home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!gatech!paladin.american.edu!auvm!VTVM1.BITNET!HARVEYRJ
- Message-ID: <STAT-L%92081416395359@VM1.MCGILL.CA>
- Newsgroups: bit.listserv.stat-l
- Date: Fri, 14 Aug 1992 16:34:01 EDT
- Sender: "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
- From: "r. j. harvey" <HARVEYRJ@VTVM1.BITNET>
- Subject: standard errors of B
- Lines: 14
-
- dear list:
- I would like to find a computational formula for computing the standard
- error of multiple regression weights WITHOUT having access to the inverse
- of the correlation matrix, AND without having to solve for the R-square
- computed without the given predictor. it seems like in the "old days" there
- must have been a way to do this; all current texts I can find make reference
- only to methods based on the inverse of R and/or on knowing the semipartial
- correlation (or other statistic based on the R-square with and without that
- predictor in the model).
- anyone have any ideas?
-
- r. j. harvey
- psychology
- virginia tech internet: harveyrj @ vtvm1.cc.vt.edu
-