[R] NaNs in Nested Mixed Model

[Johan Stenberg]

Thank you Thierry for your kind answer!

If you don't mind I would like to ask a follow-up question. In your
suggestions I get P-values for "Species". However, I am really not
interested in that factor per se. Would it make sense to use this
model instead if I am only interested in "Genotype"?

> model<-glmer(Nymphs~Genotype+(1|Species/Genotype),family=poisson)

All the best,
Johan

2011/3/9 ONKELINX, Thierry <Thierry.ONKELINX at inbo.be>:
> Dear Johan,
>
> A few remarks.
>
> - R-sig-mixed models is a better list for asking questions about mixed model.
> - I presume that Nymphs is the number of insects? In that case you need a generalised linear (mixed) model with poisson family
> - What are you interessed in? The variability among genotypes or the effect of each genotype.
>        You can achieve the first with a glmm like glmer(Nymphs ~ Species + (1|Genotype), family = "poisson"). Genotype will be implicitly nested in Species.  Note that since you have only 4 genotypes, you will not get very reliable estimates of the genotype variance.
>        For the latter you cannot use a mixed model so you need a simple glm(Nymphs ~ Species/Genotype, family = "poisson"). Note that several coefficients will be NaN, because you cannot estimate them.
>
> Best regards,
>
> Thierry
>
> ----------------------------------------------------------------------------
> ir. Thierry Onkelinx
> Instituut voor natuur- en bosonderzoek
> team Biometrie & Kwaliteitszorg
> Gaverstraat 4
> 9500 Geraardsbergen
> Belgium
>
> Research Institute for Nature and Forest
> team Biometrics & Quality Assurance
> Gaverstraat 4
> 9500 Geraardsbergen
> Belgium
>
> tel. + 32 54/436 185
> Thierry.Onkelinx at inbo.be
> www.inbo.be
>
> To call in the statistician after the experiment is done may be no more than asking him to perform a post-mortem examination: he may be able to say what the experiment died of.
> ~ Sir Ronald Aylmer Fisher
>
> The plural of anecdote is not data.
> ~ Roger Brinner
>
> The combination of some data and an aching desire for an answer does not ensure that a reasonable answer can be extracted from a given body of data.
> ~ John Tukey
>
>
>> -----Oorspronkelijk bericht-----
>> Van: r-help-bounces at r-project.org
>> [mailto:r-help-bounces at r-project.org] Namens Johan Stenberg
>> Verzonden: dinsdag 8 maart 2011 16:52
>> Aan: r-help at r-project.org
>> Onderwerp: [R] NaNs in Nested Mixed Model
>>
>> Dear R users,
>>
>> I have a problem with something called "NaNs" in a nested mixed model.
>>
>> The background is that I have studied the number of insect
>> nymphs emerging from replicated Willow genotypes in the
>> field. I have 15 replicates each of 4 Willow genotypes
>> belonging two 2 Willow species.
>> Now I want to elucidate the effect of Willow genotype on the
>> number of emerging nymphs. Previously I performed a simple
>> one-way anova with "genotype" as explanatory factor and
>> "number of nymphs emerging" as dependent variable, but the
>> editor of the journal I've submitted this piece to wants me
>> to nest Willow genotype within Willow species before he
>> accepts the paper for publication [Species*Genotype(Species)].
>>
>> The fact that I didn't include "Willow species" as a factor
>> in my initial analysis reflects that I am not very interested
>> in the species factor per se - I am just interested in if
>> genetic variation in the host plant is important, but
>> "species" is of course a factor that structures genetic diversity.
>>
>> I thought the below model would be appropriate:
>>
>> > model<-lme(Nymphs~Species*Genotype,random=~1|Species/Genotype)
>>
>> ...but I then get the error message "Error in MEEM(object, conLin,
>> control$niterEM) : Singularity in backsolve at level 0, block 1"
>>
>> I then tried to remove "Genotype" from the fixed factors, but
>> then I get the error message "NaNs produced".
>>
>> > model<-lme(Nymphs~Species,random=~1|Species/Genotype)
>> > summary(model)
>> Linear mixed-effects model fit by REML
>>  Data: NULL
>>        AIC      BIC    logLik
>>   259.5054 269.8077 -124.7527
>>
>> Random effects:
>>  Formula: ~1 | Species
>>         (Intercept)
>> StdDev:   0.9481812
>>
>>  Formula: ~1 | Genotype %in% Species
>>         (Intercept) Residual
>> StdDev:   0.3486937 1.947526
>>
>> Fixed effects: Nymphs ~ Species
>>                  Value Std.Error DF   t-value p-value
>> (Intercept)   2.666667  1.042243 56  2.558585  0.0132
>> Speciesviminalis -2.033333  1.473954  0 -1.379510     NaN
>>  Correlation:
>>              (Intr)
>> Speciesviminalis -0.707
>>
>> Standardized Within-Group Residuals:
>>        Min         Q1        Med         Q3        Max
>> -1.4581821 -0.3892233 -0.2751795  0.3439871  3.1630658
>>
>> Number of Observations: 60
>> Number of Groups:
>>           Species Genotype %in% Species
>>             2             4
>> Warning message:
>> In pt(q, df, lower.tail, log.p) : NaNs produced
>> ***********
>>
>> Do you have any idea what these error messages mean in my
>> case and how I can get around them?
>>
>> Thank you on beforehand! (data set attached).
>>
>> Johan
>>