Joshua F. Wiley, Ph.D.

If it is any help, I'm not sure 'bias' is the optimal description here. In general: `mean(Y) != ginv(mean(g(y)))` as a specific example: `mean(Y) != exp(mean(log(Y)))` this is not really...

This would be a great feature. Great to see it planned. Minor note: since the objects are already serialized, they are ripe for easy & fast compression using `fst` (see...

Has any progress been made on this? I am having very similar issues. For my personal use, I use `cmdstanr` which works well. However, I am working on a package...

Not sure if this would be any help, but here is code I used to calculate population averaged values out of `brms`. - https://github.com/JWiley/brmsmargins/blob/main/src/integratemvn.cpp - https://github.com/JWiley/brmsmargins/blob/main/src/integratere.cpp It is fairly straight...

Hi, It is looking for rdata and usevariables to be passed. The intention of mplusModeler() and mplusObject() is that the data are in R and get created for Mplus for...

**Answer 1.** The random effects are only integrated out when `effect = "integrateoutRE"`. **Answer 2.** - `effects = "fixedonly"` will set all REs = 0. So this gives you estimates...

@isabellaghement I do not think it is off base or unreasonable to consider narrow or broad inference. I have not given the narrow sense a lot of consideration as that...

@vincentarelbundock Intriguing. I had not seen the sample new levels. Yes that sounds on target. Agreed you'd not have identical replication due to random sampling. Can you control the number...

@vincentarelbundock Very cool. I've no problems with these discussions continuing here on GitHub. This is looking great. I haven't looked enough through the rest of the `brms` framework, but indeed...

@vincentarelbundock one more question on your implementation --- how are you handling AMEs for continuous variables? For 100 new levels to be sampled, the entire raw dataset would get expanded...