Date:         Mon, 14 Jul 2003 15:14:18 -0400
Reply-To:     Jay Weedon <jweedon@EARTHLINK.NET>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Jay Weedon <jweedon@EARTHLINK.NET>
Organization: http://extra.newsguy.com
Subject:      Re: Sign test and SignRank test
Content-Type: text/plain; charset=us-ascii

On 14 Jul 2003 11:01:34 -0700, accpwong@inet.polyu.edu.hk (Antonio)
wrote:

>I have a stupid question. What is the difference between sign test and
>Signrank test? Please advice, Thanks. Based on my understanding,
>
>sign test is similar to one sample t-test
>sign rank test is similar to paired sample t-test
>
>Am I correct?

Either test could both be used in either situation (as could the
t-test: the paired t-test is simply a one-sample test applied to
difference scores). The principal difference is that the signed-rank
test uses information on the relative magnitude of differences,
whereas the sign test weights all differences the same if they are in
the same direction. In other words, the signed-rank test uses more
information about the data than the other, and is more powerful in
many cases. It is however somewhat vulnerable to failure when the
distribution of (difference) scores shows significant skew, in which
case the sign test may be preferred.

JW