[R-sig-ME] Why se.fit differ in predict.glm and predict.glmmadmb?

[Ben Bolker]

  Not sure, this will be worth looking into ...

On 16-05-03 04:51 PM, Xandre wrote:
> Dear list,
> 
> I am running a GLM (family="binomial") without random effects using both 
> glm and glmmadmb.
> 
> Summaries are almost identical, however when I used the predict function 
> as follows:
> 
> predict(glm1,newdatos1,type="link",se.fit =T)
> 
> predict(admb1,newdatos1,type="link",se.fit =T)
> 
> I realized that se.fit differ a lot between them, admb se.fit resulted 
> much much higher (fit is almost identical). This is just and example of 
> what I found:
> 
> glm1$se.fit	admb1$se.fit
> 0.04290869	0.2676562
> 0.04435600	0.2733130
> 0.04095631	0.2728592
> 0.03402992	0.2718389
> 0.03000669	0.2713617
> 0.03633637	0.2722059
> 
> Maybe I'm missing something or I am making a big mistake. Any help with 
> this?
> 
> 
> Many thanks,
> 
> Alexandre Alonso
> 
> 
> 	[[alternative HTML version deleted]]
> 
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>