[R-sig-ME] homogeneous glmmTMB models

[ben pelzer]

Thanks for your reply Ben, I've tried your "maps" suggestion for a Toeplitz
model and that works fine. Indeed it looks scary, as you say, but then
again this is also true for the way it is done in gls from the nlme
package, where Toeplitz is specified as an ARMA structure (which I do not
understand too, just apply it). The glmmTMB summary shows the estimated
variances and correlations under the Random Effects, which is not the case
with gls. Hence I like the glmmTMB results more. For gls I use covmat <-
getVarcov(model, individual="3") for a case observed at all time points,
and then covmat must be converted into correlations; all in all not really
nice in a course. Thanks again!

	[[alternative HTML version deleted]]