confidence intervals for brmultinom? about brglm2 HOT 1 CLOSED

Thanks for reporting this. This is just an oversight on my behalf in implementing a confint method for brmultinom. I will do this in the upcoming version. In the meantime, you can get confidence intervals using the non-exported function from nnet:

nnet:::confint.multinom(modelD, level = 0.95)

As expected the lower endpoints here (with the default level = 0.95) are the same to

sD <- summary(modelD)
sD$coefficients - qnorm(0.975) * sD$standard.errors

and the upper to

sD$coefficients + qnorm(0.975) * sD$standard.errors

PS:

1. Just noticed that you send me an email on that, which unfortunately ended up in the wrong directory. Sorry. Here is the right place for these things
2. I could not reproduce your example because I could not access "birthcenterdata". If you find any other issues, perhaps try to produce a minimally reproducible example using the examples in the help files (?brmultinom in this case).

from brglm2.