May 05, 2005

B.S. detector: replace batteries annually

Some people, when they're faced with a statistic that seems spectacularly hard to credit and against all reason, spend a lot of time trying to confirm/disconfirm it, analyse it, try and figure out if there was some misquotation/misunderstanding involved, and generally beat it to a pancake. For instance, the allegation that 99% of all child porn users are also Trekkies. You can spend your day figuring out what bizarre statistical confluence of events could possibly lead to such a remarkable correlation. Or you could just assume, as I automatically did, that the police sergeant in question was blowing smoke out his ass. A passing knowledge of police sergeants and their general comfort and familiarity with scientifically acceptable levels of statistical and factual accuracy is helpful, but not exactly required.

Or you could try and figure out why, when a person writes a paper saying that fully 99 per cent of climate scientists do not explicitly back the global warming consensus, that that paper was not published? Was it political? Is science now for sale? Or you could just conclude the paper-writer's method must have been obviously full of crap if that's the actual number he came up with, and go back to some more useful activity, confident that somebody, somewhere will actually read the rejected paper some day, and find the inevitable simple error, whether of math or bias that led to such a completely and obviously improbable end-result.

Looky here...

Posted by BruceR at 05:37 PM