Re: [Fwd: nobel]

["Dr. Bruce R. McFarling" <ecbm@xxxxxxxxxxxxxxxxxxx>]

On Thu, 10 Oct 2002 11:24:19 -0400, Sven R Larson
<slarson@xxxxxxxxxxxx> wrote:

>Paul has rightfully pointed out that Enron shows us how vulnerable our
>sources of statistics can be. I definitely agree, but instead of
>dismissing empirical research I prefer to take the same approach to
>statistics as Churchill did to democracy: it is rude and crude and has
>appalling shortcomings, but it's the best we have.

What's appalling is when the statistical techniques are
combined with unsubstantiated a priori's to arrive at
econometrics.  If I had a nickle for every time an
econometric analysis rested on failing to reject with
confidence levels that are themselves based on the premise
that the functional form of the regression was correct,
without any critical argument to establish that the
relationship is functional at all, I would already be set
up on some tropical island somewhere (self-funding my
research, of course).  In a partnership between a
statistician and a scientist, its the scientist's
reputation at stake if a garbage empirical framework
leads to GIGO statistical results.  In economics, it
sometimes seems that a new wrinkle that supports a new
family of GIGO statistical results is cause for
celebration.