Date:         Fri, 12 Jun 1998 16:21:37 +1000
Reply-To:     John Reece <reece@RMIT.EDU.AU>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@UGA.CC.UGA.EDU>
From:         John Reece <reece@RMIT.EDU.AU>
Subject:      Re. t-test with covariates?
Content-Type: text/plain; charset="us-ascii"

This problem can be reframed as a within-subjects ANOVA with varying
covariates, which is pretty easy to do with MANOVA (I'm a Mac user, so I
have no knowledge of doing this in GLM). I can't think of any reason why
you would want to approach it using t-tests specifically, and in terms of
decision making and ease, treating it in the way I suggested should get you
what you want.
 
>I'm trying to conduct a paired sample t-test with a twist.  I want to
>adjust each of the two variables in the t-test by another variable before
>conducting the test.  I understand there's no option to do so in the t-test
>command (is there?)...
>
>Specifically I want to test the difference in means between variables A and
>B when variable A is adjusted for the impact of variable A' and variable B
>is adjusted for the impact of variable B'.
>
>My initial thought was to run a linear regression predicting A with A' and
>another predicting B with B'.  I would save the unstandardized residuals
>and test the means of the unstandardized residuals.  Well, that didn't
>work, because the means both ended up being almost zero (theoretical range
>of A and B was 0-10).
>
>It dawned on me that maybe I should run the same regressions without the
>constant in the equation.  I did so, and now I get variables with means of
>less than one, but still greater than zero.  Things still don't seem right.
>
>Does anyone have any suggestions?  Is looking at the unstandardized
>residuals without using a constant correct?  If not, what is the proper
>solution???
+----------------------------------------------------------------+
|John Reece, PhD                                                 |
|Dept. of Psychology & Intellectual Disability Studies           |
|RMIT University                        Phone:   +61.3.9925.7512 |
|PO Box 71                              Fax:     +61.3.9925.7303 |
|Bundoora VIC 3083                      mailto:reece@rmit.edu.au |
|AUSTRALIA       http://www.rmit.edu.au/departments/ps/reece.htm |
+----------------------------------------------------------------+