[R] anova(lme.model)

[Bert Gunter]

Sounds to me like you should really be seeking help from your local
statistician, not this list. What you request probably cannot be done.

What is wrong with what you get from lme, whose results seem fairly
clear whether the P values are accurate or not?

Cheers,
Bert





On Sat, Nov 6, 2010 at 4:04 AM, "Sibylle Stöckli"
<sibylle.stoeckli at gmx.ch> wrote:
> Dear R users
>
> Topic: Linear effect model fitting using the nlme package (recomended by Pinheiro et al. 2008 for unbalanced data set).
>
> The R help provides much info about the controversy to use the anova(lme.model) function to present numerator df and F values. Additionally different p-values calculated by lme and anova are reported. However, I come across the same problem, and I would very much appreciate some R help to fit an anova function to get similar p-values compared to the lme function and additionally to provide corresponding F-values. I tried to use contrasts and to deal with the ‚unbalanced data set’.
>
> Thanks
> Sibylle
>
>> Kaltenborn<-read.table("Kaltenborn_YEARS.txt", na.strings="*", header=TRUE)
>>
>>
>> library(nlme)
>
>> model5c<-lme(asin(sqrt(PropMortality))~Diversity+ Management+Species+Height+Height*Diversity, data=Kaltenborn, random=~1|Plot/SubPlot, na.action=na.omit, weights=varPower(form=~Diversity), subset=Kaltenborn$ADDspecies!=1, method="ML")
>
>> summary(model5c)
> Linear mixed-effects model fit by maximum likelihood
>  Data: Kaltenborn
>  Subset: Kaltenborn$ADDspecies != 1
>        AIC       BIC   logLik
>  -249.3509 -205.4723 137.6755
>
> Random effects:
>  Formula: ~1 | Plot
>        (Intercept)
> StdDev:  0.06162279
>
>  Formula: ~1 | SubPlot %in% Plot
>        (Intercept)   Residual
> StdDev:  0.03942785 0.05946185
>
> Variance function:
>  Structure: Power of variance covariate
>  Formula: ~Diversity
>  Parameter estimates:
>    power
> 0.7302087
> Fixed effects: asin(sqrt(PropMortality)) ~ Diversity + Management + Species +      Height + Height * Diversity
>                      Value  Std.Error  DF   t-value p-value
> (Intercept)       0.5422893 0.05923691 163  9.154585  0.0000
> Diversity        -0.0734688 0.02333159  14 -3.148896  0.0071
> Managementm+      0.0217734 0.02283375  30  0.953562  0.3479
> Managementu      -0.0557160 0.02286694  30 -2.436532  0.0210
> SpeciesPab       -0.2058763 0.02763737 163 -7.449198  0.0000
> SpeciesPm         0.0308005 0.02827782 163  1.089210  0.2777
> SpeciesQp         0.0968051 0.02689327 163  3.599602  0.0004
> Height           -0.0017579 0.00031667 163 -5.551251  0.0000
> Diversity:Height  0.0005122 0.00014443 163  3.546270  0.0005
>  Correlation:
>                 (Intr) Dvrsty Mngmn+ Mngmnt SpcsPb SpcsPm SpcsQp Height
> Diversity        -0.867
> Managementm+     -0.173 -0.019
> Managementu      -0.206  0.005  0.499
> SpeciesPab       -0.253  0.085  0.000  0.035
> SpeciesPm        -0.239  0.058  0.001  0.064  0.521
> SpeciesQp        -0.250  0.041 -0.001  0.032  0.502  0.506
> Height           -0.518  0.532 -0.037 -0.004  0.038  0.004  0.033
> Diversity:Height  0.492 -0.581  0.031 -0.008 -0.149 -0.099 -0.069 -0.904
>
> Standardized Within-Group Residuals:
>        Min          Q1         Med          Q3         Max
> -2.99290873 -0.60522612 -0.05756772  0.62163049  2.80811502
>
> Number of Observations: 216
> Number of Groups:
>             Plot SubPlot %in% Plot
>               16                48
>
>> anova(model5c)
>                 numDF denDF   F-value p-value
> (Intercept)          1   163 244.67887  <.0001
> Diversity            1    14   1.53025  0.2364
> Management           2    30   6.01972  0.0063
> Species              3   163  51.86699  <.0001
> Height               1   163  30.08090  <.0001
> Diversity:Height     1   163  12.57603  0.0005
>>
>
> --
>
> ______________________________________________
> 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.
>



-- 
Bert Gunter
Genentech Nonclinical Biostatistics