[R-sig-ME] Assessing Normality for Mixed Models

[Lionel]

Sorry forgot one function in the code for the check of random effect 
normality:
qqnorm(unlist(ranef(model)))
qqline(unlist(ranef(model)))

Lionel

On 20/05/2014 21:25, Lionel wrote:
> Hi Jacob,
>
> You should do similar normality check as you would do for linear 
> models, I usually use the qqplot, you can use qqnorm(resid(model)) and 
> qqline(resid(model)). Then another assumptions from linear mixed 
> models is that the random effect are normally distributed with a mean 
> of 0, you can use qqnorm(unlist(ranef(model))) and 
> qqline(unlist(ranef(model))) if you have only one random term.
>
> So two things should be normally distributed in linear mixed models: 
> the residuals and the random effects.
>
> When you have a low number of level in the random effects normality 
> will in some case not be reached just due to the small number of 
> levels, I am not aware of ways to account for this, I would either 
> include the random effect as fixed effect or use simulation.
>
> Sincerely yours,
> Lionel
>
>
> On 20/05/2014 20:59, AvianResearchDivision wrote:
>> Hi All,
>>
>> After doing some extensive googling, searching for ways to assess 
>> normality
>> for linear mixed models, I can honestly say my head is swimming in
>> different proposed techniques that may or may not be valid. Also, when
>> reading the literature, I find that few studies that use linear mixed
>> models and random regression actually explicitly address how they assess
>> normality.  What are the rules with normality with mixed models (if 
>> there
>> are any) and what are your techniques to assess normality?  Any input 
>> that
>> you can provide would be great and hopefully we help to settle my 
>> mind on
>> this issue.
>>
>> Thank you,
>> Jacob
>>
>>     [[alternative HTML version deleted]]
>>
>> _______________________________________________
>> R-sig-mixed-models at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models