[R] MuMIn for GLM Negative Binomial Model

[Peter Ehlers]

On 2012-06-24 12:40, Rhodes, Christopher J. wrote:
> Hello
>
> I am not able to use the MuMIn package (version 1.7.7) for multimodel inference with a GLM Negative Binomial model (It does  work when I use GLM Poisson). The GLM Negative Binomial gives the following error statement:
>
> Error in get.models(NBModel, subset = delta<  4) :
>    object has no 'calls' attribute
>
> Here is the unsuccessful Negative Binomial code.
>>
>> BirdNegBin<-glm(B~Covariate1 + Covariate2 + Covariate3 + Covariate4 + Covariate5 + Covariate6
>> + Covariate7, data = SquareSumLowland,family = negative.binomial(theta = 1))
>>
>> NegBinDredge<- dredge(BirdNegBin)
>> confset.d4<- get.models(BirdNegBin, subset = delta<  4)
>> model.avg(confset.d4)
>> confset.95p<- get.models(BirdNegBin, cumsum(weight)<= .95)
>> avgmod.95p<- model.avg(confset.95p)
>
> Any assistance would be appreciated.
>

get.models() expects an object returned by dredge().
You're feeding it an object returned by glm().

Peter Ehlers