computeNonExperimental: Compute Model-implied CMSE Elements

In trbrick/CMSE: CMSE: Causal Mean Squared Error index

Description

Computes b.nonexperimental and a.nonexperimental values for the CMSE computation.

Usage

computeNonExperimental(
  ctl,
  models,
  posttest,
  outcomes,
  intervention = NA,
  ...,
  returnModels = FALSE
)

Arguments

ctl	the control group only of the experimental data set
models	a list of mxModel objects (like those returned from makeARPanelModels) for fitting
posttest	character string containing the name of the posttest true score (see details), or a vector of the same length as models containing the true scores for each
outcomes	character vector including names of all additional outcomes, or a list the same length as models with outcomes for each model
intervention	character vector containing the name of the intervention. If present, the intervention will be isolated to avoid causing misfit.
...	additional arguments to mxTryHard()
returnModels	(default FALSE) whether to return the models themselves in addition to the CMSE elements.

Details

This function returns the (standardized) model-implied causal effect of a change to the posttest change of one (standardized) unit. CMSE computation requires this to be performed only on the control group. If the intervention exists in the model, it is isolated by removing all outgoing paths. This prevents it from influencing the paths. Its variance is bounded above zero to ensure that the model-implied covariance matrix does not become non-positive-definite.

#' @import MICr

Value

a list containing the model-implied effect of the (standardized) posttest on each (standardized) outcome

trbrick/CMSE documentation built on March 17, 2021, 4:15 a.m.