Extract exogenous variables (predictors), endogenous variables (outcomes),
latent variables (random effects), manifest (observed) variables from a
lvm object.
vars(x,...)
endogenous(x,...)
exogenous(x,...)
manifest(x,...)
latent(x,...)
# S3 method for lvm
exogenous(x, xfree = TRUE,...) <- value
# S3 method for lvm
exogenous(x,variable,latent=FALSE,index=TRUE,...)
# S3 method for lvm
latent(x,clear=FALSE,...) <- valuelvm-object
Additional arguments to be passed to the low level functions
list of variables to alter
Logical defining whether latent variables without parents should be included in the result
For internal use only
Logical indicating whether to add or remove latent variable status
For internal use only
Formula or character vector of variable names.
Vector of variable names.
vars returns all variables of the lvm-object including
manifest and latent variables. Similarily manifest and latent
returns the observered resp. latent variables of the model.
exogenous returns all manifest variables without parents, e.g.
covariates in the model, however the argument latent=TRUE can be used
to also include latent variables without parents in the result. Pr. default
lava will not include the parameters of the exogenous variables in
the optimisation routine during estimation (likelihood of the remaining
observered variables conditional on the covariates), however this behaviour
can be altered via the assignment function exogenous<- telling
lava which subset of (valid) variables to condition on. Finally
latent returns a vector with the names of the latent variables in
x. The assigment function latent<- can be used to change the
latent status of variables in the model.
endogenous, manifest,
latent, exogenous, vars
g <- lvm(eta1 ~ x1+x2)
regression(g) <- c(y1,y2,y3) ~ eta1
latent(g) <- ~eta1
endogenous(g)
#> [1] "y1" "y2" "y3"
exogenous(g)
#> [1] "x1" "x2"
identical(latent(g), setdiff(vars(g),manifest(g)))
#> [1] TRUE