[R] help with polytomous logistic regression

[John Fox]

Dear Anne,

There's an anova() method for multinom objects that will allow you to get
likelihood-ratio tests. Likewise, the Anova() function in the car package
has a method for multinom objects.

As to plotting, I have a paper, coauthored with Bob Andersen, on producing
"effect" plots (partial plots) for polytomous logistic regression, including
multinomial and proportional-odds logistic regression. The paper, at
<http://socserv.socsci.mcmaster.ca/jfox/logit-effect-displays.pdf>, has an R
function for computing such effects, including standard errors. There are
examples in the paper as well. Eventually, I intend to incorporate this into
the effects package to make the graphs more or less automatic.

Regards,
 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 Anne
> Sent: Friday, January 07, 2005 8:53 AM
> To: R list
> Subject: [R] help with polytomous logistic regression
> 
> Hi!
> 
> I'm trying to do some ploytomous logistic regression using 
> multinom() in the nnet package, but am a bit confused about 
> interpretation of the results Is it possible to get the 
> following quantities: 
> 
> I:  maximum likelihood estimates to test for fit of model and 
> significance of each predictor
> 
> (I would like to produce a table of the following type)
> Analysis of Variance: MLE    (values are non sensical, I know!)
>                df      chi2           p
> intercept   2       57            .003
> v1            4       89            .876
> v2            2         7            .05
> LR        110     450            0.93
> 
> II:  chi square values and corresponding p-values for each 
> level of the predictor variates (I'm OK for the estimates and 
> se from summary,)
> 
>  
>                level estimate     se        chi2         p   
> intercept   1        5              .9         32          .002
>                 2       6.3            .8         31          .03
> v1             3       89            .876       43          .001
>                 4       65            .05         67          
> 0.001      
>  v2...
> 
> 
> Is there a convenient way to plot the results? (I'd like to 
> display visually the effects of the predictors: any sugestion?)
> 
> OK, I probably miss something here (no experience with (non 
> ordinal) polytomous logistic regression yet!)
> 
> Thank  very much to all
> 
> Anne
> 
> ----------------------------------------------------
> Anne Piotet
> Tel: +41 79 359 83 32 (mobile)
> Email: anne.piotet at m-td.com
> ---------------------------------------------------
> M-TD Modelling and Technology Development PSE-C
> CH-1015 Lausanne
> Switzerland
> Tel: +41 21 693 83 98
> Fax: +41 21 646 41 33
> --------------------------------------------------
>  
> 	[[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