[R-sig-ME] help with ordinal MCMCglmm

[Jarrod Hadfield]

Hi Achaz,

I think the code below does what you want.

CP <- cbind(-Inf, 0, mod1$CP, Inf)  # Obtain posteior cutpoints

Sol <-mod1$Sol  # Obtain posterior fixed effects (I think there are two?)

VCV<-rowSums(mod1$VCV) # Total variance

prob.with<-mcmc(matrix(NA, 1000,15))
prob.without<-mcmc(matrix(NA, 1000,15))


for (i in 1:15) {
prob.with[,i] <-  pnorm(CP[, i + 1] - Sol[,1], 0, sqrt(VCV + 1)) -  
pnorm(CP[, i] -Sol[,1], 0, sqrt(VCV + 1))
prob.without[,i] <-  pnorm(CP[, i + 1] - Sol[,2], 0, sqrt(VCV + 1)) -  
pnorm(CP[, i] -Sol[,2], 0, sqrt(VCV + 1))
}

Cheers,

Jarrod


Quoting Achaz von Hardenberg <achaz.hardenberg at gmail.com>:

> dear all,
> I would like to ask your help on the following model:
> I have data for the size of zooplankon in lakes with (Y) and without  
> (N) introduced fish. there are 15 size classes with 1 the smallest  
> and 15 the largest. Number of zooplankton samples in each class are  
> as follows (cumulated over all lakes):
>
>> Ctable
>
>          N     Y
>   1   7901 24174
>   2   2869  2331
>   3    964   737
>   4   1573   621
>   5   1857   584
>   6   2073   777
>   7   2078   548
>   8    919   171
>   9    485    28
>   10   239     3
>   11    64     0
>   13    30     0
>   14     3     0
>   15     1     0
>
> To test the hypothesis that fish eat the biggest zooplanktons,
> as I have repeated measures in different lakes, I thought to model  
> this data as follows,:
>
> prior=list(R=list(V=1, fix=1), G=list(G1=list(V=1, nu=0)))
> mod1<-MCMCglmm(CLASS~Fish-1,  
> random=~lago,family="ordinal",data=dimensioni.dat, prior=prior)
>
> where CLASS is the size class and "Fish" the variable indicating if  
> the lake has or not introduced fish.
>
> The model seems to run fine, but I have trouble in obtaining the  
> expected probabilities for each size category in lakes with and  
> without introduced fish... Anybody can help me out?
>
> your help would be greatly appreciated!
>
> cheers,
> Achaz von Hardenberg
> Gran Paradiso National PArk. Italy
>
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>



-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.