[R-sig-ME] Error message when handling complete separation

[Ben Bolker]

   This might be fixable/hackable with bglmer, but we'd need a
reproducible example.  (Does tightening the prior help at all, e.g.
normal(cov=diag(4,4)) ?)

On 2019-10-21 11:40 a.m., Amal Dahounto wrote:
> Ok;
> thank you so much for reply and your suggestion.
> Regards,
> 
> Le lun. 21 oct. 2019 à 13:32, D. Rizopoulos <d.rizopoulos using erasmusmc.nl> a
> écrit :
> 
>> In GLMMadaptive you can solve separation issues by including a penalty
>> term for the coefficients. You may find an example on how to do this at the
>> bottom of this vignette:
>> https://drizopoulos.github.io/GLMMadaptive/articles/GLMMadaptive_basics.html
>>
>>
>> Best,
>> Dimitris
>>
>> *From: *C. AMAL D. GLELE <altessedac2 using gmail.com>
>> *Date: *Monday, 21 Oct 2019, 13:21
>> *To: *R SIG Mixed Models <r-sig-mixed-models using r-project.org>
>> *Subject: *[R-sig-ME] Error message when handling complete separation
>>
>> Hi, all.
>> When trying to handle a complete separation case
>> ( initialmodel<-glmer(resp~treatment+(1|net),family=binomial,data=mydata) ,
>> where: treatment is a factor with 4 levels; net has 4 levels;
>> resp<-cbind(,)
>> Warning message:
>> unable to evaluate scaled gradient Hessian is numerically singular:
>> parameters are not uniquely determined
>> Outputs: huge estimates, huges sderror and p-value=1 everywhere
>> )
>> to solve this, I tried the following fit;
>> mod2kisBW_unsep1<-
>>
>> bglmer(respkisBW~treatment+(1|net),data=conedata1kisBW,family=binomial,fixef.prior
>> = normal(cov=diag(9,4)))
>> but, I get the error message below:
>>
>> Error in length(value <- as.numeric(value)) == 1L :
>>   pwrssUpdate did not converge in (maxit) iterations
>>
>> 1) what are the possible causes of such a problem?
>> 2) how can I figure it out?
>> Thanks, in advance.
>> Regards,
>>
>>         [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> R-sig-mixed-models using r-project.org mailing list
>>
>> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstat.ethz.ch%2Fmailman%2Flistinfo%2Fr-sig-mixed-models&data=02%7C01%7Cd.rizopoulos%40erasmusmc.nl%7Cdf91bdf080be490a9aeb08d75618c929%7C526638ba6af34b0fa532a1a511f4ac80%7C0%7C1%7C637072536761480378&sdata=o6YoF5%2BXMcfDkQWjq7iKPVcKi3AjK6Wpq2InrrjA5Os%3D&reserved=0
>>
> 
>