[R-lang] Re: is it possible to use lmer function with a polytomous response variable?

[Roger Levy]

Just a follow-up on this: it is not terribly hard to do Bayesian  
multinomial logit models for unordered polytomous response variables  
using JAGS and the R package rjags, once you get used to JAGS.  I have  
some material covering this ground in the following draft book chapter:

   http://idiom.ucsd.edu/~rlevy/textbook/chapter_8.pdf

and I can add more information if it'd be useful.  There are some  
issues that need to be worked out to deal with random slopes and  
interactions, but if you're OK with random intercepts, then it's not  
too bad at all.

Roger

P.S. the close timing of the two posts is because both posts had been  
sitting around in the moderator-must-clear bin for a few days :-)

On Apr 9, 2010, at 12:26 PM, T. Florian Jaeger wrote:

> Dear Oezge,
>
> interesting timing, since Roger just posted something about ordered  
> response variables. So far lmer does not deal with the type of model  
> you need. If you can reasonably assume that your four outcomes are  
> ordered in some way, you can use the package that Roger just posted  
> (see http://www.cran.r-project.org/web/packages/ordinal// 
> ordinal.pdf). If you cannot assume order AND need random effects,  
> thinks get considerably more complicated.
>
> Austin Frank posted about multilevel multinomial regression models  
> for unordered responses on our lab blog: http://hlplab.wordpress.com/2009/05/07/multinomial-random-effects-models-in-r/ 
> . There have been a lot of updates over the months though (the post  
> is from 2007). Maybe he'll read this and will feel sufficiently  
> nudged to post an update (there's now a couple of useful packages  
> for what you need to do).
>
> For an example, you can look at Tily and Piantadosi's 2009 cogsci  
> paper (http://www.stanford.edu/~hjt/tily09cogsci.pdf).
>
> HTH,
>
> Florian
>
> On Thu, Apr 8, 2010 at 9:19 PM, <gurcanli@cogsci.jhu.edu> wrote:
> Dear R-lang-ers,
>
> In my data I have a polytomous response variable which has 4 levels (4
> different verb categories). I want to analyze the data by using lmer
> function with fixed and random effects. Is it possible? If not, what  
> other
> package can I use? Since I am new to this type of analysis, I would
> appreciate if you could give the details of the R-code.
>
> Thanks,
> Özge Gürcanl&#305;
>
>

--

Roger Levy                      Email: rlevy@ling.ucsd.edu
Assistant Professor             Phone: 858-534-7219
Department of Linguistics       Fax:   858-534-4789
UC San Diego                    Web:   http://ling.ucsd.edu/~rlevy