I’ve found a few recent posts about lying with statistics:

## Skewing Statistics for Politics

This entry over at Good Math is about the selective reporting of statistics to make your argument look better but ignoring other statistics that make your argument look wrong. In the meantime, he takes on Powerline.

## Anti-mercury warriors descending further into the depths

While I don’t agree with Orac very often (and I have been known to throw some vitriol his way) and still question his overly sanguine apparent reliance on clinical trials, I think his dicussion of multiple comparisons is pertinent here. I’ve been meaning to cover the topic for a while now. There are some subtle issues that he glosses over because, well, they’re subtle and probably very hard to explain. Despite my minor quibble with his argument (apparently caused from late-night hand-waving), it does serve to illustrate that multiple comparison issues are not something just cooked up in a statistician’s lab to give other people headaches. There are real consequences to this problem, and it deserves a lot more discussion.

## Dishonest Dembski:the Universal Probability Bound

Another of Mark’s “bad math” expositions, he takes on the so-called Universal Probability Bound. The Universal Probability Bound is such a stupid concept, pretty much invented in modern times for the sake of mathematically arguing for intelligent design at least from what I can tell. (See my opinion on ID here.) (I’m not sure while Emile Borel was calculating UPBs.)

At any rate, trying to make a probabilistic argument for the proof of a creator sounds pretty weak to me. Statistics was invented to analyze gambling, and later turned out to be highly useful for the physical sciences. One day science and statistics may advance far enough to start examining supernatural issues, but we’re definitely not there yet by any stretch of the imagination. There seem to be only a few skirmishes, such as young earth theory (which isn’t really part of creation theory), where the two come into conflict. The rest of what scientists have to say about the supernatural (no matter what they do say) really seems philosophical to me. And mathematically arguing for the existence of a creator really seems like you’re stretching mathematics too far. Mathematics and statistics are just tools we invented to try to explain the world, and they really need to be left at just that. (Integral Naked’s 6/15/2006 podcast has a similar take on quantum physics and the Unmanifest.)

I don’t have a problem with people who believe in a creation theory. (I myself believe in one.) I do have a problem with people who try to mold mathematics and science to try to “prove” their theory.

Technorati Tags: statistics

Filed under: Lying with statistics |