[R-sig-ME] Log likelihood of a glmer() binomial model .

Rolf Turner r@turner @end|ng |rom @uck|@nd@@c@nz
Mon Apr 22 01:48:54 CEST 2019

On 22/04/19 4:27 AM, Juho Kristian Ruohonen wrote:

> Rolf: I'm no Prof but a lowly grad student of an unrelated field, so 
> take all my input with a grain of salt.

Well, you will very likely *be* a Prof. someday.  And your input seems 
very sound to me.  (I am a retired "Honorary" Research Fellow, which 
seems to me to be even lowlier than a grad student; grad students 
generally get at least a *bit* of monetary compensation, in the form of 
scholarships or other stipends. :-) )

> As alluded to by Ben, predict() can certainly provide fitted 
> probabilities for the validation set with the random effects taken into 
> account. This is achieved by the re.form = NULL argument. However -- and 
> I'll be happy to be corrected on this -- the problem is that dbinom() 
> will calculate a (log)likelihood of the observed responses assuming a 
> regular binomial PMF, which does not apply in the case of a 
> mixed-effects model. Thus the result will not equal the loglikelihood 
> that is maximized in the fitting process, i.e. will not equal 
> logLik(validationFit) unless it's a standard logistic GLM.

Thanks.  I'm going to have to chew this over a bit more .... Takes me a 
while to get my head around these issues.



Honorary Research Fellow
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276

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