Date: Wed, 30 Jun 1999 09:24:35 -0500
Reply-To: Tim Dunsworth <Tim.Dunsworth@METROSTATE.EDU>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Tim Dunsworth <Tim.Dunsworth@METROSTATE.EDU>
Subject: Re: Median Test
Content-Type: text/plain; charset=US-ASCII
I assume since it provided both measures that you are comparing just two independent samples? (if there is a generalized Yates correction for table larger than 2 x 2, I am not aware of it) In that case the choice depends on total N (if Yates is needed at all, it is only for "small" N), and on your opinion about the value of the Yates correction. My copy of the base system users manual has a brief discussion of this issue and a couple of references to "the controversy" that you might want to look into. I tend to look at them both but report the uncorrected value. Since Yates is more conservative, if both are significant that tends to confirm the result and make me more confident in it. HTH.
>>> c dommeyer <vcmkt001@CSUN.EDU> 06/29 6:27 PM >>>
When I get the output for median test, two significance values are
reported - one based on the Chi-Square value and one based on Yates'
Continuity Correction. Which significance value should I use and why?