Editore"s Note
Tilting at Windmills

Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

April 26, 2003
By: Kevin Drum

SHORTER TIM LAMBERT....Say it after me: "John Lott is a hack, John Lott is a hack....."

Here's the nickel version of today's installment: a couple of guys have written an article for the Stanford Law Review in which they say that Lott's statistics are all wet. Now, econometricians are forever telling other econometricians that their models are no good because they have failed to take into account some obscure variable or another, so us laymen could be excused for nodding off at this point and just waiting a couple of years to see if a scholarly consensus emerges.

But there's more, and this doesn't take an advanced degree to understand: basically, the SLR article took Lott's model and applied it to more recent data. Result: nada. So Lott and a couple of other guys responded by doing their own analysis, and their conclusion is that the model does work. Unfortunately, it's not just a matter of competing analyses: Lott and his partners actually miscoded the data, and by coincidence it was systematically miscoded to favor their hypothesis every time. Can you imagine?

Lott has been using this miscoded data for a while, but it turns out that he took his name off the SLR response before it was published, so presumably he knew what was going on and didn't want to put his name to it in a journal article. Or maybe he was just covering his ass. Or maybe the sheer force of the SLR argument has actually convinced him to change his mind.

(Ha ha, just kidding on that last one, folks. I do slay myself sometimes, yes I do.)

As usual, Tim Lambert has all the gory details.

Kevin Drum 6:10 PM Permalink | Trackbacks

Bookmark and Share
 
Comments




 

 

Read Jonathan Rowe remembrance and articles
Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

Advertise in WM



buy from Amazon and
support the Monthly