Thursday, January 14, 2010

Bikini Statistics: aka Jesse Livermore gives me a good reason to post another installment


When rats are fed GMO food, do they show signs of organ damage? You get your treatment group, your control group, your outcome measures, and you do an unpaired t-test between the two. These guys use "nonparametric methods," "Principle Component Analysis," and a bunch of other unnecessarily advanced statistics. Basically, they tried doing it the right way, failed to find what they were looking for, and pulled out the fancy (wrong) techniques to get the result they wanted.

Photo link here. Previous BS post here.
Statistics are like a bikini. What they present is suggestive, but what they conceal is vital--Aaron Levenstein

No comments:

Post a Comment