[R] effect.clm

[Fox, John]

Dear David,

Effect.clm() isn't properly passing optional arguments like xlevels. The problem occurs in some other Effect() and effect() methods as well, and the fix seems simple. 

I've corrected these functions in the development version of effects package on R-Forge. We'll need to test more carefully before sending the revised package to CRAN. In the meantime, you should be able to install effects from R-Forge via install.packages("effects", repos="http://R-Forge.R-project.org"), after waiting a day or so for R-Forge to build the package.

Thanks for reporting this bug.

John

-----------------------------
John Fox, Professor
McMaster University
Hamilton, Ontario
Canada L8S 4M4
Web: socserv.mcmaster.ca/jfox



> -----Original Message-----
> From: R-help [mailto:r-help-bounces at r-project.org] On Behalf Of David Barron
> Sent: August 16, 2016 8:16 AM
> To: r-help at r-project.org
> Subject: [R] effect.clm
> 
> I'm having a problem using the xlevels option with effect (or Effect) with
> objects of class clm, which seems to have no impact on the output (whereas it
> does for models of class polr).  Can anyone suggest how to fix this?
> 
> Thanks,
> David
> 
> Here's an example:
> 
> mod.wvs <- MASS::polr(poverty ~ gender + religion + degree +
> country*poly(age,3), data=WVS)
> 
>  Effect('age', mod.wvs, xlevels = list(age = 1:10))
> 
> Re-fitting to get Hessian
> 
> 
> age effect (probability) for Too Little
> age
>         1         2         3         4         5         6         7
>   8         9        10
> 0.8403661 0.8270171 0.8133710 0.7995001 0.7854780 0.7713778 0.7572714
> 0.7432277 0.7293120 0.7155850
> 
> age effect (probability) for About Right age
>         1         2         3         4         5         6         7
>   8         9        10
> 0.1295679 0.1399772 0.1505377 0.1611853 0.1718564 0.1824888 0.1930231
> 0.2034039 0.2135803 0.2235068
> 
> age effect (probability) for Too Much
> age
>          1          2          3          4          5          6
>  7          8          9         10
> 0.03006598 0.03300572 0.03609134 0.03931456 0.04266563 0.04613341
> 0.04970548 0.05336837 0.05710766 0.06090822
> 
> But with clm:
> 
> +                 data=WVS)
> > Effect('age', mod.wvs, xlevels = list(age = 1:10))
> 
> age effect (probability) for Too Little
> age
>        20        30        40        50        60        70        80
>  90
> 0.5980510 0.5268304 0.4936603 0.4803375 0.4674829 0.4359217 0.3684641
> 0.2589983
> 
> age effect (probability) for About Right age
>        20        30        40        50        60        70        80
>  90
> 0.3031130 0.3453414 0.3629620 0.3696094 0.3757686 0.3897338 0.4129734
> 0.4227230
> 
> age effect (probability) for Too Much
> age
>         20         30         40         50         60         70
> 80         90
> 0.09883598 0.12782815 0.14337772 0.15005316 0.15674848 0.17434449
> 0.21856247 0.31827868
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> 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.