[R-sig-ME] Multinomial mixed-effect model in nnet

[Joehanes, Roby (NIH/NHLBI) [F]]

Hi Jeffrey:

Your concern about lack of multiple observations on about half of the
samples is indeed valid, but nnet is not a mixed model package. I think you
should clarify what you would like to accomplish with your data and what
specifically you want out of the mixed model to your analysis. If you want
to establish association between "summercategory" and "n.year" after
accounting for the elkid, then you can apply mixed model analysis. You may
want to look into glmmADMB package for doing so.

Hope this helps,
Roby

On 12/10/12 7:00 PM, "Jeffrey Warren" <jeffrey_warren at fws.gov> wrote:

> Hello,
> 
> 
> 
> We¹re exploring temporal change in summer range selection in radio-collared
> elk during a 3 decade period. Elk are marked on the wintering grounds and
> tracked back to one of 5 summer ranges. There are 180 individuals; most (n
> = 121) individuals were monitored >=2 years, resulting in a total of 546
> summer range observations from the 180 elk. We¹ve run multinomial
> regression models with year as a predictor in packages Œmlogit¹ and Œnnet¹
> using only the last summer range choice for each individual (n = 180):
...
[snip]