I’ve received a lot of help for fitting models like this from others. e.g., see here .
I’m trying to fit a multivariate outcomes model in brms in long format. The wide-format equation looks like this.
bf(y_correct ~ 0 + (1 | p | subject)) + bf(y_error ~ 0 + (1 | p | subject)) + set_rescor(TRUE)
If re-written in a long-format, so correct (y_correct) and error trial scores (y_error) are in the same column (y) with event type (correct, error) in a separate one, I end up with the following syntax.
bf(y ~ 0 + event + (0 + event | subject), sigma ~ 0 + event)
This gives me all of the estimates I need, except for covariances between the sigma for correct and sigma for error. (i.e., the same output cor in
summary() thanks to
set_rescor(TRUE) from the wide-format above).
Anyone have any recommendations?
VarCorr() does not include the sigmas.
I apologize if I missed something in a vignette somewhere… I’ve been working on this for a while now…
Thank you for you thoughts!
- Operating System: OS X 10.15.6
- brms Version: 2.13.5