[S] Replacement for ks.gof

Steve Craighead (USNATL4T@ibmmail.com)
Thu, 27 Aug 1998 08:32:13 EDT


Is there a better hypothesis test for whether two separate samples are from the
same distribution than the ks.gof test?

The test needs to address these issues:

1. Is the support of the two sample distributions nearly identical?
2. The ks.gof test examines the probability of the largest
deviation between the two sample distributions. I need a method
that would examine the cumulative deviations between the two
sample distributions.
3. The test could be used on truncated distributions that would allow
one to examine similarity between various subregions of the
sample distributions.
4. The test should be nonparameteric, or at least accommodate
nonparameteric analysis.
5. The sample sizes of the sample distributions can be different.
6. It would be nice if it was coded in S or S+.
7. You don't have to be a rocket scientist to understand how to use it.
(It is okay for the test to have been designed by a rocket scientist)
8. There is a written reference justifying the test.

Steve Craighead Email: usnatL4t@ibmmail.com
Nationwide Life Ins. Co. or craighs@nationwide.com
One Nationwide Plaza 1-27-01 Fax: (614) 249-0725
Columbus, Ohio USA 43215-2220 Phone:(614) 249-8775
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news