home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math.stat
- Path: sparky!uunet!think.com!ames!news.hawaii.edu!pmk
- From: pmk@busadm1.cba.hawaii.edu (PAUL KAZENSKI)
- Subject: 2 vs 1 system of linear equations
- Message-ID: <pmk.37@busadm1.cba.hawaii.edu>
- Sender: root@news.Hawaii.Edu (News Service)
- Nntp-Posting-Host: bus85.cba.hawaii.edu
- Organization: College Business Administration, University of Hawaii
- Date: Wed, 16 Dec 1992 03:36:05 GMT
- Lines: 18
-
- I've received a question I cannot provide a good answer to:
-
- There are 2 systems of equations comprised of 3 equations each that
- have been estimated recursively with "good" results; it is being
- suggested that these be combined into a single system and estimated
- simultaneously in order to capture additional information from
- (possibly) correleated error terms between the two systems.
-
- Since the task is non-trivial, the questioner asks if there are tests
- which can be performed to determine **if** it is necessary to combine
- systems. He has tried correlation analysis of the residuals--which
- indicates no correlation--and regressed error terms against all others
- with the same results.
-
- If anyone knows of tests/references, I would appreciate either your post or
- e-mail. I will pass it on and summarize to the net.
-
- Mahalo Nui Loa!
-