1
$\begingroup$

I'm running a glmm with the lme4 package in R that looks like the following:

glmer(y ~ factor1 + factor2 + (1 | RE), data=data, poisson)

And I get the following warning message:

Warning message:
In mer_finalize(ans) : gr cannot be computed at initial par (65)

Both glmer(y ~ factor1 + (1 | RE), data=data, poisson) and glmer(y ~ factor2 + (1 | RE), data=data, poisson) work just fine without a warning message. Can someone explain to me what this warning means, why I'm getting it, and what, if anything, I might do about it?

$\endgroup$
5
  • $\begingroup$ rank deficient model matrix? e.g. lack of balance, missing cell values $\endgroup$ Mar 8, 2012 at 21:34
  • 3
    $\begingroup$ I think this means the gradient at the initial parameter value is non-finite or NaN. Try changing the initial values. $\endgroup$
    – Macro
    Mar 9, 2012 at 13:10
  • $\begingroup$ @Macro your comment looks like the answer to this question. Did you find anything else Frank? Would one of you mind putting this as an answer? $\endgroup$
    – Corvus
    Feb 26, 2013 at 16:26
  • 1
    $\begingroup$ @Corone, Basically all I did was read the warning message. I usually only answer if I can offer something above and beyond the obvious. If you think resolving the unanswered question is a high priority then you have my permission to post my comment as an answer. I think what is more important is that this question is off topic. I'm voting to close. $\endgroup$
    – Macro
    Feb 26, 2013 at 20:08
  • $\begingroup$ I agree then close is sensible $\endgroup$
    – Corvus
    Feb 26, 2013 at 20:13

0