Coyote Blog | Warren Meyer | A Truly Bad Study
Imagine this study: An academic who is a strong Democrat wants to do a study to discover if Republicans suffer from a psychological tendency to bizarre conspiracy theories. OK, the reasonable mind would already be worried about this. The academic says his methodology will be an online survey of the first 1000 people who reply to him from the comment sections of certain blogs. This is obviously terrible -- a 12-year-old today understands the problems with such online surveys. But the best part is that he advertises the survey only on left-wing sites like the Daily Kos, telling anyone from those heavily Democratic sites that if they self-identify as Republicans, they can take this survey and their survey responses will be published as typical of Republicans. Anyone predict what he would get?
It is hard to believe that even in this post-modern academic world, that such a piece of garbage could get published. But it did. The only difference is that the academic was a strong believer in global warming, he was writing about skeptics, and sought out survey respondents only on strong-believer sites. What makes this story particularly delicious is the juxtaposition of the author's self-appointed role as defender of science with his atrocious scientific methodology. The whole story is simply amazing, and you can read about it at JoNova's site.
Wow. This is astonishingly poorly done. Just astoundingly bad. Not even wrong levels of badness.
I've actually read quite a few of Lewandowsky's cog sci and neural nets papers. It's solid work. He's certainly capable of being intelligent and rational. That only makes this study all the more worrisome. If he did sloppy CogSci/NN stuff and then did sloppy climate research, I could at least conclude he's just a perennial half-wit. Better that this study be vomited up by some intellectual slob than a sharp thinker who decided to let himself be sloppy and dishonest as soon as politics entered the picture.