[R-sig-ME] glmer and start values help

[Adam D. I. Kramer]

Hi,

I'm trying to muck about a bit with the start values for a glmer model.
Here's what I've set:

> ST
[[1]]
             (Intercept)
(Intercept)      -0.779

> fef
[1] 0.04519647 0.01011628
> has_pos <- glmer(has_fpos_d1 ~ has_upos_d1 + (1 | userid), data=x,
+ family='binomial', verbose=TRUE, control=list(msVerbose=TRUE,
+ maxIter=500, maxFN=5000), start=list(ST=ST, fixef=fef))
   0:     10415953.: 0.608675 0.0451965 0.0101163
   1:     10305257.:  0.00000 -0.139832 -0.361935

...as you can see, the two fixed effects startvalues are as I said they
should be at step 0, but the random effect start value is not. (Or at least,
it seems really weird that the fixef params would be spot-on the same if the
report from step 0 is actually the report of the values after the
iteration.)

The help page is pretty clear on the topic:

   If the list contains components named 'fixef' and/or 'ST', these are used
   as the starting values for those slots.

Any thoughts?

(has_fpos_d1 and has_upos_d1 are both logical.)

R is [1] "R version 2.13.1 (2011-07-08)"
lme4 is     note = {R package version 0.999375-41}

--Adam