bdlvm provides a simple interface for building latent variable models on top of brms. Currently, it works through the mi() missing variable indicator, simply providing a wrapper around its use.
Download from this repo:
devtools::install_github("bdlvm-project/bdlvm-pkg")Specify a latent variable model and transform it into a brms formula in two simple steps:
library(bdlvm)
cfa_formula <- lv(x ~ items(y, 3))
bdlvm_parse(cfa_formula)
# x | mi() ~ 1
# yi1 ~ mi(x)
# yi2 ~ mi(x)
# yi3 ~ mi(x) Done! The result can be used in brms::brm() as usual. Just make sure your data.frame has the corresponding manifest variables and columns consisting solely of NA_real_ for each latent variable.
See the documentation at ?lv() for more details.