Friday, November 23, 2007

The problem with debunking bad science...

... is you need to not fuck it up.

If you're going to flame science papers for bad mathematics, you better not do it yourself!

Junkfood Science quotes Dr John Ioannidis as follows:
"the probability that the research finding is true is only 1.5".

Sandy seems unimpressed by this figure she quotes.

[No, I didn't quote-mine there and leave off an important word like "percent"... the quote continues " — hardly any higher...".]

I don't know about you, but I'm very impressed by a probability larger than 1.

Having something occur more than 100% of the time seems quite dramatic enough to me.

Exactly how much more often than "occurs 150% of the time" would it take to make you guys think the effect was strong?

Maybe Junk Food Science should check the quote, because statisticians tend to get a bit skeptical when probabilities don't come between 0 and 1.

In normal circumstances, I wouldn't post something like that here, I'd put it in comments on the original blog, but junkfoodscience doesn't seem to allow comments. Doesn't even give an email address. Since I can't say it to them quietly, I will say it to everyone loudly.

No comments: