[R-sig-ME] glmer and influence.me - complaining about nAGQ==0

[Ben Bolker]

    Don't have much to add to John's comments. You can see 
vignette("lmerperf") for a few suggestions on improving performance.

   I'm a little surprised that your response variable is "logRT" *and* 
you have a log-link; that seems like double-logging?  (I was going to 
suggest that if you aren't wedded to the Gamma model, a log-Normal model 
(lmer(log(logRT) ~ ...) would probably be a lot faster ...)

   It's possible that other platforms (glmmTMB, Julia::MixedModels.jl) 
would be faster ... but then you might be stuck without influence 
diagnostics again ...

On 4/24/21 3:52 PM, John Fox wrote:
> Dear Catia,
> 
> I believe I understand what's going on: influence.merMod() works by 
> updating the model -- it calls update() -- using the parameter estimates 
> for the fit to the full data as start-values. The idea is to decrease 
> the computation necessary for each re-fit.
> 
> It probably would have been better if I had made this an option, with 
> using the start-values as the default, but perhaps there's a work-around:
> 
> You may be able to fit your model without specifying nACQ by instead 
> specifying the estimates that you obtained as start-values. Then glmer() 
> may converge quickly. If it doesn't, that would, I suppose, indicate 
> some sort of ill-conditioning.
> 
> Ben might well have something to add.
> 
> I hope this helps,
>   John
> 
> John Fox, Professor Emeritus
> McMaster University
> Hamilton, Ontario, Canada
> web: https://socialsciences.mcmaster.ca/jfox/
> 
> On 2021-04-24 2:29 p.m., Cátia Ferreira De Oliveira via 
> R-sig-mixed-models wrote:
>> Dear Professor Bolker,
>>
>> I am really sorry for bothering you but I have been quite stuck and have
>> posted my question both on reddit and cross validated but nothing has 
>> come
>> from it. Do you know if there is a good way of running dfbetas for glmer
>> with a gamma distribution?  The model runs without issues,
>>
>> lmer_log11 <- glmer(logRT ~ Probability*Block*Session*testing + (1 +
>> Block * Probability * Session|Participant), data= Data, family =
>> Gamma(link = "log"),
>> control=glmerControl(optimizer="bobyqa",optCtrl=list(maxfun=1000000)),
>> nAGQ = 0)
>>
>> yet when I run the influence function I get this error:
>>
>> influentialcases <- influence(glmer_log11, "Participant")
>>
>> Error in glmer(formula = RT ~ Probability * Block * Session * testing 
>> +  :
>>    should not specify both start$fixef and nAGQ==0
>>
>> I have tried to remove the "nAGQ==0" but the model is now taking 10 
>> days to
>> converge, a lot more than usual. Given this, I haven't been able to check
>> if the influential cases will run normally.
>>
>> Do you have any idea why this is happening? I would very much
>> appreciate your help!
>>
>> Thank you so much!
>>
>> Best wishes,
>>
>>     [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> R-sig-mixed-models using r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>>
> 
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models