[ADMB Users] Output from glmm.admb

[H. Skaug]

Nina,

Output is supposed to be available at a higher level from
within R.


Here are some examples:

>fit = glmm.admb(y~Base*trt+Age+Visit,random=~Visit,group="subject",data=epil2,family="nbinom")
>fit2 = glmm.admb(y~Base*trt+Age,random=~Visit,group="subject",data=epil2,family="nbinom")

Now:
>anova(fit2,fit)

gives you:

Analysis of Variance Table

Model 1: y ~ Base * trt + Age
Model 2: y ~ Base * trt + Age + Visit
  NoPar  LogLik Df -2logQ P.value
1     8 -625.87
2     9 -624.57  1    2.6 0.10686

Also, try

summary(fit)

(For standard deviations of the fixed effects, you maybe have to
to the .std file.)

To get AIC you have to type

> aic = -2*logLik(fit) -2*10

where you have to count the number of parameters yourself.

The complete list of functions (with their own help pages) is:

> library(help=glmmADMB)

anova.glmm.admb         Anova for glmm.admb objects
epil2                   Seizure Counts for Epileptics
glmm.admb               Generalized Linear Mixed Models using AD Model
                        Builder
glmmADMB-package        What the package does (short line) ~~ package
                        title ~~
glmmADMB-package        Generalized Linear Mixed Models using AD Model
                        Builder
logLik.glmm.admb        logLik.glmm.admb
plot.glmm.admb          plot.glmm.admb
print.glmm.admb         Print glmm.admb objects
ranef.glmm.admb         Extract Random Effects
summary.glmm.admb       summary.glmm.admb



Hans

On Mon, Mar 29, 2010 at 1:36 PM, Nina Bhola <nina.bhola at gmail.com> wrote:
> Hi
>
> I have been trying to implement the glmm.admb to my data.
> I have a model which works great with my data.
>
> However, i don't know exactly what some of the output is.
> For example, the parameters together with estimated standard
> deviations in the file nbmm.std. are not clear. How do you know which
> estimate belongs to which parameter?
>
> How do you extract AIC values?
>
> The -log-likelihood is in the file nbmm.par. that can be used for a
> likelihood ratio test on the significance of adding the parameter T to
> the model. What do you mean by T?
>
> The file nbmm.rep contains the predicted mean for each observation in
> the second column of the list. What are these means? The reason i ask
> is that the glmm.admb, should back-transform the estimates to give the
> us means, which should be the ones in the the nbmm.rep output file,
> however, if i exp(estimate) i do not get the same result as what is in
> this file? I get different predicted means which i am inclined to
> believe since i back transformed them myself! So what are these means
> in the output file?
>
> What does  Tau  r2/mu   r2/(mu*tau)  mean?
>
> Can we generate confidence intervals since there are no p values?
>
> Hope you can clarify these issues
> _______________________________________________
> Users mailing list
> Users at admb-project.org
> http://lists.admb-project.org/mailman/listinfo/users
>