I haven't seen a reply to this, so I will offer a comment in case
you haven't already solved the problem.
Have you consulted the "Mixed-Effects Bible for S-Plus / R",
Pinheiro and Bates (2000) Mixed-Effects Models in S and S-Plus
(Springer)? If yes, have you worked through appropriate portions of the
book and the companion script files available with the standard R
distribution in "~R\library\nlme\scripts"? I just did "grep
'pdB' *.*"
in that directory and found 5 uses of pdBlocked, 3 in ch04.R, and 1 each
in ch06.R and ch08.R. Also, RSiteSearch("pdBlocked with 2 random
effects") produced 69 hits for me just now. You may not find anything
useful there, but 69 does not seem to large a list to search, and there
seems like a reasonable chance of finding something useful there.
Beyond this, a recommended approach to difficult questions like
this is to try to simplify it to the maximum extent possible. For
example, it sounds to me like your question, "can I use pdBlocked with
only 2 random effects", could be answered without the complexity of
'nlme'. What phony data set can you generate with the minimum number of
observations and variables that could help answer this question? The
process of producing such a simplified example might produce an answer
to your question. If it doesn't, then you can submit that simple,
self-contained example to this list. Doing so will increase the chances
of a useful reply.
I know this doesn't answer your question, but I hope it helps.
Best Wishes,
Spencer Graves
Daniel O'Shea wrote:> I am examining the following nlme model.
>
> asymporig<-function(x,th1,th2)th1*(1-exp(-exp(th2)*x))
>
mod1<-nlme(fa20~(ah*habdiv+ad*log(d)+ads*ds+ads2*ds2+at*trout)+asymporig(da.p,th1,th2),
> fixed=ah+ad+ads+ads2+at+th1+th2~1,
> random=th1+th2~1,
> start=c(ah=.9124,ad=.9252,ads=.5,ads2=-.1,at=-1,th1=2.842,th2=-6.917),
> data=pca1.grouped)
>
> However, the two random effects (th1 and th2) which describe the asymptotic
relationship between richness (fa20) and area (da.p) are correlated: -0.904 with
approximate 95% ci of -0.99 to -.32.
> I examined the anova of mod1 with both random effects and mod2 with just
th1 and mod1 is preferred. I also examined pdDiag(th1 + th2~1) for another
model (mod3) and based on the anova the original mod1 is preferred.
>
> My question is can I use pdBlocked with only 2 random effects or should I
and if so how I would specify that in the model or perhaps the 95% ci for
correlation is wide enough to ignore???
>
> Dan
>
> ______________________________________________
> 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
> and provide commented, minimal, self-contained, reproducible code.
>