[R-sig-ME] MCMCglmm priors for ordinal response
v@@t@vr|n|de@ @end|ng |rom uc|@@c@uk
Tue Aug 6 11:43:34 CEST 2019
I am having some trouble understanding how to use priors for a GLMM where the response variable is ordinal (a scale from 1-6) and the explanatory variables are a mixture of factor and continuous variables. I have found a few posts on how to fix certain parameters in the “prior” argument of the MCMCglmm function, but it is not still very clear in my head. Also, if there is poor chain convergence that cannot be fixed by increasing the iteration number, changing the prior seems the next logical step, but I cannot understand how to make an educated guess of what the change should be. Could anyone please provide some valuable insight?
Thank you and apologies if this is a slightly simple issue – I am relatively inexperienced in this field.
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models