Date:         Thu, 11 Nov 1999 10:36:12 -0600
Reply-To:     "Nichols, David" <nichols@SPSS.COM>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Nichols, David" <nichols@SPSS.COM>
Subject:      Re: searching statistical test
Comments: To: Marek Fuchs <marek.fuchs@KU-EICHSTAETT.DE>

If these are four independent samples on the same variable where the
standard linear model assumptions are plausible, then a one way ANOVA with
four groups can be run, and a contrast of the form 1 -1 -1 1 fitted to give
what you want. This is the same thing as what was elsewhere suggested about
setting it up as a 2x2 and testing for an interaction.

David Nichols
Principal Support Statistician and
Manager of Statistical Support
SPSS Inc.

> -----Original Message-----
> From: Marek Fuchs [mailto:marek.fuchs@KU-EICHSTAETT.DE]
> Sent: Wednesday, November 10, 1999 12:30 PM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: [SPSSX-L] searching statistical test
>
>
> Hello --
>
> I'm looking for a test in order to compare two differences.
>
> I have two independent samples S1 and S2. For each sample I have a
> quantitative measure A. Using t-test I can compute the significance of
> the mean difference DIFF1_2 = mean_A(S1) - mean_A(S2).
>
> In addition, I have two other independent samples S3 and S4. Again, I
> have the measure A and DIFF3_4 = mean_A(S3) - mean_A(S4).
>
> The goal is to determine whether the two differences DIFF1_2
> and DIFF3_4
> are the same or whether they differ significantly. However, I have no
> standard deviation for the differences. What test is
> applicable to this
> situation? How can I estimate the standard deviation?
>
> I appreciate any hint or help.
> Thanks in advance
> Marek Fuchs
>
> Sorry for cross posting
>