[R-sig-ME] parameter estimates for all factor levels MCMCglmm
dp at dparis.com
Mon Dec 5 05:29:12 CET 2016
I'm new to Bayesian stats and the MCMCglmm package. I'm trying to understand how to interpret the results from a fitted model. I've fitted a model with a binomial response (reject/select), two categorical fixed effects (taxon with 16 levels, and size class with 3 levels), and a single random effect (bird ID). My model is mixing well and the results fit the data. My problem is that I'm not getting estimates for all levels of the fixed effects (19). How do I get the post mean and CIs for all levels in order to correctly interpret/write up the results?
As I have (near) complete separation in the data, I've used the fixed effect prior structure suggested in the Course Notes, fixed the residuals, and removed the global intercept.
prior.1 = list(
B = list(mu = rep(0, 18), V = (diag(18)) * (1 + pi^2/3)),
R = list(fix=1, V=1, n = k - 1),
G = list(G1 = list(V = 1, n = 1))
m.1 <- MCMCglmm(Selected ~ -1 + Arthropod + Size, random = ~bID, family = "categorical", prior = prior.1, data = type.selected, verbose = FALSE, nitt = 5e+05, burnin = 5000, thin = 100)
Thank you for your guidance,
School of Environmental Sciences
Institute for Land, Water and Society
Charles Sturt University
PO Box 789
Albury NSW 2640
M: +61 424 451 858?
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models