Date:         Wed, 24 Feb 1999 17:07:31 -0500
Reply-To:     MARIA_ELENA_GARCIA@WCJ.COM
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         Maria Elena Garcia <MARIA_ELENA_GARCIA@WCJ.COM>
Subject:      Re: Comparisons of frequency distributions for full and cut-d
Comments: To: Stephen Greaves <sgreaves@MINDSPRING.COM>
Content-Type: text/plain; charset=US-ASCII

Hi Stephen:
I think you can use the Kolmogorov Smirnov statistics, it compares the vertical
distances between two empirical distributions so you can use it for deciding
whether they have the same distribution or one is bigger than the other.
About the Wald-Wolfowitz statistic I can't tell you anything.
Marķa Elena.
 
____________________Reply Separator____________________
Subject:    Comparisons of frequency distributions for full and cut-down
Author: Stephen Greaves <sgreaves@MINDSPRING.COM>
Date:       24/02/99 1:16 PM
 
My problem concerns comparing whether two empirical frequency distributions
are significantly different.  The distributions are based on a full dataset
and a subset of that dataset (cut-down data set) so they do not represent
 
independent samples.  I am aware of the various tests for comparing
distributions (Chi-Square, Kolm-Smirnov, Wald-Wolfowitz) but I am not sure
which, if any, are appropriate for this scenario.
 
Does anyone have any comments on the appropriate test in this case?
 
Thanks in advance