[R] Regression of categorical data

Achim Zeileis Achim.Zeileis at uibk.ac.at
Wed Aug 14 12:20:28 CEST 2013


On Tue, 13 Aug 2013, Walter Anderson wrote:

> I have a set of survey data where I have answers to identify preference of 
> three categories using three questions
>
> 1) a or b?
> 2) b or c?
> 3) a or c?
>
> and want to obtain weights for each of the preferences
>
> something like X(a) + Y(b) + Z(c) = 100%

You could use a Bradley-Terry model for paired comparisons. This estimates 
a so-called "worth parameter" for each alternative/object (a, b, c) and 
these are typically scaled to sum to 1.

A wide collection of Bradley-Terry models including regressors is provided 
in "BradleyTerry2", see http://www.jstatsoft.org/v48/i09/.

Basic Bradley-Terry models as well as other extended models for 
preferences are provided in "prefmod", see 
http://www.jstatsoft.org/v48/i10/.

Regression trees based on Bradley-Terry models are available in 
"psychotree", see example("bttree", package = "psychotree") and the 
references on the corresponding manual page.

Finally, the "eba" package implements further extensions of the 
Bradley-Terry model.

hth,
Z

> I am at a loss how how to calculate this from the data.  Any help would be 
> appreciated!
>
> ______________________________________________
> R-help at r-project.org 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.
>



More information about the R-help mailing list