home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.math.stat:2726 bit.listserv.stat-l:2329
- Newsgroups: sci.math.stat,bit.listserv.stat-l
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!cs.utexas.edu!torn!watserv2.uwaterloo.ca!watserv1!phatak
- From: phatak@watserv1.uwaterloo.ca (Aloke Phatak)
- Subject: Estimating bias
- Message-ID: <C0I8pr.9I5@watserv1.uwaterloo.ca>
- Summary: How to estimate bias in biased regression?
- Keywords: bias, biased regression
- Organization: University of Waterloo, Dept. of Chemical Engineering
- Date: Thu, 7 Jan 1993 22:20:15 GMT
- Lines: 27
-
-
- Just wondering - in using a biased estimator in the usual linear
- regression model given by
-
- y = Xb + e
-
- it's often possible to come up with an expression (as in principal
- component regression, for example) for the bias of the estimator.
- Unfortunately, it usually depends on b, the unknown parameter of
- interest. My question is this: are there any methods to estimate
- the bias given an analytical expression for it, even if that
- expression depends on b? I came across an article in the latest
- Technometrics in which the authors show how bootstrapping can be
- used to estimate bias, but I'm looking for simple methods that are
- not computationally intensive.
-
- Any suggestions or pointers to references would be appreciated.
-
- Aloke Phatak
-
- __
- +----------------------------------------------------------------------------+
- | Aloke Phatak |
- | Department of Chemical Engineering |
- | University of Waterloo phatak@watserv1.uwaterloo.ca |
- | Waterloo, Ontario CANADA N2L 3G1 (519) 885-1211, ext. 6982 |
- +----------------------------------------------------------------------------+
-