
From
http://listserv.uga.edu/cgi-bin/wa?A2=ind0111&L=spssx-l&F=&S=&P=23150


Date:         Tue, 20 Nov 2001 09:58:02 +0100
Reply-To:     Asesora Bioestadstica
              <bioestadistica@ERESMAS.NET>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Asesora Bioestadstica
              <bioestadistica@ERESMAS.NET>
Subject:      Re: asymptotic significance
Content-Type: text/plain; charset=us-ascii

Dear Rose

"Asymptotic" means that the p-value aproaches the real value as sample size
increases. For Kruskal-Wallis test, the asymptotic value is valid if sample
size is above 5 in all groups, so in your case it is OK. For Mann-Whitney's U
test, the asymptotic value is valid if both samples involved have sizes above
20 cases (it is not the situation here). One problem with the exact p-value is
that  it is not corrected for ties (several individuals with the same value)
so, if you have a lot of "tied" data, the exact p value will under-estimate the
real significance (will give higher values that the real ones). By the way, if
you are using Mann-Whitney's U test as a multiple comparison procedure, you
should consider adjusting p values with Bonferroni, Sidak or other methods. I
recommend you the program adjustp.exe from the freeware PEPI 4.0 statistical
package, available for download at:

http://www.sagebrushpress.com

The manual for the 3.0 version is available here:

http://www.brixtonbooks.demon.co.uk

Best regards

Marta Garcia-Granero

Rose Martini ha escrito:

> Dear list members,
>
> I am presently using SPSS for nonparametric statistics (my N is low: 9, 8,
> 5 per group and my distribution in non-nomal). I am specifically using the
> Kurskal-Wallis test followed by a Mann-Whitney U. I have a couple of
> questions re: the SPSS output:
>
> 1.My output reports  Asymp sig. for the Kruskal Wallis. Can someone explain
> to be what an asymptotic significance is?
>
> 2. The Mann-Whitney output reports an Asymp. Sig. (2-tailed) and a Exact
> Sig. (2*(1-tailed Sig.)). Can someone explain to me what the difference
> between these two are?
>
> Thank you,
>
> ..Rose..