omitted variables in logistic regression

Article View: bit.listserv.stat-l

Article #1000

Back to bit.listserv.stat-l Previous Next

#1000

From: Peter A. Lachenbruch
Date: Tue, 27 Jul 1993 23:01

5 lines
1045 bytes

At convergence, the coefficients are the solutions to equations of the form
(X'WX)**-1X'WY, so one might consider looking at the transformed variables
Z=W**(1/2)Y and U=W**(1/2)X and carrying out the same analysis that you did
for linear regression. I believe the analogy carries through exactly.
Tony Lachenbruch

Technical Details

Message-ID: <STAT-L%93072719011553@VM1.MCGILL.CA>

Path:

arc1-ca.pugleaf.net!gmd.de!xlink.net!howland.reston.ans.net!usenet.ins.cwru.edu!usenet.mcs.kent.edu!news.ysu.edu!psuvm!auvm!UCLAMVS.BITNET!IARSPAL

Show Full Headers

Complete Article Headers

X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: f5efb,846ecee2d250ce48,start
X-Google-Attributes: gidf5efb,public
X-Google-ArrivalTime: 1993-07-28 04:50:21 PST
Path: arc1-ca.pugleaf.net!gmd.de!xlink.net!howland.reston.ans.net!usenet.ins.cwru.edu!usenet.mcs.kent.edu!news.ysu.edu!psuvm!auvm!UCLAMVS.BITNET!IARSPAL
Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
Newsgroups: bit.listserv.stat-l
Return-Path: <@AUVM.AMERICAN.EDU,@VTBIT.CC.VT.EDU:owner-stat-l@VM1.MCGILL.CA>
Message-ID: <STAT-L%93072719011553@VM1.MCGILL.CA>
Date: Tue, 27 Jul 1993 16:01:00 PDT
Sender: Stat-l Discussion List <STAT-L@MCGILL1.BITNET>
From: "Peter A. Lachenbruch" <IARSPAL@UCLAMVS.BITNET>
Subject: omitted variables in logistic regression
Lines: 4