Date:         Mon, 22 May 2006 10:49:52 +0200
Reply-To:     Marta García-Granero
              <biostatistics@terra.es>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         Marta García-Granero
              <biostatistics@terra.es>
Organization: Asesoría Bioestadística
Subject:      Re: Degrees of Freedom in Excel
In-Reply-To:  <329A68716B57D54E8D39FD3F8A4A84DF06276544@um-mail0136.unimaas.nl>
Content-Type: text/plain; charset=ISO-8859-15

Hi Gjalt-Jorn,

PGP> I am trying to make an Excel file which replicates SPSS' independent
PGP> t-test procedures using means, standard deviations and sample sizes.
PGP> I've almost got it, but for some reason I seem unable to implement the
PGP> calculation for the degrees of freedom (Df) that SPSS uses when the
PGP> variances of the two samples differ significantly.

I wrote SPSS code for that (available at:
http://www.spsstools.net/Syntax/T-Test/DoT-TestWithSummaryData.txt )

PGP> Does anybody know what this formula should look like in SPSS?

The formula for Welch DF that I know is this one:

     (S²a/Na + S²b/Nb)²
DF= ----------------------
    (S²a/Na)²   (S²b/Nb)²
    --------- + ---------
     (Na-1)      (Nb-1)

Using it I get the same result (up to 3 decimal places) that the one
provided by T-TEST procedure.

You can find the description of Welch test and the formula here:

http://web.uccs.edu/lbecker/spss80/ttest.htm#5.%20Independent%20Samples:%20Running%20the%20t%20Test

HTH

Marta