliminate hidden causes as best we can, which is clearly not going to be terribly possible when you're studying past events.
I was just wondering Adam, if the reason that my Missus talks rubbish so often, is that, years ago, she was fashioned from some idiots rib?
If t'is so, that God chappie's got a lot to answer for.
I couldn't possibly comment.
Post a Comment