On Fri, 06 Dec 2002 17:05:50 +0000, Jerry Dallal wrote:

> Often, K-S like tests are of limited value because they are most
> sensitive to departures from normality in the center of the
> distribution.  When normality is an issue, it usually involves
> behavior in the tails, which is where KS-like tests are least
> sensitive.

It hasn't been mentioned in the thread, I think. But I know that's true,
and that the Anderson-Darling test (which was mentioned) is somewhat
better in that respect.
Thanks for the remark though!

Koen
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to