Re: Theory question on kolmogorov-smirnov

From: Glen (glenbarnett_at_geocities.com)
Date: 10/27/04 

Date: 26 Oct 2004 23:09:16 -0700

AJ <99atlantic@gmail.com> wrote in message news:<ggzfd.38382$Pl.31116@pd7tw1no>...
> For the normal_CDF(x), is that normal using mean 0, variance of 1, or using
> the paramaters from the list that comprises the x's. so mean(x), stdev(x) ?

The K-S is a test for a completely specified distribution - so you use
the mean and variance given in the null hypothesis.

If you want to do it for a general normal (i.e. unspecified
parameters), the distribution of the KS test statistic is no longer
appropriate. The corresponding "fit the parameters" version of the K-S
for the normal is called the Lilliefors test.

Glen