[R-sig-ME] Anomalous results with glmer().

Rolf Turner r.turner at auckland.ac.nz
Wed May 28 05:02:40 CEST 2014


Yes.  That makes quite a bit of sense.  Phoneme types are going to be 
partially confounded with words.  A glimmer of light is starting to seep 
through the cracks.

Still leaves a bit of a puzzle as to why the 0.999999-0 and 1.1-7 
results are so different.



On 28/05/14 14:49, David Duffy wrote:
> On Wed, 28 May 2014, Rolf Turner wrote:
>> Some months back I sent an inquiry to this list concerning the
>> analysis of some linguistics data with which I am involved.
>> essentially zero.  Which is silly.
>> If I fit the same model using lme4 version 1.1-7 (and ignore the
>> warning about failure to converge) I get sensible looking estimates of
>> the variances of the random effects, but an impossibly wrong estimate
>> of at least one of the fixed effect coefficients.  (The estimate says
>> that the success probability is larger for phoneme type "Mclus" than
>> it is for the baseline type "Fclus".  However a raw tabulation show
>> that the success probability for Mclus is much, much smaller than for
>> Fclus.
>                    MClus
> glm               -2.6950  (0.10455)
> glmer (+Stud)     -2.74464 (0.10536)
> glmer (+Words)     0.36826 (0.19783)
> glmer (+S+W)       0.33574 (0.19981)
> glmmML (+Stud)    -2.7444 (0.10546)
> glmmML (+Words)    0.3683 (0.19816)
> It's words that will always get you into trouble...

More information about the R-sig-mixed-models mailing list