[R-sig-ME] MCMCglmm: Correct prior?

Katrijn Delaruelle Katrijn.Delaruelle at UGent.be
Mon Nov 23 16:59:52 CET 2015


Hello everyone,

 

I'm trying to run a multinomial MCMCglmm, but I'm having some problems with
specifying the prior.

My model consists of three levels and my dependent variable consists of four
categories.

 

I use the following model, in analogy to  this formulation:
https://hlplab.wordpress.com/2009/05/07/multinomial-random-effects-models-in
-r/

 

 

k <- length(levels(marital))

I <- diag(k-1)

J <- matrix(rep(1, (k-1)^2), c(k-1, k-1))

 

 

Model1_Multi <-
MCMCglmm(marital~-1+trait+trait:Parent+trait:edu*trait:cohort.+
trait:edu*trait:age

                         +trait:employment+trait:Income, random
=~idh(trait):cntry+ idh(trait):PC +
idh(trait):cohorts,rcov=~idh(trait):units,family = "categorical", 

                         prior=list(R=list(V=0.5 * (I + J), n = 4,fix=1),
G=list(G1=list(V=diag(3), nu=0.002), G2=list(V=diag(3), nu=0.002),
G3=list(V=diag(3), nu=0.002))), data=data2, nitt=5000,  burnin =1000)

 

This model works well, but I don't know if I generated a correct prior (e.g.
Did I use correct values to specify V and n?)

I would be very thankful if someone could help me.

 

Kind regards,

Katrijn Delaruelle

Vakgroep Sociologie

Universiteit Gent

 

 

 

 


	[[alternative HTML version deleted]]



More information about the R-sig-mixed-models mailing list