semMatrixAlgebra: Extract or calculate with model matrices

In semPlot: Path Diagrams and Visual Analysis of Various SEM Packages' Output

Description

This function can be used to extract or calculate with model matrices given a "semMatriModel" object (from modelMatrices) or a "semPlotModel" object or any of the input types that can be used in semPlotModel directly.

If the model is not specified it is attempted to be identified by the given algebra.

Usage

semMatrixAlgebra(object, algebra, group, simplify = TRUE, model, endoOnly = FALSE)

Arguments

object	A "semMatriModel" object (from modelMatrices) or a "semPlotModel" object or any of the input types that can be used in semPlotModel directly.
algebra	An R expression to use.
group	Groups the algebra should be used on. If more than one a list is returned with the result for each group.
simplify	If TRUE and only one group is used, return output as is instead of in a list.
model	Model to be used in modelMatrices, "mplus", "ram" or "lisrel"
endoOnly	Only needed when the model is "lisrel", sets all variables to endogenous.

Details

The "lisrel" model uses the following matrix names: LY, TE, PS, BE, LX, TD, PH, GA, TY, TX, AL and KA.

The "mplus" model uses the following matrix names: Lambda, Nu, Theta, Kappa, Alpha, Beta, Gamma and Psi.

The "ram" model uses the following matrix names: F, A and S.

Value

A list containing output per group

Author(s)

Sacha Epskamp <[e-mail]>

See Also

semPlotModel semPlotModel-class modelMatrices lisrelModel ramModel

Examples

## Mplus user guide SEM example:
outfile <- tempfile(fileext=".out")
download.file("http://www.statmodel.com/usersguide/chap5/ex5.11.out",outfile)

# Plot model:
semPaths(outfile,intercepts=FALSE)

# Obtain latent regressions (mplus)
semMatrixAlgebra(outfile, Beta)

# mplus model implied covariance:
mat1 <- semMatrixAlgebra(outfile, 
  Lambda %*% Imin(Beta, TRUE) %*% Psi %*% t(Imin(Beta, TRUE)) %*% t(Lambda) + Theta)

# Lisrel model implied covariance:
mat2 <- semMatrixAlgebra(outfile, 
  LY %*% Imin(BE, TRUE) %*% PS %*% t(Imin(BE, TRUE)) %*% t(LY) + TE, endoOnly = TRUE)

# RAM model implied covariance:
mat3 <- semMatrixAlgebra(outfile, 
                 F %*% Imin(A,TRUE) %*% S %*% t(Imin(A, TRUE)) %*% t(F))

## Not run: 
# Plot:
library("qgraph")

pdf("Models.pdf",width=15,height=5)
layout(t(1:3))
qgraph(round(cov2cor(mat1),5), maximum=1, edge.labels=TRUE, layout = "spring", 
  cut = 0.4, minimum = 0.1)
title("Mplus model")
qgraph(round(cov2cor(mat2),5), maximum=1, edge.labels=TRUE, layout = "spring", 
  cut = 0.4, minimum = 0.1)
title("LISREL model")
qgraph(round(cov2cor(mat3),5), maximum=1, edge.labels=TRUE, layout = "spring", 
  cut = 0.4, minimum = 0.1)
title("RAM model")
dev.off()

## End(Not run)
# They are the same.

semPlot documentation built on May 30, 2017, 6:29 a.m.

Visually inspect implied and observed correlations

semPlot index

Plot path diagram for SEM models.