Skip to contents

Fit a latent variable model using bias-reducing methods

Source: R/11-brlavaan.R
brlavaan.Rd

Fit a latent variable model using bias-reducing methods

Usage

brlavaan(
  model,
  data,
  estimator = "ML",
  estimator.args = list(rbm = "implicit", plugin_pen = NULL),
  information = "observed",
  lavfun = "sem",
  ...
)

Arguments

model

A description of the user-specified model. Typically, the model is described using the lavaan model syntax. See lavaan::model.syntax for more information. Alternatively, a parameter table (eg. the output of the lavaan::lavaanify() function) is also accepted.

data

An optional data frame containing the observed variables used in the model. If some variables are declared as ordered factors, lavaan will treat them as ordinal variables.

estimator

The estimator to use. Currently only "ML" is supported.

estimator.args

A list containing RBM arguments. Possible arguments are

  • rbm: The type of RBM method to use. One of "none", "explicit", or "implicit" (although, short forms are accepted, e.g. "exp", "iRBM", etc.)

  • plugin_pen: The type of penalty to use. One of NULL, "pen_ridge", or "pen_ridge_bound".

  • info_pen The type of information matrix to use for the penalty term.

  • info_bias The type of information matrix to use for the bias term of the explicit reduced bias method.

information

The type of information matrix to use for calculation of standard errors. Defaults to "observed", although "expected" and "first.order" is also permitted.

lavfun

The lavaan function to use. Default is "sem".

...

Additional arguments to pass to the lavaan::lavaan function.

Value

An object of class brlavaan which is a subclass of the lavaan::lavaan class.

Details

The pen_ridge() function applies a ridge regression style penalty $f(x) = || x ||^2$ that shrinks the parameters to zero. The pen_ridge_bound() function applies a penalty that shrinks the parameters to the bounds of the parameter space. The bounds are calculated by {lavaan} – see the paper for more details.

The info_pen and info_bias arguments default to "observed". It is not recommended to change this to say "expected", especially for small sample sizes as the bias-reducing properties of the estimators are not guaranteed.