According to Dean what he did was correcting one way in the first experiment, then correcting another way in the second, and a third way in the third time, then retroactively applied his goofy (my words) methods back to the previous sets of data, essentially poisoning the data.
By his own admission on the skeptiko podcast he did something which anyone with statistical knowledge knows absolutely kills the credibility of his analysis.
Last edited by Opcnup; 08-08-2007 at 02:17 AM..
|