Date:         Fri, 3 Nov 2000 08:38:05 -0500
Reply-To:     "C. Mitchell Dayton" <cd4@UMAIL.UMD.EDU>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "C. Mitchell Dayton" <cd4@UMAIL.UMD.EDU>
Subject:      Re: partial correlations
Comments: To: Friedrich Funke <Friedrich.Funke@RZ.UNI-JENA.DE>
In-Reply-To:  <NEBBIIADMENHGLKMAJFLMEIPCDAA.sff@uni-jena.de>
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII

        I would use the Hotelling t test (Annals of
Mathematical Statistics, 1940, volume 11, p 271-283) and
adjust the degrees of freedom by 1 (i.e., n-4 rather than
n-3). On the off chance that you don't have the 1940 volume
of the Annals, a more "modern" reference is H. Walker & J.
Lev, Statistical Inference, Holt, 1953. Also, you can find
the formula (in a PDF file) on my course web site:
http://www.education.umd.edu/EDMS/EDMS651Notices.html
Perhaps someone can provide more recent references.

On Fri, 3 Nov 2000 11:00:06 +0100 Friedrich Funke
<Friedrich.Funke@RZ.UNI-JENA.DE> wrote:

> hi,
> i want to test the difference of two partial correlations. the "dependent"
> variable is the same and the controled variable is the same as well. any
> hints?
>
> r(ab.c)    vs    r(db.c)
> thanx in advance,
> fritz.
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> dipl.psych. friedrich funke
>
> friedrich-schiller-universität jena
> institut für psychologie
> methodenlehre und evaluationsforschung
> steiger 3 haus 1
> d-07743 jena
>
> tel.: +49 / (0)3641 / 945-235
> fax.: +49 / (0)3641 / 945-232
> http://www.uni-jena.de/~sff
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

----------------------
C. Mitchell Dayton, Professor
Department of Measurement & Statistics
           and
Joint Program in Survey Methodology
cd4@umail.umd.edu