...because false positives become endemic in any research that leverages huge datasets and effectively infinite processing power. This tendency to throw every variable against every other variable to manufacture 'theory' is pernicious in the extreme. It leads to false positive hell. These 'cats say it's even worse than that because since in financial economics the data is standard and available to all any novel finding is likely a false positive so the statistical test for significance should be a T value exceeding 3.0 rather than the traditional 2.0. I think a lot of biomedical research should face the same standard. Psychological and social research probably needs an even higher standard because most of it is absolute ideological (and I'll let you guess which one) goal driven crap. Richard Thaler and Cass Sunstein: yeah, I'm lookin' at you - your days of blowing questionable statistical claims out your keisters and getting the gun toters to do your bidding are numbered. Nudge that school boys.
The more our glorious stupor state has funded research the more dreckish it has become. Free money from bureaucracy corrupts everything it touches. Everything. Because free money has no valid human agent behind it demanding a return for the investment - you know, what we call in the private sector 'results'. Read the whole damned thing. Here.
No comments:
Post a Comment