[19186] in s-news-athena
[S] multivariate regression models
daemon@ATHENA.MIT.EDU (In-Sun Nam)
Sat Sep 11 04:38:16 1999
Date: Sat, 11 Sep 1999 18:32:54 +1000
From: In-Sun Nam <i.nam@fsc.qut.edu.au>
To: S Mail Server <s-news@wubios.wustl.edu>
Message-Id: <37DA13B6.2466B83D@fsc.qut.edu.au>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Dear all,
Would anyone suggest general multivariate models for two or more
strongly correlated (paired) dependent variables,
- when they may have different probability distribution (not
necessarily though),
- the nature of the correlation may be more complex (than a linear
correlation), and
- they may have a common set of explanatory variables, while each
response variable may have its own independent covariates (not
necessarily though).
Additionally, if the two responses are not paired initially, but there
is/are a common covariate or two, is it a common practice to match the
responses according to those common covariates?
Or does anyone can comment on the curds and whey method (Breiman and
Friedman), related to the above issues?
I will greatly appreciate any directly relevant comments, suggestions,
even some references.
Sincerely,
In-Sun
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
