home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!IUBACS.BITNET!SMCNULTY
- Original_To: EDSTATL
- Message-ID: <EDSTAT-L%92111009192066@NCSUVM.CC.NCSU.EDU>
- Newsgroups: bit.listserv.edstat-l
- Date: Tue, 10 Nov 1992 09:15:00 EST
- Reply-To: SMCNULTY@IUBACS.BITNET
- Sender: "Statistics Education Discussion" <EDSTAT-L@NCSUVM.BITNET>
- From: SMCNULTY@IUBACS.BITNET
- Subject: Averaging beta weights
- Lines: 18
-
- I posted this question to the list a week or so ago, but some hardware
- problems on this end lead me to believe it never made it. At any rate,
- the recent questions about the Fisher r to z transformation and confidence
- intervals for B's suggest now would be a good time to ask again:
-
- Is any transformation (analogous to the Fisher r to z) necessary before
- averaging standardized regression weights? I have a dataset in which
- we've computed a within-subjects regression weight for each case
- (regressing self-report's of several traits on ratings by another person
- and an earlier set of self-ratings) and we'd like to be able to say
- something about the central tendency of the resulting distribution of
- beta weights. Can we simply compute the mean? If a transformation is
- necessary, what form would it take? Thanks in advance for your advice.
-
- Shawn E. McNulty
-
- bitnet: smcnulty@iubacs
- internet: smcnulty@ucs.indiana.edu
-