Endless Quibblings of our Times, I. I see that the Lancet Article That Will Not Die is back under discussion again. My own minor contribution to all this was here, wherein I tried unsuccessfully to get the tag-team duo of Daniel Davies and Tim Lambert to justify the way in which they were reporting the summary statistics of the piece. (At least, I didn't think they justified it. Discussion starts down towards the bottom of the comment thread.)

Obviously, the proper way to deal with this is for other parties to do follow-up studies - only then will we be able to assess how accurate the initial study was.

My own guess is, "not very". But we shall see.

(By the way, is it true what schoolchildren say -- that if you go into the bathroom, turn off all the lights, look into the mirror and chant "Lancet" over and over, Tim Lambert will appear?)


Posted by David Fleck at 27 March 2005 11:45 AM
Comments

A lot of this argument comes down to the statement: "How dare you challenge my data." Which means one should challenge the data. There's also the essential circularity of asserting that the results are too important to throw out bad data, and of replying to a challenge of the data with an assertion that the challenger doesn't understand sophisticated analysis techniques.


Posted by: Jonathan on March 27, 2005 11:31 PM

Yes, this paper seems to have become quite the nexus of bad statistical thought and practice. I'm glad there are those (e.g., Shannon) willing to take on the Sisyphean task of pointing out its problems in the face of those who have been relentlessly misrepresenting it.


Posted by: David Fleck on March 29, 2005 06:25 AM

Oh no -- you have raised up the Lambert beast!

Posted by: Andrea Harris on March 29, 2005 06:33 AM

Post a comment
Name:


Email Address:


URL:


Comments:


Remember info?