[R] enhanced question / standardized coefficients

[Michael Dewey]

At 15:51 07/02/2007, Brian S Cade wrote:
>There was a nice paper in The American Statistician by Johan Bring (1994.
>How to standardize regression coefficients.  The American Statistician
>48(3):209-213) pointing out that comparing ratios of t-test statistic
>values (for null hypothesis that parameter = 0) is equivalent to comparing
>ratios of standardized coefficients where standardization is based on the
>partial (conditional) standard deviations of the parameter estimates.  And
>this is equivalent to thinking about the incremental improvement in
>R-squared that is obtained by including a variable in the regression model
>after all others are already in the model.   It would seem possible to
>extend this idea to categorical factor variables with more than 2 levels
>(>1 indicator variable), given the relation between an F and t-test
>statistic.

You may also be interested in
http://www.tfh-berlin.de/~groemp/rpack.html
As well as her package relaimpo this also links to her article in JSS 
which I found thought provoking.


>Any way something to think about, though there are no doubt still
>limitations in trying to equate effects of variables measured on disparate
>scales.
>
>Brian
>
>Brian S. Cade
>
>U. S. Geological Survey
>Fort Collins Science Center
>2150 Centre Ave., Bldg. C
>Fort Collins, CO  80526-8818
>
>email:  brian_cade at usgs.gov
>tel:  970 226-9326
>
>
>
>"John Fox" <jfox at mcmaster.ca>
>Sent by: r-help-bounces at stat.math.ethz.ch
>02/07/2007 07:49 AM
>
>To
>"'Simon P. Kempf'" <simon.kempf at web.de>
>cc
>r-help at stat.math.ethz.ch
>Subject
>Re: [R] enhanced question / standardized coefficients
>
>
>
>
>
>
>Dear Simon,
>
>In my opinion, standardized coefficients only offer the illusion of
>comparison for quantitative explanatory variables, since there's no deep
>reason that the standard deviation of one variable has the same meaning as
>the standard deviation of another. Indeed, if the variables are in the
>same
>units of measurement in the first place, permitting direct comparison of
>unstandardized coefficients, then separate standardization of each X is
>like
>using a rubber ruler.
>
>That said, as you point out, it makes no sense to standardize the dummy
>regressors for a factor, so you can just standardize the quantitative
>variables (Y and X's) in the regression equation.
>
>I hope that this helps,
>  John
>
>--------------------------------
>John Fox
>Department of Sociology
>McMaster University
>Hamilton, Ontario
>Canada L8S 4M4
>905-525-9140x23604
>http://socserv.mcmaster.ca/jfox
>--------------------------------
>
> > -----Original Message-----
> > From: r-help-bounces at stat.math.ethz.ch
> > [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Simon P. Kempf
> > Sent: Wednesday, February 07, 2007 9:27 AM
> > To: r-help at stat.math.ethz.ch
> > Subject: [R] enhanced question / standardized coefficients
> >
> > Hello,
> >
> >
> >
> > I would like to repost the question of Joerg:
> >
> >
> >
> > Hello everybody,
> >
> > a question that connect to the question of Frederik Karlsons
> > about 'how to stand. betas'
> > With the stand. betas i can compare the influence of the
> > different explaning variables. What do i with the betas of
> > factors? I can't use the solution of JohnFox, because there
> > is no sd of an factor. How can i compare the influence of the
> > factor with the influence of the numeric variables?
> >
> > I got the same problem. In my regression equation there are
> > several categorical variables and  I would like to compute
> > the standard coefficients. How can I do this?
> >
> >
> >
> > Simon
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >                [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-help at stat.math.ethz.ch mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide
> > http://www.R-project.org/posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
>
>______________________________________________
>R-help at stat.math.ethz.ch mailing list
>https://stat.ethz.ch/mailman/listinfo/r-help
>PLEASE do read the posting guide
>http://www.R-project.org/posting-guide.html
>and provide commented, minimal, self-contained, reproducible code.
>
>
>         [[alternative HTML version deleted]]

Michael Dewey
http://www.aghmed.fsnet.co.uk