Message-ID: <Pine.A41.4.44.0211151059560.32848-100000@homer10.u.washington.edu>
Date: 2002-11-15T19:02:50Z
From: Thomas Lumley
Subject: analysing data with case weights: rejoinder
In-Reply-To: <x2fzu2vf1s.fsf@biostat.ku.dk>
On 15 Nov 2002, Peter Dalgaard BSA wrote:
> Thomas Lumley <tlumley at u.washington.edu> writes:
>
> > On Fri, 15 Nov 2002, bojaniss wrote:
> >
> > > Hello,
> > >
> > > Based on replies (which I'm gratefull for) I guess the only solution
> > > is to create a package, that will contain function to build
> > > lm()-like linear models, and allow weighting cases (as I described
> > > two days ago).
> > >
> > > This would also alow buliding those models for data in aggregated
> > > form without a need for using e.g. expand$grid().
> > >
> >
> > There's no problem with data in aggregated form, only with probability
> > weighting.
>
> Depends on what you mean with "aggregation" doesn't it? If you have
> "replication weights" (counts of data with given value), I think you
> still have problems.
>
There's no problem for categorical data, which is the case where you
most likely to have counts of data with given value.
With a linear model you can aggregate further to get an average value over
each covariate pattern and then have variance weights
-thomas
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._