Message-ID: <5F267D06-4E39-40DE-AA37-ABA3C8859F81@ed.ac.uk>
Date: 2018-06-13T18:08:09Z
From: Jarrod Hadfield
Subject: MCMCglmm bivariate with offset
In-Reply-To: <91fa7964-2dbe-0dcc-b356-c27501a0ac20@gmail.com>
Hi,
I think you have 11 fixed effects, not 5. Replace 5 with 11 and then it should work. Also, you should expect the posterior to be sensitive to the prior if you have nu=3 unless there is a lot of information in the data.
Cheers,
Jarrod
> On 13 Jun 2018, at 15:28, Walid Mawass <walidmawass10 at gmail.com> wrote:
>
> Hello everyone,
>
> I am working on a bivariate MCMCglmm model and I want to include an
> offset term for my second response variable. I already know that for the
> offset, i have to fix the coefficient to 1 through the prior but it is
> not working in my case. I have 4 fixed effects fitted for both response
> variables and the offset is fitted for the second one using
> /at.level(trait,2)/.
>
> prior1 <- list(B=list(V=diag(5)*1e8, mu=rep(0,5)), R=list(V=diag(2),
> nu=3), G=list(G1=list(V=diag(2), nu=3),G2=list(V=diag(2), nu=3)))
> prior1$B$mu[5]<- 1
> prior1$B$V[5,5]<- 1e-8
>
> is this the proper way to set the prior? or should the fixed effects
> matrix have different dimensions since my model is bivariate:
>
> model_multi <- MCMCglmm(cbind(AFR, OffMortality)~trait-1 + trait:COEFPAR
> + trait:I(COEFPAR*COEFPAR) + trait:TWIN + trait:YEARM+
> at.level(trait,2):log(FERTILITY), random =
> ~us(trait):animal+us(trait):MOTHERW, rcov = ~us(trait):units, data =
> IAC, pedigree = prunedPed, family = c("gaussian", "poisson"), nitt =
> 3500000, burnin = 500000, thin = 3000, prior = prior1, verbose = FALSE,
> pr=TRUE)
>
> Thank you in advance for any advice.
>
> --
> Walid Mawass
>
> Ph.D. candidate in Cellular and Molecular Biology
>
> Population Genetics Laboratory
>
> University of Qu?bec at Trois-Rivi?res
> 3351, boul. des Forges, C.P. 500
> Trois-Rivi?res (Qu?bec) G9A 5H7
> Telephone: 819-376-5011 poste 3384
>
>
> [[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.