Confirmatory Factor Analysis with R

A Draft document using lavaan, sem, and OpenMx

Bruce Dudek
2019-07-11
Contents

Preface 3

1 Introduction and R Setup 4

1.1 Caveat on this document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Prepare and Describe the Data 6

2.1 The Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Numeric and Graphical Description of the Data . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Bivariate Characteristics of the data set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Covariances and Zero Order Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3 Using the lavaan package for CFA 15

3.1 Implement the CFA, First Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Generate a second model and compare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3 Compare Model 1 and Model 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.4 An additional perspective on estimation and optimization . . . . . . . . . . . . . . . . . . . . 26

4 Using the sem package for CFA 30

4.1 Example one . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.2 Example two . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.3 Compare the two CFA models produced by sem . . . . . . . . . . . . . . . . . . . . . . . . . 40

5 Using the OpenMx Package for CFA 41

5.1 First OpenMx Model - Single Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.2 Second OpenMx Model - the bifactor model . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.3 Third OpenMx Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.4 Compare the OpenMx models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

6 Summary and Reproducibility 54

2
Preface

The data set and the models evaluated are those used by James Boswell in his APSY613 Multivariate
Analysis class in the Psychology Department at the University at Albany. The data set is the WISC-R data
set that the multivariate statistics textbook by the Tabachnick textbook (Tabachnick et al., 2019) employs for
confirmatory factor analysis illustration. The goal of this document is to outline rudiments of Confirmatory
Factor Analysis strategies implmented with three different packages in R. The illustrations here attempt to
match the approach taken by Boswell with SAS. The document is targeted to UAlbany graduate students
who have already had instruction in R in their introducuctory statistics courses.
This book/monograph uses the bookdown package (Xie, 2018a) for R (R Core Team, 2018), which was
built on top of rmarkdown (Allaire et al., 2018) and knitr (Xie, 2015). RStudio (RStudio Team, 2015)
was used for all writing and programming.

3
Chapter 1

Introduction and R Setup

This short monograph outlines three approaches to implementing Confirmatory Factor Analysis with R,
by using three separate packages. The illustration is simple, employing a 175 case data set of scores on
subsections of the WISC. The idea is to fit a bifactor model where the two latent factors are the verbal and
performance constructs. In this primary two-factor model, each observed variable is associated with only one
latent factor. Then a second model is fit. It includes a path from both latent factors to one of the variables.
Comparisons of models are then performed.

Several R packages are required for the implementations outlined in the succeeding chapters. Since CFA
is implemented as a structural equation model, commercial software (e.g., LISREL, EQS, SAS) as well as
open-source approaches to CFA all use SEM routines. The three primary R packages to illustrate CFA are
lavaan, sem and OpenMx, along with the drawing package, semPlot. One major advantage of using
R for implementation of these methods is that semPlot provides a user-friendly method for producing
path diagrams of many styles by simply taking a model object from the CFA fitting functions of the other
packages.

Other “housekeeping” packages are loaded here, but the three analytical packages for CFA are loaded at
the point in the sequence of their usage since some common function names are shared - thus load order is
important.
library(car)
library(semPlot)
library(psych)
library(knitr)
library(kableExtra)
library(MVN)
library(dplyr)
library(magrittr)
library(tidyr)
library(corrplot)
library(ggraph)

Package citations for packages loaded here (in the above order): car (Fox et al., 2018), semPlot (Epskamp
and with contributions from Simon Stuber, 2017), psych (Revelle, 2019), knitr (Xie, 2018b), kableExtra
(Zhu, 2019), MVN (Korkmaz et al., 2018), dplyr (Wickham et al., 2018), magrittr (Bache and Wickham,
2014), tidyr (Wickham and Henry, 2018), corrplot (Wei and Simko, 2017)

Package citations for packages loaded elsewhere in this document: bookdown (Xie, 2018a), rmarkdown
(Allaire et al., 2018), sem (Fox et al., 2017), lavaan (Rosseel, 2018), OpenMx (Boker et al., 2019)

4
1.1. CAVEAT ON THIS DOCUMENT 5

1.1 Caveat on this document

The present treatment of the CFA procedures is not intended to be an exhaustive analysis of this particular
data set. Nor is it intended to be a thorough treatment of the CFA approaches available in R, CFA in
general, or SEM in general. Rather, it is intended as a bit more than a simple introduction to CFA using
R (and by implication, the nice capabilities of for Structural Equation Modeling). It provides students, who
have a basic understanding of how to use R, with a reasonable introduction to CFA modeling code. The R
appraoches can then be compared to their class coverage of the same analysis, done with SAS. This document
provides some capabilities that may not have been covered in class, and it misses others. The learning curve
for software is never at an asymptote……

1.2 Resources
The following list will provide a good start for those needing a broader in CFA modeling and more detailed
sources for the primary packages empoyed in this document.
• A comprehensive textbook treatment of SEM and CFA: (Tabachnick et al., 2019)
• Tim Brown’s well-regarded book on CFA: (Brown, 2015)
• Rosseel’s extensive original article on lavaan: (Rosseel, 2012)
• El-Sheik, et al on a comparison of software for SEM: (El-Sheikh et al., 2017)
• Narayanan’s review of eight SEM software approaches (Narayanan, 2012)
• Espkamp’s helpful original article on the semPlot package: (Narayanan, 2012)
In addition, the following internet resources can be helpful.
• Lavaan package home: [http://lavaan.ugent.be/]
• Google Group for Lavaan: [https://groups.google.com/forum/#!forum/lavaan]
• OpenMx package home: [https://openmx.ssri.psu.edu/wiki/projects]
• OpenMx package online forums: [https://openmx.ssri.psu.edu/forums]
• SEM package page on CRAN: [https://cran.r-project.org/web/packages/sem/index.html]
• lavaan package page on CRAN: [https://cran.r-project.org/web/packages/lavaan/index.html]
• OpenMx package page on Cran: [https://cran.r-project.org/web/packages/OpenMx/index.html]
• MVN package page on Cran: [https://cran.r-project.org/web/packages/MVN/index.html]
• semPlot package page on Cran: [https://cran.r-project.org/web/packages/semPlot/index.html]
Chapter 2

Prepare and Describe the Data

This chapter prepares the data set and does some univariate and multivariate description of its characteristics
prior to the CFA implementation in later chapters. Both numeric and graphical description and inference
about distribution shape are quickly available with R functions from the psych and MVN packages.

2.1 The Data Set

The 175 case data set (no missing observations) is loaded from a .csv file. The .csv file was exported from
the SPSS system file that is available from the website for the Tabachnick textbook (Tabachnick et al., 2019)
It has eleven subscales from the WISC:
• info (Information)
• comp (Comprehension)
• arith (Arithmetic)
• simil (Similarities)
• vocab (Vocabulary)
• digit (Digit Span)
• pictcomp (Picture Completion)
• parang (Picture Arrangement)
• block (Block Design)
• object(Object Assembly)
• coding (Coding - not sure if it is A or B, or a combination)
The user may recognize these scales as commonly discussed subtests of the WISC. The first 6 variables
comprise the set of manifest variables for the latent factor known as Verbal. The last five are associated
with Performance.
The original data file also contains an ID variable that is dropped for the working object created as wisc2
here.
# import the primary data file
wisc1 <- read.csv("wisc1.csv")
knitr::kable(head(wisc1),booktabs=TRUE,format="markdown")

ID info comp arith simil vocab digit pictcomp parang block object coding
3 8 7 13 9 12 9 6 11 12 7 9
4 9 6 8 7 11 12 6 8 7 12 14
5 13 18 11 16 15 6 18 8 11 12 9
6 8 11 6 12 9 7 13 4 7 12 11
7 10 3 8 9 12 9 7 7 11 4 10

6
2.2. NUMERIC AND GRAPHICAL DESCRIPTION OF THE DATA 7

ID info comp arith simil vocab digit pictcomp parang block object coding
8 11 7 15 12 10 12 6 12 10 5 10

# create the working data frame by removing the ID variable

wisc2 <- wisc1[,2:12]

A note about tables in this document: Many of the tables generated by the various R functions in this
document are reformatted so that they do not appear as the plain text that is typically output into the R
console. The kable function in the knitr package permits formatting that is well-rendered with rmarkdown
and bookdown document production. kable is used frequently.

2.2 Numeric and Graphical Description of the Data

We can explore univariate characteristics of the data with summaries, plots, and evaluation of normality
characteristics

2.2.1 Univariate descriptive statistics from the psych package.

knitr::kable(describe(wisc2,type=2,fast=T),booktabs=TRUE,format="markdown")

vars n mean sd min max range se

info 1 175 9.4971 2.9123 3 19 16 0.22015
comp 2 175 10.0000 2.9653 0 18 18 0.22416
arith 3 175 9.0000 2.3069 4 16 12 0.17439
simil 4 175 10.6114 3.1836 2 18 16 0.24066
vocab 5 175 10.7029 2.9327 2 19 17 0.22169
digit 6 175 8.7314 2.7042 0 16 16 0.20442
pictcomp 7 175 10.6800 2.9342 2 19 17 0.22181
parang 8 175 10.3714 2.6597 2 17 15 0.20105
block 9 175 10.3143 2.7098 2 18 16 0.20484
object 10 175 10.9029 2.8440 3 19 16 0.21498
coding 11 175 8.5486 2.8721 0 15 15 0.21711

2.2.2 Univariate Distribution Tests and Plots plus Evaluation of Multivariate

Normality
The MVN package provides univariate and multivariate normality tests. It is an efficient way to explore
characteristics of a set of variables. Several options are available for testing both univariate and Multivariate
normality. First, explicit calls for univariate and multivariate tests are made, and then an approach is shown
that obtains all at once plus a useful set of plots.
x_vars <- wisc2
# use the mvn function for an extensive evaluation
# note that different kinds of tests can be specified with changes in the arguments
result <- mvn(data= x_vars, mvnTest="mardia", univariateTest="AD")
kable(result$univariateNormality, booktabs=TRUE, format="markdown")

Test Variable Statistic p value Normality

Anderson-Darling info 1.2049 0.0037 NO
Anderson-Darling comp 1.3546 0.0016 NO
8 CHAPTER 2. PREPARE AND DESCRIBE THE DATA

Test Variable Statistic p value Normality

Anderson-Darling arith 1.8656 1e-04 NO
Anderson-Darling simil 0.9336 0.0175 NO
Anderson-Darling vocab 1.4992 7e-04 NO
Anderson-Darling digit 2.5087 <0.001 NO
Anderson-Darling pictcomp 1.1472 0.0052 NO
Anderson-Darling parang 1.3391 0.0017 NO
Anderson-Darling block 1.5225 6e-04 NO
Anderson-Darling object 1.1593 0.0049 NO
Anderson-Darling coding 1.2217 0.0034 NO

kable(result$multivariateNormality, booktabs=TRUE, format="markdown")

Test Statistic p value Result

Mardia Skewness 348.496907430775 0.00673137335852766 NO
Mardia Kurtosis 2.30789388829536 0.0210050390817529 NO
MVN NA NA NO

kable(mvn(data= x_vars,univariatePlot="histogram"), booktabs=TRUE, format="markdown")

Test Statistic p value Result

Mardia Skewness 348.496907430775 0.00673137335852766 NO
Mardia Kurtosis 2.30789388829536 0.0210050390817529 NO
MVN NA NA NO

Test Variable Statistic p value Normality

Shapiro-Wilk info 0.9814 0.0194 NO
Shapiro-Wilk comp 0.9810 0.0170 NO
Shapiro-Wilk arith 0.9709 0.0010 NO
Shapiro-Wilk simil 0.9866 0.0937 YES
Shapiro-Wilk vocab 0.9789 0.0093 NO
Shapiro-Wilk digit 0.9693 0.0007 NO
Shapiro-Wilk pictcomp 0.9836 0.0373 NO
Shapiro-Wilk parang 0.9814 0.0193 NO
Shapiro-Wilk block 0.9800 0.0127 NO
Shapiro-Wilk object 0.9853 0.0638 YES
Shapiro-Wilk coding 0.9820 0.0230 NO

n Mean Std.Dev Median Min Max 25th 75th Skew Kurtosis

info 175 9.4971 2.9123 10 3 19 8 11.5 0.08341 -0.08148
comp 175 10.0000 2.9653 10 0 18 8 12.0 0.08547 0.32620
arith 175 9.0000 2.3069 9 4 16 7 10.5 0.39097 -0.17897
simil 175 10.6114 3.1836 11 2 18 9 12.0 0.02233 -0.22985
vocab 175 10.7029 2.9327 10 2 19 9 12.0 0.26702 0.28751
digit 175 8.7314 2.7042 8 0 16 7 11.0 0.26631 0.07205
pictcomp 175 10.6800 2.9342 11 2 19 9 13.0 -0.07150 0.29116
parang 175 10.3714 2.6597 10 2 17 9 12.0 -0.19909 -0.05850
2.2. NUMERIC AND GRAPHICAL DESCRIPTION OF THE DATA 9

n Mean Std.Dev Median Min Max 25th 75th Skew Kurtosis

block 175 10.3143 2.7098 10 2 18 9 12.0 -0.21972 0.50196
object 175 10.9029 2.8440 11 3 19 9 13.0 -0.12289 0.14745
coding 175 8.5486 2.8721 9 0 15 6 11.0 -0.05283 -0.45376
Density

Density

Density

Density
0.00

0.00

0.00

0.00

5 10 20 0 5 10 4 8 12 16 5 10 15

info comp arith simil

Density

Density

Density

Density
0.00

0.00

0.00

0.00

5 10 20 0 5 10 15 5 10 20 5 10 15

vocab digit pictcomp parang

Density

Density

Density
0.00

0.00

0.00

5 10 15 5 10 20 0 5 10 15

block object coding

10 CHAPTER 2. PREPARE AND DESCRIBE THE DATA

2.2.3 Multivariate Outlier tests

The MVN package permits a good array of diagnostic tests/plots for univariate/multivariate shape and
outliers.

First, multivariate outliers are examined with the Mahalanobis distance:

result2 <- mvn(data=x_vars,multivariateOutlierMethod="quan")

Chi−Square Q−Q Plot

1
Outliers (n=16)
25

Non−outliers (n=159) 2
3
Chi−Square Quantile

4
5
6
87
20

11 10 9
12
13
14
1615
15
10

Quantile: 21.92
5

5 10 15 20 25 30 35

Robust Squared Mahalanobis Distance

2.2. NUMERIC AND GRAPHICAL DESCRIPTION OF THE DATA 11

Next, multivariate outliers are evaluated with the Adjusted-Mahalanobis distance:

result2 <- mvn(data=x_vars,multivariateOutlierMethod="adj")

Adjusted Chi−Square Q−Q Plot

1
Outliers (n=9)
25

Non−outliers (n=166) 2
3
Chi−Square Quantile

4
5
6
987
20
15
10

Quantile: 26.051
5

5 10 15 20 25 30

Robust Squared Mahalanobis Distance

12 CHAPTER 2. PREPARE AND DESCRIBE THE DATA

2.3 Bivariate Characteristics of the data set

We can quickly explore numerical and graphical summaries of the eleven variables.

2.3.1 SPLOM
Among the many scatterplot matrix capabilies in R, John Fox’ scatterplot.matrix function in his car
package has probably been seen by most students.
scatterplotMatrix(wisc2,cex=.2,
smooth=list(col.smooth="red", spread=F, lwd.smooth=.3),
col="skyblue1",
regLine=list(lwd=.3,col="black"))

0 10 5 15 0 10 5 15 5 15
info

5
comp
0

arith

4
simil
5

vocab

5
digit
0

pictcomp

5
parang
5

block
5
object
5

coding
0

5 15 4 12 5 15 5 15 5 15 0 10

Even with some control over colors and sizes of points/lines, this SPLOM probably has too many variables
to be effective - each plot is very small. Nonetheless, the sense of fairly linear relationships among all pairs
is somewhat apparent, as is the relative univariate normality of each of the eleven.

Note that the image can be enlarged if the reader is using a pdf version of this document simply by using
the increase/decrease size capability of pdf readers. If the user is reading an html version of this document,
then try to do a right mouse click on the image and “view image” (in Windows). Then the image can be
increased in size in a browser.
2.4. COVARIANCES AND ZERO ORDER CORRELATIONS 13

2.4 Covariances and Zero Order Correlations

The covariance matrix is the basic input for the CFA algorithms outlined in later chapters.
covmatrix1 <- round(cov(wisc2),digits=3)
knitr::kable(covmatrix1,booktabs=TRUE,format="markdown")

info comp arith simil vocab digit pictcomp parang block object coding
info 8.481 4.034 3.322 4.758 5.338 2.720 1.965 1.561 1.808 1.531 0.059
comp 4.034 8.793 2.684 4.816 4.621 1.891 3.540 1.471 2.966 2.718 0.517
arith 3.322 2.684 5.322 2.713 2.621 1.678 1.052 1.391 1.701 0.282 0.598
simil 4.758 4.816 2.713 10.136 5.022 2.234 3.450 2.524 2.255 2.433 -0.372
vocab 5.338 4.621 2.621 5.022 8.601 2.334 2.456 1.031 2.364 1.546 0.842
digit 2.720 1.891 1.678 2.234 2.334 7.313 0.597 1.066 0.533 0.267 1.344
pictcomp 1.965 3.540 1.052 3.450 2.456 0.597 8.610 1.941 3.038 3.032 -0.605
parang 1.561 1.471 1.391 2.524 1.031 1.066 1.941 7.074 2.532 1.916 0.289
block 1.808 2.966 1.701 2.255 2.364 0.533 3.038 2.532 7.343 3.077 0.832
object 1.531 2.718 0.282 2.433 1.546 0.267 3.032 1.916 3.077 8.088 0.433
coding 0.059 0.517 0.598 -0.372 0.842 1.344 -0.605 0.289 0.832 0.433 8.249
14 CHAPTER 2. PREPARE AND DESCRIBE THE DATA

We can use the Corrplot package to produce a useful combination of a schematic and correlation matrix.
mat1 <- cor(wisc2)
corrplot(mat1,type="upper",tl.pos="tp")
corrplot(mat1,add=T,type="lower", method="number",

pictcomp
col="black", diag=FALSE,tl.pos="n", cl.pos="n")

parang

coding
object
vocab
comp

block
simil
arith

digit
info
1
info
0.8
comp 0.47
0.6
arith 0.49 0.39

simil 0.51 0.51 0.37 0.4

vocab 0.63 0.53 0.39 0.54 0.2

digit 0.35 0.24 0.27 0.26 0.29 0

pictcomp 0.23 0.41 0.16 0.37 0.29 0.08 −0.2

parang 0.2 0.19 0.23 0.3 0.13 0.15 0.25 −0.4

block 0.23 0.37 0.27 0.26 0.3 0.07 0.38 0.35

−0.6
object 0.18 0.32 0.04 0.27 0.19 0.03 0.36 0.25 0.4
−0.8
coding 0.01 0.06 0.09−0.04 0.1 0.17−0.070.04 0.11 0.05
−1
Chapter 3

Using the lavaan package for CFA

One of the primary tools for SEM in R is the lavaan package. It permits path specification with a simple
syntax.

3.1 Implement the CFA, First Model

Using the lavaan package, we can implemnt directly the CFA with only a few steps. Since this document
contains three different packages’ approach to CFA, the packages used for each will be loaded at that point,
so as to not have confusion over common function names.
library(lavaan)

3.1.1 Define and fit the first model

The latent variables and their paths to the manifest variables are initially defined as simple textual specifi-
cations. The =~ symbol is the key to defining paths. We have two latent variables and no manifest variable
has duplicate paths from both latents. This is a so-called “simple structure.”
Note that this text string uses variable names that match what is in the wisc2 data set.
wisc.model1 <- "verbal =~ info + comp + arith + simil + vocab + digit
performance =~ pictcomp + parang + block + object + coding"

Fit the model and obtain the basic summary. Note that in this default approach, the latent factors are
permitted to covary and the model estimates this covariance.
One R syntax note….. the format here to call the cfa function (lavaan::cfa(.....)) is employed to ensure
no ambiguity that the correct cfa function is the one from the lavaan package. This precludes confusion
when multiple packages contain functions with the same name as is the case with both lavaan and sem
which is also used in this document. Even though sem is loaded later in this document, if there is a chance
that it may simultaneously exist in the R environment with lavaan then the approach here is safer.
fit1 <- lavaan::cfa(wisc.model1, data=wisc2,std.lv=TRUE)
summary(fit1, fit.measures=T,standardized=T)

## lavaan 0.6-3 ended normally after 24 iterations

##
## Optimization method NLMINB
## Number of free parameters 23
##
## Number of observations 175

15
16 CHAPTER 3. USING THE LAVAAN PACKAGE FOR CFA

##
## Estimator ML
## Model Fit Test Statistic 70.640
## Degrees of freedom 43
## P-value (Chi-square) 0.005
##
## Model test baseline model:
##
## Minimum Function Test Statistic 519.204
## Degrees of freedom 55
## P-value 0.000
##
## User model versus baseline model:
##
## Comparative Fit Index (CFI) 0.940
## Tucker-Lewis Index (TLI) 0.924
##
## Loglikelihood and Information Criteria:
##
## Loglikelihood user model (H0) -4491.822
## Loglikelihood unrestricted model (H1) -4456.502
##
## Number of free parameters 23
## Akaike (AIC) 9029.643
## Bayesian (BIC) 9102.433
## Sample-size adjusted Bayesian (BIC) 9029.600
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.061
## 90 Percent Confidence Interval 0.033 0.085
## P-value RMSEA <= 0.05 0.233
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.059
##
## Parameter Estimates:
##
## Information Expected
## Information saturated (h1) model Structured
## Standard Errors Standard
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## verbal =~
## info 2.206 0.200 11.029 0.000 2.206 0.760
## comp 2.042 0.210 9.709 0.000 2.042 0.691
## arith 1.300 0.172 7.555 0.000 1.300 0.565
## simil 2.232 0.225 9.940 0.000 2.232 0.703
## vocab 2.250 0.200 11.225 0.000 2.250 0.770
## digit 1.053 0.212 4.967 0.000 1.053 0.390
## performance =~
## pictcomp 1.742 0.242 7.187 0.000 1.742 0.595
3.1. IMPLEMENT THE CFA, FIRST MODEL 17

## parang 1.253 0.224 5.582 0.000 1.253 0.473

## block 1.846 0.222 8.311 0.000 1.846 0.683
## object 1.605 0.236 6.800 0.000 1.605 0.566
## coding 0.207 0.254 0.814 0.416 0.207 0.072
##
## Covariances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## verbal ~~
## performance 0.589 0.075 7.814 0.000 0.589 0.589
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .info 3.566 0.507 7.034 0.000 3.566 0.423
## .comp 4.572 0.585 7.815 0.000 4.572 0.523
## .arith 3.602 0.420 8.571 0.000 3.602 0.681
## .simil 5.096 0.662 7.702 0.000 5.096 0.506
## .vocab 3.487 0.506 6.886 0.000 3.487 0.408
## .digit 6.162 0.680 9.056 0.000 6.162 0.848
## .pictcomp 5.526 0.757 7.296 0.000 5.526 0.646
## .parang 5.463 0.658 8.298 0.000 5.463 0.777
## .block 3.894 0.640 6.083 0.000 3.894 0.533
## .object 5.467 0.719 7.600 0.000 5.467 0.680
## .coding 8.159 0.874 9.335 0.000 8.159 0.995
## verbal 1.000 1.000 1.000
## performance 1.000 1.000 1.000

3.1.2 Obtain coefficients

It is possible to obtain the output from the above summary function that does not contain the parameter
coefficients part of the table. If that had been the implementation, then the coefficients can be obtained
simply, but it is better to obtain the full table of parameter estimates and errors, etc.
# obtain only the coefficients
kable(coef(fit1), booktabs=TRUE, format="markdown")

3.1.3 Complete parameter listing

Instead of directly printing the parameter table, I prefer to reformat it with kable when using R Markdown.
Without using kable, the code would be the following:
parameterEstimates(fit1,standardized=T)

Format the table and clean it up using kable.

parameterEstimates(fit1, standardized=TRUE) %>%
filter(op == "=~") %>%
select('Latent Factor'=lhs, Indicator=rhs, B=est, SE=se, Z=z, 'p-value'=pvalue, Beta=std.all) %>%
knitr::kable(digits = 3, booktabs=TRUE, format="markdown", caption="Factor Loadings")

Latent Factor Indicator B SE Z p-value Beta

verbal info 2.206 0.200 11.029 0.000 0.760
verbal comp 2.042 0.210 9.709 0.000 0.691
verbal arith 1.300 0.172 7.555 0.000 0.565
verbal simil 2.232 0.225 9.940 0.000 0.703
verbal vocab 2.250 0.200 11.225 0.000 0.770
18 CHAPTER 3. USING THE LAVAAN PACKAGE FOR CFA

Latent Factor Indicator B SE Z p-value Beta

verbal digit 1.053 0.212 4.967 0.000 0.390
performance pictcomp 1.742 0.242 7.187 0.000 0.595
performance parang 1.253 0.224 5.582 0.000 0.473
performance block 1.846 0.222 8.311 0.000 0.683
performance object 1.605 0.236 6.800 0.000 0.566
performance coding 0.207 0.254 0.814 0.416 0.072

3.1.4 Residuals correlation matrix

Residuals can be examined.
cor_table <- residuals(fit1, type = "cor")$cov
#cor_table[upper.tri(cor_table)] <- # erase the upper triangle
#diag(cor_table) <- NA # erase the diagonal 0's
knitr::kable(cor_table, digits=3,format="markdown", booktabs=TRUE) # makes a nice table and rounds every

info comp arith simil vocab digit pictcomp parang block object coding
info 0.000 -0.058 0.065 -0.021 0.040 0.049 -0.036 -0.010 -0.076 -0.068 -0.025
comp -0.058 0.000 0.002 0.025 0.000 -0.034 0.165 -0.006 0.091 0.092 0.031
arith 0.065 0.002 0.000 -0.028 -0.047 0.048 -0.043 0.069 0.045 -0.145 0.066
simil -0.021 0.025 -0.028 0.000 -0.003 -0.015 0.123 0.102 -0.021 0.034 -0.071
vocab 0.040 0.000 -0.047 -0.003 0.000 -0.006 0.016 -0.082 -0.012 -0.071 0.067
digit 0.049 -0.034 0.048 -0.015 -0.006 0.000 -0.062 0.040 -0.084 -0.095 0.156
pictcomp -0.036 0.165 -0.043 0.123 0.016 -0.062 0.000 -0.033 -0.025 0.026 -0.115
parang -0.010 -0.006 0.069 0.102 -0.082 0.040 -0.033 0.000 0.028 -0.014 0.004
block -0.076 0.091 0.045 -0.021 -0.012 -0.084 -0.025 0.028 0.000 0.013 0.058
object -0.068 0.092 -0.145 0.034 -0.071 -0.095 0.026 -0.014 0.013 0.000 0.012
coding -0.025 0.031 0.066 -0.071 0.067 0.156 -0.115 0.004 0.058 0.012 0.000

3.1.5 Plot the residuals.

norm plot
# extract the residuals from the fit1 model
# get rid of the duplicates and diagonal values
# create a vector for a
res1 <- residuals(fit1, type = "cor")$cov
res1[upper.tri(res1,diag=T)] <- NA
v1 <- as.vector(res1)
v2 <- v1[!is.na(v1)]
qqPlot(v2,id=F)
3.1. IMPLEMENT THE CFA, FIRST MODEL 19

0.15
0.05
v2

−0.05
−0.15

−2 −1 0 1 2

norm quantiles

3.1.6 Modification Indices

Modification indices provide……
kable(modificationIndices(fit1, sort.=TRUE, minimum.value=3), booktabs=TRUE, format="markdown")

lhs op rhs mi epc sepc.lv sepc.all sepc.nox

32 performance =~ comp 9.8232 0.93721 0.93721 0.31696 0.31696
62 arith ~~ object 6.3624 -0.94453 -0.94453 -0.21285 -0.21285
37 info ~~ comp 5.2214 -0.98473 -0.98473 -0.24388 -0.24388
81 digit ~~ coding 4.9267 1.20723 1.20723 0.17025 0.17025
51 comp ~~ pictcomp 4.6854 0.96931 0.96931 0.19284 0.19284
85 pictcomp ~~ coding 4.5748 -1.22491 -1.22491 -0.18242 -0.18242
31 performance =~ info 4.4766 -0.59552 -0.59552 -0.20507 -0.20507
40 info ~~ vocab 4.4029 0.91242 0.91242 0.25875 0.25875
73 vocab ~~ parang 4.1459 -0.79773 -0.79773 -0.18277 -0.18277
38 info ~~ arith 4.1333 0.69702 0.69702 0.19450 0.19450
67 simil ~~ parang 3.3732 0.83050 0.83050 0.15741 0.15741
70 simil ~~ coding 3.2711 -0.95602 -0.95602 -0.14827 -0.14827
66 simil ~~ pictcomp 3.2695 0.86025 0.86025 0.16212 0.16212

3.1.7 Path Diagram for the bifactor Model 1

The semPlot package provides a capability of drawing path diagrams for cfa and other sem models. The
semPaths function will take a cfa model object and draw the diagram, with several options available. The
20 CHAPTER 3. USING THE LAVAAN PACKAGE FOR CFA

diagram produced here takes control over font/label sizes, display of residuals, and color of paths/coefficients.
These and many more control options are available. There is a challenge in producing these path diagrams
to have font sizes large enough for most humans to read. I’ve taken control of the font sizes for the “edges”
with a cex argument. But this causes overlap in the values if the default layout is used. I found that “circle2”
worked best here.
# Note that the base plot, including standardized path coefficients plots positive coefficients green
# and negative coefficients red. Red-green colorblindness issues anyone?
# I redrew it here to choose a blue and red. But all the coefficients in this example are
# positive,so they are shown with the skyblue.
# more challenging to use colors other than red and green. not in this doc
semPaths(fit1, residuals=F,sizeMan=7,"std",
posCol=c("skyblue4", "red"),
#edge.color="skyblue4",
edge.label.cex=1.2,layout="circle2")

cdn inf

obj cmp
0.07 0.76
0.57 0.69
blc art
0.68 0.57
prf 0.59 vrb

0.47 0.70

prn 0.60 0.77 sml

0.39

pct vcb
dgt

# or we could draw the paths in such a way to include the residuals:

#semPaths(fit1, sizeMan=7,"std",edge.color="skyblue4",edge.label.cex=1,layout="circle2")
# the base path diagram can be drawn much more simply:
#semPaths(fit1)
# or
semPaths(fit1,"std")
3.2. GENERATE A SECOND MODEL AND COMPARE 21

1.00 1.00

0.59

vrb prf

0.76 0.69 0.57 0.70 0.77 0.39 0.60 0.47 0.68 0.57 0.07

inf cmp art sml vcb dgt pct prn blc obj cdn

0.42 0.52 0.68 0.51 0.41 0.85 0.65 0.78 0.53 0.68 0.99

3.1.8 orthognal fit comparison

Compare to a one factor solution.
fit1orth <- lavaan::cfa(wisc.model1, data=wisc1,orthogonal=T)
kable(anova(fit1,fit1orth), booktabs=TRUE, format="markdown")

Df AIC BIC Chisq Chisq diff Df diff Pr(>Chisq)

fit1 43 9029.6 9102.4 70.64 NA NA NA
fit1orth 44 9065.6 9135.2 108.61 37.97 1 0

3.2 Generate a second model and compare

Next, generate a second CFA model. Thre are theoretical reasons why paths from both latent factors to
“comp” might be warranted. The “comp” variable also has the largest Modification index in the first model.
This second modelimplements this one-path/parmaeter change.

3.2.1 Add a path (Perf to comp) and Fit the second CFA model
Define the addditional path in the model text string.
wisc.model2 <- 'verbal =~ info + comp + arith + simil + vocab + digit
performance =~ pictcomp + parang + block + object + coding + comp'

Fit the model and obtain the basic summary

22 CHAPTER 3. USING THE LAVAAN PACKAGE FOR CFA

fit2 <- lavaan::cfa(wisc.model2, data=wisc1,std.lv=TRUE)

summary(fit2, fit.measures=T,standardized=T)

## lavaan 0.6-3 ended normally after 26 iterations

##
## Optimization method NLMINB
## Number of free parameters 24
##
## Number of observations 175
##
## Estimator ML
## Model Fit Test Statistic 60.642
## Degrees of freedom 42
## P-value (Chi-square) 0.031
##
## Model test baseline model:
##
## Minimum Function Test Statistic 519.204
## Degrees of freedom 55
## P-value 0.000
##
## User model versus baseline model:
##
## Comparative Fit Index (CFI) 0.960
## Tucker-Lewis Index (TLI) 0.947
##
## Loglikelihood and Information Criteria:
##
## Loglikelihood user model (H0) -4486.823
## Loglikelihood unrestricted model (H1) -4456.502
##
## Number of free parameters 24
## Akaike (AIC) 9021.645
## Bayesian (BIC) 9097.600
## Sample-size adjusted Bayesian (BIC) 9021.600
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.050
## 90 Percent Confidence Interval 0.016 0.077
## P-value RMSEA <= 0.05 0.465
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.054
##
## Parameter Estimates:
##
## Information Expected
## Information saturated (h1) model Structured
## Standard Errors Standard
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
3.2. GENERATE A SECOND MODEL AND COMPARE 23

## verbal =~
## info 2.256 0.199 11.318 0.000 2.256 0.777
## comp 1.491 0.254 5.877 0.000 1.491 0.504
## arith 1.307 0.172 7.584 0.000 1.307 0.568
## simil 2.205 0.226 9.748 0.000 2.205 0.695
## vocab 2.273 0.201 11.329 0.000 2.273 0.777
## digit 1.075 0.212 5.068 0.000 1.075 0.399
## performance =~
## pictcomp 1.790 0.239 7.495 0.000 1.790 0.612
## parang 1.189 0.224 5.317 0.000 1.189 0.448
## block 1.823 0.219 8.334 0.000 1.823 0.675
## object 1.633 0.233 7.010 0.000 1.633 0.576
## coding 0.200 0.253 0.793 0.428 0.200 0.070
## comp 0.884 0.266 3.324 0.001 0.884 0.299
##
## Covariances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## verbal ~~
## performance 0.533 0.081 6.594 0.000 0.533 0.533
##
## Variances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## .info 3.343 0.502 6.665 0.000 3.343 0.396
## .comp 4.331 0.554 7.819 0.000 4.331 0.495
## .arith 3.584 0.420 8.533 0.000 3.584 0.677
## .simil 5.217 0.675 7.726 0.000 5.217 0.518
## .vocab 3.383 0.508 6.655 0.000 3.383 0.396
## .digit 6.116 0.677 9.032 0.000 6.116 0.841
## .pictcomp 5.358 0.741 7.231 0.000 5.358 0.626
## .parang 5.620 0.663 8.480 0.000 5.620 0.799
## .block 3.979 0.623 6.385 0.000 3.979 0.545
## .object 5.376 0.707 7.603 0.000 5.376 0.668
## .coding 8.162 0.874 9.337 0.000 8.162 0.995
## verbal 1.000 1.000 1.000
## performance 1.000 1.000 1.000

3.2.2 Obtain coefficients

Coefficients can be obtained simply with the coef function, but it is better to do the full parameter listing
in the next section
knitr::kable(coef(fit2),booktabs=TRUE, format="markdown")

It is simple to obtain the full parameter list, but I prefer to use kable for tables when I can.
parameterEstimates(fit2,standardized=TRUE)

Reformat the table and clean it up by using kable.

parameterEstimates(fit2, standardized=TRUE) %>%
filter(op == "=~") %>%
select('Latent Factor'=lhs, Indicator=rhs, B=est, SE=se, Z=z, 'p-value'=pvalue, Beta=std.all) %>%
knitr::kable(digits = 3, format="markdown", booktabs=TRUE, caption="Factor Loadings")

Latent Factor Indicator B SE Z p-value Beta

verbal info 2.256 0.199 11.318 0.000 0.777
24 CHAPTER 3. USING THE LAVAAN PACKAGE FOR CFA

Latent Factor Indicator B SE Z p-value Beta

verbal comp 1.491 0.254 5.877 0.000 0.504
verbal arith 1.307 0.172 7.584 0.000 0.568
verbal simil 2.205 0.226 9.748 0.000 0.695
verbal vocab 2.273 0.201 11.329 0.000 0.777
verbal digit 1.075 0.212 5.068 0.000 0.399
performance pictcomp 1.790 0.239 7.495 0.000 0.612
performance parang 1.189 0.224 5.317 0.000 0.448
performance block 1.823 0.219 8.334 0.000 0.675
performance object 1.633 0.233 7.010 0.000 0.576
performance coding 0.200 0.253 0.793 0.428 0.070
performance comp 0.884 0.266 3.324 0.001 0.299

3.2.3 Residuals correlation matrix

Residuals can be examined.
cor_table2 <- residuals(fit2, type = "cor")$cov

#cor_table[upper.tri(cor_table)] <- # erase the upper triangle

#diag(cor_table) <- NA # erase the diagonal 0's
knitr::kable(cor_table2, digits=3,format="markdown",booktabs=TRUE) # makes a nice table and rounds every

info comp arith simil vocab digit pictcomp parang block object coding
info 0.000 -0.049 0.053 -0.026 0.021 0.036 -0.023 0.016 -0.050 -0.054 -0.022
comp -0.049 0.000 0.015 0.049 0.015 -0.029 0.059 -0.068 -0.014 -0.005 0.021
arith 0.053 0.015 0.000 -0.025 -0.054 0.043 -0.030 0.091 0.068 -0.131 0.069
simil -0.026 0.049 -0.025 0.000 -0.002 -0.017 0.143 0.132 0.012 0.056 -0.067
vocab 0.021 0.015 -0.054 -0.002 0.000 -0.016 0.032 -0.054 0.018 -0.053 0.071
digit 0.036 -0.029 0.043 -0.017 -0.016 0.000 -0.055 0.053 -0.071 -0.088 0.158
pictcomp -0.023 0.059 -0.030 0.143 0.032 -0.055 0.000 -0.025 -0.031 0.011 -0.115
parang 0.016 -0.068 0.091 0.132 -0.054 0.053 -0.025 0.000 0.049 -0.005 0.006
block -0.050 -0.014 0.068 0.012 0.018 -0.071 -0.031 0.049 0.000 0.011 0.060
object -0.054 -0.005 -0.131 0.056 -0.053 -0.088 0.011 -0.005 0.011 0.000 0.013
coding -0.022 0.021 0.069 -0.067 0.071 0.158 -0.115 0.006 0.060 0.013 0.000

3.2.4 Modification Indices for Model 2

Modification indices provide………………..
kable(modificationIndices(fit2, sort.=TRUE, minimum.value=3), booktabs=TRUE, format="markdown")

lhs op rhs mi epc sepc.lv sepc.all sepc.nox

34 performance =~ simil 6.7637 0.79457 0.79457 0.25030 0.25030
62 arith ~~ object 5.5514 -0.87412 -0.87412 -0.19915 -0.19915
81 digit ~~ coding 4.9756 1.20983 1.20983 0.17124 0.17124
85 pictcomp ~~ coding 4.7421 -1.23384 -1.23384 -0.18659 -0.18659
66 simil ~~ pictcomp 3.8109 0.92525 0.92525 0.17501 0.17501
52 comp ~~ parang 3.7688 -0.85250 -0.85250 -0.17279 -0.17279
73 vocab ~~ parang 3.6900 -0.75282 -0.75282 -0.17265 -0.17265
67 simil ~~ parang 3.5789 0.86835 0.86835 0.16037 0.16037
57 arith ~~ vocab 3.3775 -0.64063 -0.64063 -0.18399 -0.18399
3.2. GENERATE A SECOND MODEL AND COMPARE 25

lhs op rhs mi epc sepc.lv sepc.all sepc.nox

61 arith ~~ block 3.2569 0.60993 0.60993 0.16152 0.16152
38 info ~~ arith 3.2166 0.62088 0.62088 0.17938 0.17938
70 simil ~~ coding 3.1652 -0.94894 -0.94894 -0.14543 -0.14543

3.2.5 Path Diagram for Model 2

The semPlot package provides a capability of drawing path diagrams for cfa and other sem models. The
semPaths function will take a cfa model object and draw the diagram, with several options available. The
diagram produced here takes control over font/label sizes, display of residuals, and color of paths/coefficients.
These and many more control options are available. There is a challenge in producing these path diagrams
to have font sizes large enough for most humans to read. I’ve taken control of the font sizes for the “edges”
with a cex argument. But this causes overlap in the values if the default layout is used. I found that “circle2”
worked best here.
# Note that the base plot, including standardized path coefficients plots positive coefficients green
# and negative coefficients red. Red-green colorblindness issues anyone?
# I redrew it here to choose a blue and red. But all the coefficients in this example are
# positive,so they are shown with the skyblue.
# more challenging to use colors other than red and green. not in this doc
semPaths(fit2, residuals=F,sizeMan=7,"std",
posCol=c("skyblue4", "red"),
#edge.color="skyblue4",
edge.label.cex=1.2,layout="circle2")

cdn inf

obj art
0.07 0.78
0.58 0.57
blc sml
0.67 0.69
prf 0.53 vrb

0.45 0.78

prn 0.61 0.40 vcb

0.30 0.50

pct dgt
cmp
26 CHAPTER 3. USING THE LAVAAN PACKAGE FOR CFA

# or we could draw the paths in such a way to include the residuals:

#semPaths(fit1, sizeMan=7,"std",edge.color="skyblue4",edge.label.cex=1,layout="circle2")
# the base path diagram can be drawn much more simply:
#semPaths(fit2)
# or
semPaths(fit2,"std")

1.00 1.00

0.53

vrb prf

0.78 0.57 0.69 0.78 0.40 0.50 0.30 0.61 0.45 0.67 0.58 0.07

inf art sml vcb dgt cmp pct prn blc obj cdn

0.40 0.68 0.52 0.40 0.84 0.50 0.63 0.80 0.54 0.67 1.00

3.3 Compare Model 1 and Model 2

Model 1 is first/base. Model 2 adds a path from Perf to comp… Compare these models.
kable(anova(fit1,fit2), booktabs=TRUE, format="markdown")

Df AIC BIC Chisq Chisq diff Df diff Pr(>Chisq)

fit2 42 9021.6 9097.6 60.642 NA NA NA
fit1 43 9029.6 9102.4 70.640 9.998 1 0.00157

3.4 An additional perspective on estimation and optimization

Subtle differences in algorithmic strategies for Structural Equation Modeling exist among software packages.
Users are often familiar with only the default approaches and will only be moved to learn other approaches
when the default approach fails to converge or produces problematic models. The student first confronted
with these methods will often assume that the same SEM model evauated in different software (e.g., LIS-
REL, MPlus, EQS, lavaan, sem, OpenMx) will produce the same model outcome. This may not be a safe
3.4. AN ADDITIONAL PERSPECTIVE ON ESTIMATION AND OPTIMIZATION 27

assumption. Different “dialects” exist for the various software products. The commercial products may use
algorithms that are proprietary and not available to understand their approach. The open source products
(e.g., lavaan) have made source code available for inspection.
The perceptive reader will notice that the default solutions given by the three R packages employed in this
document (lavaan, sem, OpenMx) all give the same values for parameter estimates and goodnes of fit
statistics for the same two models run with each. However, there are slight differences in these quantities
compared to the published LISREL analysis of the same data in the Tabachnik, et al textbook ((2019)).
Other readers will have noticed that the LISREL output in this textbook matches SAS output that they
may have seen with class coverage of the CFA topic. The R packages, while agreeing with each other, vary
slightly from the LISREL and SAS values. The degree of difference is slight, but its existence may puzzle
some students. The answer to understanding these differences goes well beyond the scope of this document
and involves an advanced understanding of the computational algorithms employed. Even though all of the
approaches are Maximum Liklihood methods some optimization and estimation strategies can differ.
The lavaan package permits some insight into this with one of the argments available for the cfa function
used in this chapter. The reader might want to examine the help docs for this function: ?cfa. That help
page directs readers to another help document on Lavaan Options (lavoptions). There, one can find an
argument that can be passed to cfa, called “mimic”. Here is the text from that section, describing the
various “mimic” possibilities:
“If”Mplus“, an attempt is made to mimic the Mplus program. If”EQS“, an attempt is made to mimic the
EQS program. If”default“, the value is (currently) set to to”lavaan“, which is very close to”Mplus“.”
The reader may have been exposed to a SAS Proc Calis approach to this problem that employed the default
Method called LINEQS. The following rerun of the first model from this chapter above, employs the mimic
argument to be specified as “EQS”. The product of this model is a set of parameter values and fit statistics
(e.g., Chi Sq) that match the SAS output (and the Tabachnick et al LISREL output) exactly. Demonstrating
the equivalence with the addition of the mimic argument does not fully explain why such differences originally
existed, but that, again, is well beyond the scope of this document. The reference section to this document
includes some articles that address these differences (El-Sheikh et al., 2017; Narayanan, 2012). Rosseel
(2012) has discussed use of the ‘mimic’ function that guides lavaan to emulate the approach of some of the
commercial products. His website is also a valuable resource on this [http://lavaan.ugent.be/].
wisc.model1 <- "verbal =~ info + comp + arith + simil + vocab + digit
performance =~ pictcomp + parang + block + object + coding"

fit1eqs <- lavaan::cfa(wisc.model1, data=wisc2,std.lv=TRUE, mimic="EQS")

summary(fit1eqs, fit.measures=T,standardized=T)

## lavaan 0.6-3 ended normally after 25 iterations

##
## Optimization method NLMINB
## Number of free parameters 23
##
## Number of observations 175
##
## Estimator ML
## Model Fit Test Statistic 70.236
## Degrees of freedom 43
## P-value (Chi-square) 0.005
##
## Model test baseline model:
##
## Minimum Function Test Statistic 516.237
## Degrees of freedom 55
## P-value 0.000
28 CHAPTER 3. USING THE LAVAAN PACKAGE FOR CFA

##
## User model versus baseline model:
##
## Comparative Fit Index (CFI) 0.941
## Tucker-Lewis Index (TLI) 0.924
##
## Loglikelihood and Information Criteria:
##
## Loglikelihood user model (H0) -4497.337
## Loglikelihood unrestricted model (H1) -4462.018
##
## Number of free parameters 23
## Akaike (AIC) 9040.675
## Bayesian (BIC) 9113.465
## Sample-size adjusted Bayesian (BIC) 9040.631
##
## Root Mean Square Error of Approximation:
##
## RMSEA 0.060
## 90 Percent Confidence Interval 0.033 0.085
## P-value RMSEA <= 0.05 0.239
##
## Standardized Root Mean Square Residual:
##
## SRMR 0.059
##
## Parameter Estimates:
##
## Information Expected
## Information saturated (h1) model Structured
## Standard Errors Standard
##
## Latent Variables:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## verbal =~
## info 2.212 0.201 10.997 0.000 2.212 0.760
## comp 2.048 0.212 9.682 0.000 2.048 0.691
## arith 1.304 0.173 7.534 0.000 1.304 0.565
## simil 2.238 0.226 9.911 0.000 2.238 0.703
## vocab 2.257 0.202 11.193 0.000 2.257 0.770
## digit 1.056 0.213 4.952 0.000 1.056 0.390
## performance =~
## pictcomp 1.747 0.244 7.166 0.000 1.747 0.595
## parang 1.257 0.226 5.566 0.000 1.257 0.473
## block 1.851 0.223 8.287 0.000 1.851 0.683
## object 1.609 0.237 6.780 0.000 1.609 0.566
## coding 0.208 0.256 0.811 0.417 0.208 0.072
##
## Covariances:
## Estimate Std.Err z-value P(>|z|) Std.lv Std.all
## verbal ~~
## performance 0.589 0.076 7.792 0.000 0.589 0.589
##
## Variances:
3.4. AN ADDITIONAL PERSPECTIVE ON ESTIMATION AND OPTIMIZATION 29

## Estimate Std.Err z-value P(>|z|) Std.lv Std.all

## .info 3.586 0.511 7.014 0.000 3.586 0.423
## .comp 4.599 0.590 7.793 0.000 4.599 0.523
## .arith 3.623 0.424 8.547 0.000 3.623 0.681
## .simil 5.125 0.667 7.680 0.000 5.125 0.506
## .vocab 3.507 0.511 6.866 0.000 3.507 0.408
## .digit 6.198 0.686 9.030 0.000 6.198 0.848
## .pictcomp 5.558 0.764 7.276 0.000 5.558 0.646
## .parang 5.494 0.664 8.275 0.000 5.494 0.777
## .block 3.916 0.646 6.066 0.000 3.916 0.533
## .object 5.499 0.726 7.578 0.000 5.499 0.680
## .coding 8.206 0.882 9.309 0.000 8.206 0.995
## verbal 1.000 1.000 1.000
## performance 1.000 1.000 1.000
Chapter 4

Using the sem package for CFA

In this chapter, we use the sem package to implement the same two CFA analyses that we produced with
lavaan in chapter 3. sem provides an equally simple way to obtain the models and only the basics are
shown here. The code in this chapter is modeled after a document by James Steiger

4.1 Example one

Once again, the bifactor model with Verbal and Performance factors is specified. Each manifest factor has
a path from only one of the two factors.

4.1.1 Data Setup

In sem, it is helpful to have covariance and correlation matrices available as objects, as well ans a sample
size object.
# same data file and extraction of the wisc2 data frame with only the 11 manifests
#wisc1 <- read.csv("wisc1.csv")
#wisc2 <- wisc1[,2:12]
# covariance and correlation matrices are saved as objects
covwisc <- cov(wisc2)
corwisc <-cor(wisc2)
# list of manifest variables for potential use
manifests <- names(wisc2)
# this gives an object that is the sample size
wobs <- length(wisc2)

We also need to load the package.

library(sem)

4.1.2 Define the first model

In sem, the structure of the model is created with a text string to define the paths. The specifyModel
function permits this in several ways. The simplest is to enter text as an argument. Some explanation of
the strucure is needed.
• Each line defines a path, a label for the parameter, and the starting value for the parameter value.

• This is symbolized as: , , .

30
4.1. EXAMPLE ONE 31

• Note that paths can be double or single-headed arrows.

• The unique name specified by the parameter symbol is for free parameters.

• If the parameter value is NA, then its starting point value is system determined.

• A numerical value following the NA can fix a variance at the value. E.g.,: F1 <-> F1, NA, 1 would
fix the factor variance a 1.
• Unique variances can be specified for manifest variables. e.g.,: manifest1 <-> manifest1, e1, NA. If
they are not specified, they default to “free to vary”
• Factors can be set to a fixed relationship to each other, e.g, 0, or 1. Or they can be left free (estimable)
as in the example here.
# this text could have been saved in a file and read in with the file argument to be efficient
# commented here to show the argument options
# m1.model <- specifyModel(file="sem1.txt")
m1.model <- specifyModel(text="
## Factor 1 is Verbal
Verbal -> info, t01, NA
Verbal -> comp, t02, NA
Verbal -> arith, t03, NA
Verbal -> simil, t04, NA
Verbal -> vocab, t05, NA
Verbal -> digit, t06, NA
## Factor 2 is performance
Performance -> pictcomp, t07, NA
Performance -> parang, t08, NA
Performance -> block, t09, NA
Performance -> object, t10, NA
Performance -> coding, t11, NA
## Set factor variances
Verbal <-> Verbal, NA, 1
Performance <-> Performance, NA, 1
## Set factor covariance to be estimable
Verbal <-> Performance, p1, NA"
)

## NOTE: it is generally simpler to use specifyEquations() or cfa()

## see ?specifyEquations

## NOTE: adding 11 variances to the model

This m1.model is now available to be used in the model function. The first “note” refers to the fact that the
model can be specified interactivly (I think). I found it easier to do this way, and more reproducible.
32 CHAPTER 4. USING THE SEM PACKAGE FOR CFA

4.1.3 Fit model 1 and examine the results

In this chapter, we will just look at the basics of the model fit and draw the path diagram. Other aspects of
the model can be extracted, but are passed over here to save space. The interested reader might try str(m1)
to see what is available from the model object.
# all that is required is the model specification, the covariance matrix, and the sample size
m1 <- sem(m1.model,covwisc,175)
options(digits=5)
summary(m1)

##
## Model Chisquare = 70.236 Df = 43 Pr(>Chisq) = 0.0054454
## AIC = 116.24
## BIC = -151.85
##
## Normalized Residuals
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## -1.88413 -0.41375 -0.00001 0.03511 0.45976 2.11208
##
## R-square for Endogenous Variables
## info comp arith simil vocab digit pictcomp parang
## 0.5772 0.4770 0.3193 0.4944 0.5922 0.1524 0.3545 0.2233
## block object coding
## 0.4667 0.3202 0.0052
##
## Parameter Estimates
## Estimate Std Error z value Pr(>|z|)
## t01 2.21249 0.201190 10.99700 3.9507e-28
## t02 2.04806 0.211541 9.68163 3.6090e-22
## t03 1.30357 0.173035 7.53358 4.9367e-14
## t04 2.23845 0.225846 9.91142 3.7135e-23
## t05 2.25691 0.201642 11.19269 4.4281e-29
## t06 1.05579 0.213185 4.95244 7.3287e-07
## t07 1.74699 0.243776 7.16638 7.7008e-13
## t08 1.25683 0.225785 5.56647 2.5995e-08
## t09 1.85123 0.223392 8.28693 1.1621e-16
## t10 1.60918 0.237324 6.78052 1.1974e-11
## t11 0.20761 0.255890 0.81132 4.1718e-01
## p1 0.58883 0.075569 7.79198 6.5965e-15
## V[info] 3.58620 0.511281 7.01414 2.3137e-12
## V[comp] 4.59856 0.590106 7.79277 6.5556e-15
## V[arith] 3.62254 0.423850 8.54675 1.2660e-17
## V[simil] 5.12484 0.667296 7.68001 1.5908e-14
## V[vocab] 3.50720 0.510787 6.86626 6.5906e-12
## V[digit] 6.19783 0.686345 9.03020 1.7136e-19
## V[pictcomp] 5.55770 0.763882 7.27560 3.4488e-13
## V[parang] 5.49428 0.663998 8.27454 1.2895e-16
## V[block] 3.91612 0.645638 6.06550 1.3154e-09
## V[object] 5.49872 0.725619 7.57798 3.5099e-14
## V[coding] 8.20596 0.881552 9.30853 1.2961e-20
##
## t01 info <--- Verbal
## t02 comp <--- Verbal
## t03 arith <--- Verbal
4.1. EXAMPLE ONE 33

## t04 simil <--- Verbal

## t05 vocab <--- Verbal
## t06 digit <--- Verbal
## t07 pictcomp <--- Performance
## t08 parang <--- Performance
## t09 block <--- Performance
## t10 object <--- Performance
## t11 coding <--- Performance
## p1 Performance <--> Verbal
## V[info] info <--> info
## V[comp] comp <--> comp
## V[arith] arith <--> arith
## V[simil] simil <--> simil
## V[vocab] vocab <--> vocab
## V[digit] digit <--> digit
## V[pictcomp] pictcomp <--> pictcomp
## V[parang] parang <--> parang
## V[block] block <--> block
## V[object] object <--> object
## V[coding] coding <--> coding
##
## Iterations = 74

The table listing of parameter estimates uses the labeling strategy that was defined in the modelSpecify
function. The names, such as theta01 are arbitrary.
34 CHAPTER 4. USING THE SEM PACKAGE FOR CFA

4.1.4 Can we draw the path diagram from a sem model?

The semPaths function from semPlot is capable of recognizing a model object from sem. This code is
identical to what was used in chapter 2 for the lavaan object. The fit object name is just changed to m1
here.
# Note that the base plot, including standardized path coefficients plots positive coefficients green
# and negative coefficients red. Red-green colorblindness issues anyone?
# I redrew it here to choose a blue and red. But all the coefficients in this example are
# positive,so they are shown with the skyblue.
# more challenging to use colors other than red and green. not in this doc
semPaths(m1, residuals=F,sizeMan=7,"std",
posCol=c("skyblue4", "red"),
#edge.color="skyblue4",
edge.label.cex=1.2,layout="circle2")

cdn inf

obj cmp
0.07 0.76
0.57 0.69
blc art
0.68 0.57
Prf 0.59 Vrb

0.47 0.70

prn 0.60 0.77 sml

0.39

pct vcb
dgt

# or we could draw the paths in such a way to include the residuals:

#semPaths(m1, sizeMan=7,"std",edge.color="skyblue4",edge.label.cex=1,layout="circle2")
# the base path diagram can be drawn much more simply:
#semPaths(m1)
# or
semPaths(m1,"std")
4.1. EXAMPLE ONE 35

1.00 1.00

0.59

Vrb Prf

0.76 0.69 0.57 0.70 0.77 0.39 0.60 0.47 0.68 0.57 0.07

inf cmp art sml vcb dgt pct prn blc obj cdn

0.42 0.52 0.68 0.51 0.41 0.85 0.65 0.78 0.53 0.68 0.99
36 CHAPTER 4. USING THE SEM PACKAGE FOR CFA

4.2 Example two

Model with added path still under construction in this chapter.

4.2.1 Define the second model

In the specifyModel text argument we just need to add one path, from Performance to “comp”, rerun the
model, and compare to the first.
# this text could have been saved in a file and read in with the file argument to be efficient
# commented here to show the argument options
# m1.model <- specifyModel(file="sem2.txt")
m2.model <- specifyModel(text="
## Factor 1 is Verbal
Verbal -> info, t01, NA
Verbal -> comp, t02, NA
Verbal -> arith, t03, NA
Verbal -> simil, t04, NA
Verbal -> vocab, t05, NA
Verbal -> digit, t06, NA
## Factor 2 is performance
Performance -> pictcomp, t07, NA
Performance -> parang, t08, NA
Performance -> block, t09, NA
Performance -> object, t10, NA
Performance -> coding, t11, NA
Performance -> comp, t12, NA
## Set factor variances
Verbal <-> Verbal, NA, 1
Performance <-> Performance, NA, 1
## Set factor covariance to be estimable
Verbal <-> Performance, p1, NA"
)

## NOTE: it is generally simpler to use specifyEquations() or cfa()

## see ?specifyEquations
## NOTE: adding 11 variances to the model

4.2.2 Fit model 2 and examine the results

In this chapter, we will just look at the basics of the model fit and draw the path diagram. Other aspects of
the model can be extracted, but are passed over here to save space. The interested reader might try str(m1)
to see what is available from the model object.
# all that is required is the model specification, the covariance matrix, and the sample size
m2 <- sem(m2.model,covwisc,175)
options(digits=5)
summary(m2)

##
## Model Chisquare = 60.295 Df = 42 Pr(>Chisq) = 0.033354
## AIC = 108.3
## BIC = -156.63
##
## Normalized Residuals
## Min. 1st Qu. Median Mean 3rd Qu. Max.
4.2. EXAMPLE TWO 37

## -1.7087 -0.3670 0.0000 0.0439 0.4509 2.0855

##
## R-square for Endogenous Variables
## info comp arith simil vocab digit pictcomp parang
## 0.6036 0.5046 0.3227 0.4823 0.6044 0.1589 0.3741 0.2009
## block object coding
## 0.4551 0.3315 0.0049
##
## Parameter Estimates
## Estimate Std Error z value Pr(>|z|)
## t01 2.26254 0.200471 11.28609 1.5373e-29
## t02 1.49558 0.255230 5.85973 4.6361e-09
## t03 1.31053 0.173300 7.56223 3.9620e-14
## t04 2.21107 0.227463 9.72056 2.4641e-22
## t05 2.28001 0.201829 11.29675 1.3616e-29
## t06 1.07784 0.213286 5.05351 4.3377e-07
## t07 1.79476 0.240149 7.47352 7.8078e-14
## t08 1.19223 0.224879 5.30164 1.1477e-07
## t09 1.82799 0.219975 8.30998 9.5719e-17
## t10 1.63752 0.234254 6.99037 2.7416e-12
## t11 0.20104 0.254172 0.79098 4.2896e-01
## t12 0.88667 0.267519 3.31443 9.1829e-04
## p1 0.53322 0.081099 6.57487 4.8695e-11
## V[info] 3.36224 0.505946 6.64546 3.0227e-11
## V[comp] 4.35598 0.558735 7.79614 6.3830e-15
## V[arith] 3.60434 0.423598 8.50888 1.7563e-17
## V[simil] 5.24665 0.681040 7.70388 1.3199e-14
## V[vocab] 3.40241 0.512709 6.63614 3.2200e-11
## V[digit] 6.15077 0.682983 9.00574 2.1421e-19
## V[pictcomp] 5.38850 0.747358 7.21007 5.5922e-13
## V[parang] 5.65249 0.668468 8.45589 2.7697e-17
## V[block] 4.00164 0.628568 6.36628 1.9367e-10
## V[object] 5.40673 0.713201 7.58094 3.4306e-14
## V[coding] 8.20865 0.881648 9.31057 1.2715e-20
##
## t01 info <--- Verbal
## t02 comp <--- Verbal
## t03 arith <--- Verbal
## t04 simil <--- Verbal
## t05 vocab <--- Verbal
## t06 digit <--- Verbal
## t07 pictcomp <--- Performance
## t08 parang <--- Performance
## t09 block <--- Performance
## t10 object <--- Performance
## t11 coding <--- Performance
## t12 comp <--- Performance
## p1 Performance <--> Verbal
## V[info] info <--> info
## V[comp] comp <--> comp
## V[arith] arith <--> arith
## V[simil] simil <--> simil
## V[vocab] vocab <--> vocab
## V[digit] digit <--> digit
38 CHAPTER 4. USING THE SEM PACKAGE FOR CFA

## V[pictcomp] pictcomp <--> pictcomp

## V[parang] parang <--> parang
## V[block] block <--> block
## V[object] object <--> object
## V[coding] coding <--> coding
##
## Iterations = 37

The table listing of parameter estimates uses the labeling strategy that was defined in the modelSpecify
function. The names, such as theta01 are arbitrary.
4.2. EXAMPLE TWO 39

4.2.3 Can we draw the path diagram from a sem model?

The semPaths function from semPlot is capable of recognizing a model object from sem. This code is
identical to what was used in chapter 2 for the lavaan object. The fit object name is just changed to m2
here.
# Note that the base plot, including standardized path coefficients plots positive coefficients green
# and negative coefficients red. Red-green colorblindness issues anyone?
# I redrew it here to choose a blue and red. But all the coefficients in this example are
# positive,so they are shown with the skyblue.
# more challenging to use colors other than red and green. not in this doc
semPaths(m2, residuals=F,sizeMan=7,"std",
posCol=c("skyblue4", "red"),
#edge.color="skyblue4",
edge.label.cex=1.2,layout="circle2")

cdn inf

obj art
0.07 0.78
0.58 0.57
blc sml
0.67 0.69
Prf 0.53 Vrb

0.45 0.78

prn 0.61 0.40 vcb

0.30 0.50

pct dgt
cmp

# or we could draw the paths in such a way to include the residuals:

#semPaths(m1, sizeMan=7,"std",edge.color="skyblue4",edge.label.cex=1,layout="circle2")
# the base path diagram can be drawn much more simply:
#semPaths(m1)
# or
semPaths(m2,"std")
40 CHAPTER 4. USING THE SEM PACKAGE FOR CFA

1.00 1.00

0.53

Vrb Prf

0.78 0.57 0.69 0.78 0.40 0.50 0.30 0.61 0.45 0.67 0.58 0.07

inf art sml vcb dgt cmp pct prn blc obj cdn

0.40 0.68 0.52 0.40 0.84 0.50 0.63 0.80 0.54 0.67 1.00

4.3 Compare the two CFA models produced by sem

We can use the anova function to compare the two models, as we did for lavaan.
kable(anova(m2,m1), booktabs=TRUE, format="markdown")

Model Df Model Chisq Df LR Chisq Pr(>Chisq)

m2 42 60.295 NA NA NA
m1 43 70.236 1 9.9409 0.00162
Chapter 5

Using the OpenMx Package for CFA

The OpenMX package in R is a port of the well-respected MX analytical software. It handles SEM and
can easily be used for CFA. Here, the same two models that were run in lavaan will be run again, but an
additional model will be run first.
library(OpenMx)

5.1 First OpenMx Model - Single Factor

First, fit a bad model that posits only one underlying factor

5.1.1 set new dataframe and set up basics

Some initial setups
# grab data
#wisc2 <- wisc1[2:12] # remove ID from df
# set up basics
manifests=names(wisc2)
observedCov <- cov(wisc2)
numSubjects <- nrow(wisc2)

5.1.2 Create the model using the mxModel function

This code initializes the model and sets vars/paths
latents="F1"
cfa2a <- mxModel("Common Factor Model",type="RAM",
manifestVars = manifests, latentVars = latents,
# Now set the residual variance for manifest variables
mxPath(from=manifests, arrows=2,free=T,values=1,labels=paste("error",1:11,sep="")),
# set latent factor variance to 1
mxPath(from="F1",arrows=2,free=F,values=1,labels="varF1"),
# specify factor loadings
mxPath(from="F1",to=manifests, arrows=1,
free=T,values=1,labels=paste("i", 1:11,sep="")),
# specify the covariance matrix
mxData(observed=observedCov,type="cov",numObs=numSubjects)
) # close the model

41
42 CHAPTER 5. USING THE OPENMX PACKAGE FOR CFA

5.1.3 Now use mxRun to fit the model

mxRun uses the model defined above.
cfa2a <- mxRun(cfa2a)

## Running Common Factor Model with 22 parameters

Summarize the fit:

summary(cfa2a)

## Summary of Common Factor Model

##
## free parameters:
## name matrix row col Estimate Std.Error A
## 1 i1 A info F1 2.10843 0.20489
## 2 i2 A comp F1 2.10258 0.20876
## 3 i3 A arith F1 1.27253 0.17310
## 4 i4 A simil F1 2.25978 0.22323
## 5 i5 A vocab F1 2.17498 0.20342
## 6 i6 A digit F1 1.00892 0.21308
## 7 i7 A pictcomp F1 1.35055 0.22867
## 8 i8 A parang F1 0.89834 0.21227
## 9 i9 A block F1 1.23160 0.21130
## 10 i10 A object F1 1.01784 0.22717
## 11 i11 A coding F1 0.20128 0.23480
## 12 error1 S info info 3.98738 0.54543
## 13 error2 S comp comp 4.32199 0.56950
## 14 error3 S arith arith 3.67209 0.42704
## 15 error4 S simil simil 4.97100 0.64992
## 16 error5 S vocab vocab 3.82117 0.53031
## 17 error6 S digit digit 6.25280 0.68878
## 18 error7 S pictcomp pictcomp 6.73647 0.76246
## 19 error8 S parang parang 6.22646 0.68288
## 20 error9 S block block 5.78440 0.65307
## 21 error10 S object object 7.00600 0.77245
## 22 error11 S coding coding 8.16142 0.87333
##
## Model Statistics:
## | Parameters | Degrees of Freedom | Fit (-2lnL units)
## Model: 22 44 5492.0
## Saturated: 66 0 5375.1
## Independence: 11 55 5894.3
## Number of observations/statistics: 175/66
##
## chi-square: <U+03C7>² ( df=44 ) = 116.85, p = 1.5993e-08
## Information Criteria:
## | df Penalty | Parameters Penalty | Sample-Size Adjusted
## AIC: 28.851 160.85 167.51
## BIC: -110.400 230.48 160.81
## CFI: 0.84306
## TLI: 0.80383 (also known as NNFI)
## RMSEA: 0.097268 [95% CI (0.071763, 0.12282)]
## Prob(RMSEA <= 0.05): 0.00026603
## timestamp: 2019-07-11 11:27:19
5.1. FIRST OPENMX MODEL - SINGLE FACTOR 43

## Wall clock time: 0.055 secs

## optimizer: CSOLNP
## OpenMx version number: 2.12.2
## Need help? See help(mxSummary)

What is in the model?

slotNames(cfa2a@output)

## NULL
names(cfa2a@output)

## [1] "matrices" "algebras"

## [3] "data" "SaturatedLikelihood"
## [5] "IndependenceLikelihood" "calculatedHessian"
## [7] "standardErrors" "gradient"
## [9] "hessian" "expectations"
## [11] "fit" "fitUnits"
## [13] "Minus2LogLikelihood" "maxRelativeOrdinalError"
## [15] "minimum" "estimate"
## [17] "infoDefinite" "conditionNumber"
## [19] "status" "iterations"
## [21] "evaluations" "mxVersion"
## [23] "frontendTime" "backendTime"
## [25] "independentTime" "wallTime"
## [27] "timestamp" "cpuTime"
44 CHAPTER 5. USING THE OPENMX PACKAGE FOR CFA

5.2 Second OpenMx Model - the bifactor model

Now fit a model that is the same as the initial model fit with lavaan in chapter 3. Two factors, verbal and
performance are established with each manifest variable uniquely specified by only one of the two latent
factors.

5.2.1 set new dataframe and set up basics

Same initial setups as above.

# grab data
wisc2 <- wisc1[2:12] # remove ID from df
# set up basics
manifests <- names(wisc2)
verbalVars <- names(wisc2[1:6])
perfVars <- names(wisc2[7:11])
latents2 <- c("verbal","perf")
#manifests <- c("verbalVars","perfVars")
observedCov <- cov(wisc2[,1:11])
numSubjects <- nrow(wisc2)

5.2.2 Create the model using the mxModel function

Essentially initializes the model and sets vars/paths

cfa2b <- mxModel("Two Factor Model",type="RAM",
manifestVars = manifests, latentVars = latents2,
# Now set the residual variance for manifest variables
mxPath(from = manifests, arrows=2,free=T,values=1, labels=paste("e",1:11,sep="")),
# set latent factor variances
mxPath(from=latents2,arrows=2,free=F,values=1,labels=c("p1","p2")),
# speciffy factor loadings
# Allow the factors to covary as per the lavaan model
mxPath(from="verbal", to="perf", arrows=2,
free=T, values=1, labels="latcov1"),
# Specify the latent factor paths to manifests
mxPath(from="verbal", to = verbalVars,
arrows=1,
free=T,values=1),
mxPath(from="perf", to = perfVars,
arrows=1,
free=T,values=1),
mxData(observed=observedCov,type="cov",numObs=numSubjects)
) # close the model
5.2. SECOND OPENMX MODEL - THE BIFACTOR MODEL 45

5.2.3 Now use ’mxRun’to fit the model

mxRun uses the model defined above
cfa2b <- mxRun(cfa2b)

## Running Two Factor Model with 23 parameters

Summarize the fit

summary(cfa2b)

## Summary of Two Factor Model

##
## free parameters:
## name matrix row col Estimate Std.Error A
## 1 Two Factor Model.A[1,12] A info verbal 2.20616 0.201309
## 2 Two Factor Model.A[2,12] A comp verbal 2.04220 0.211937
## 3 Two Factor Model.A[3,12] A arith verbal 1.29984 0.172679
## 4 Two Factor Model.A[4,12] A simil verbal 2.23205 0.225198
## 5 Two Factor Model.A[5,12] A vocab verbal 2.25045 0.200598
## 6 Two Factor Model.A[6,12] A digit verbal 1.05277 0.212308
## 7 Two Factor Model.A[7,13] A pictcomp perf 1.74200 0.246173
## 8 Two Factor Model.A[8,13] A parang perf 1.25323 0.224773
## 9 Two Factor Model.A[9,13] A block perf 1.84593 0.224562
## 10 Two Factor Model.A[10,13] A object perf 1.60457 0.236048
## 11 Two Factor Model.A[11,13] A coding perf 0.20701 0.257118
## 12 e1 S info info 3.56570 0.516752
## 13 e2 S comp comp 4.57229 0.594874
## 14 e3 S arith arith 3.60184 0.422011
## 15 e4 S simil simil 5.09555 0.666203
## 16 e5 S vocab vocab 3.48717 0.507536
## 17 e6 S digit digit 6.16241 0.680961
## 18 e7 S pictcomp pictcomp 5.52590 0.772061
## 19 e8 S parang parang 5.46287 0.658943
## 20 e9 S block block 3.89376 0.651701
## 21 e10 S object object 5.46733 0.719732
## 22 e11 S coding coding 8.15907 0.874195
## 23 latcov1 S verbal perf 0.58883 0.077178
##
## Model Statistics:
## | Parameters | Degrees of Freedom | Fit (-2lnL units)
## Model: 23 43 5445.7
## Saturated: 66 0 5375.1
## Independence: 11 55 5894.3
## Number of observations/statistics: 175/66
##
## chi-square: <U+03C7>² ( df=43 ) = 70.608, p = 0.0050089
## Information Criteria:
## | df Penalty | Parameters Penalty | Sample-Size Adjusted
## AIC: -15.392 116.61 123.92
## BIC: -151.478 189.40 116.56
## CFI: 0.94053
## TLI: 0.92393 (also known as NNFI)
## RMSEA: 0.060571 [95% CI (0.026575, 0.089595)]
## Prob(RMSEA <= 0.05): 0.23348
46 CHAPTER 5. USING THE OPENMX PACKAGE FOR CFA

## timestamp: 2019-07-11 11:27:19

## Wall clock time: 0.0322 secs
## optimizer: CSOLNP
## OpenMx version number: 2.12.2
## Need help? See help(mxSummary)

What is in the model?

slotNames(cfa2b@output)

## NULL
names(cfa2b@output)

## [1] "matrices" "algebras"

## [3] "data" "SaturatedLikelihood"
## [5] "IndependenceLikelihood" "calculatedHessian"
## [7] "standardErrors" "gradient"
## [9] "hessian" "expectations"
## [11] "fit" "fitUnits"
## [13] "Minus2LogLikelihood" "maxRelativeOrdinalError"
## [15] "minimum" "estimate"
## [17] "infoDefinite" "conditionNumber"
## [19] "status" "iterations"
## [21] "evaluations" "mxVersion"
## [23] "frontendTime" "backendTime"
## [25] "independentTime" "wallTime"
## [27] "timestamp" "cpuTime"
5.2. SECOND OPENMX MODEL - THE BIFACTOR MODEL 47

5.2.4 Can we use semPlot to draw the OpenMx model fit?

The semPlot package is very powerful and can recognize many lm and sem model objects. We can use the
identical code that we used in chapter 2 for the lavaan model.
# Note that the base plot, including standardized path coefficients plots positive coefficients green
# and negative coefficients red. Red-green colorblindness issues anyone?
# I redrew it here to choose a blue and red. But all the coefficients in this example are
# positive,so they are shown with the skyblue.
# more challenging to use colors other than red and green. not in this doc
semPaths(cfa2b, residuals=F,sizeMan=7,"std",
posCol=c("skyblue4", "red"),
#edge.color="skyblue4",
edge.label.cex=1.2,layout="circle2")

cdn inf

obj cmp
0.07 0.76
0.57 0.69
blc art
0.68 0.57
prf 0.59 vrb

0.47 0.70

prn 0.60 0.77 sml

0.39

pct vcb
dgt

# or we could draw the paths in such a way to include the residuals:

#semPaths(fit1, sizeMan=7,"std",edge.color="skyblue4",edge.label.cex=1,layout="circle2")
# the base path diagram can be drawn much more simply:
#semPaths(fit1)
# or
semPaths(cfa2b,"std")
48 CHAPTER 5. USING THE OPENMX PACKAGE FOR CFA

1.00 1.00

0.59

vrb prf

0.76 0.69 0.57 0.70 0.77 0.39 0.60 0.47 0.68 0.57 0.07

inf cmp art sml vcb dgt pct prn blc obj cdn

0.42 0.52 0.68 0.51 0.41 0.85 0.65 0.78 0.53 0.68 0.99
5.3. THIRD OPENMX MODEL 49

5.3 Third OpenMx Model

Now fit a model that is the same as the second model fit with lavaan above. Two factors, verbal and
performance, plus paths from both latents to “comp”.

5.3.1 set new dataframe and set up basics

Same initial setups as above.
# grab data
wisc2 <- wisc1[2:12] # remove ID from df
# set up basics
manifests <- names(wisc2)
verbalVars <- names(wisc2[1:6])
perfVars <- c(names(wisc2[7:11]),"comp")
latents2 <- c("verbal","perf")
#manifests <- c("verbalVars","perfVars")
observedCov <- cov(wisc2[,1:11])
numSubjects <- nrow(wisc2)

5.3.2 Create the model using the mxModel function

Essentially initializes the model and sets vars/paths
cfa2c <- mxModel("TwoFac, Comp Common",type="RAM",
manifestVars = manifests, latentVars = latents2,
# Now set the residual variance for manifest variables
mxPath(from = manifests, arrows=2,free=T,values=1, labels=paste("e",1:11,sep="")),
# set latent factor variances
mxPath(from=latents2,arrows=2,free=F,values=1,labels=c("p1","p2")),
# speciffy factor loadings
# Allow the factors to covary as per the lavaan model
mxPath(from="verbal", to="perf", arrows=2,
free=T, values=1, labels="latcov1"),
# Specify the latent factor paths to manifests
mxPath(from="verbal", to = verbalVars,
arrows=1,
free=T,values=1),
mxPath(from="perf", to = perfVars,
arrows=1,
free=T,values=1),
mxData(observed=observedCov,type="cov",numObs=numSubjects)
) # close the model

5.3.3 Now use ‘mxRun’ to fit the model

mxRun uses the model defined above
cfa2c <- mxRun(cfa2c)

## Running TwoFac, Comp Common with 24 parameters

Summarize the fit
summary(cfa2c)

## Summary of TwoFac, Comp Common

50 CHAPTER 5. USING THE OPENMX PACKAGE FOR CFA

##
## free parameters:
## name matrix row col Estimate
## 1 TwoFac, Comp Common.A[1,12] A info verbal 2.25606
## 2 TwoFac, Comp Common.A[2,12] A comp verbal 1.49130
## 3 TwoFac, Comp Common.A[3,12] A arith verbal 1.30678
## 4 TwoFac, Comp Common.A[4,12] A simil verbal 2.20475
## 5 TwoFac, Comp Common.A[5,12] A vocab verbal 2.27348
## 6 TwoFac, Comp Common.A[6,12] A digit verbal 1.07476
## 7 TwoFac, Comp Common.A[2,13] A comp perf 0.88414
## 8 TwoFac, Comp Common.A[7,13] A pictcomp perf 1.78962
## 9 TwoFac, Comp Common.A[8,13] A parang perf 1.18881
## 10 TwoFac, Comp Common.A[9,13] A block perf 1.82276
## 11 TwoFac, Comp Common.A[10,13] A object perf 1.63283
## 12 TwoFac, Comp Common.A[11,13] A coding perf 0.20047
## 13 e1 S info info 3.34302
## 14 e2 S comp comp 4.33109
## 15 e3 S arith arith 3.58375
## 16 e4 S simil simil 5.21667
## 17 e5 S vocab vocab 3.38297
## 18 e6 S digit digit 6.11561
## 19 e7 S pictcomp pictcomp 5.35770
## 20 e8 S parang parang 5.62019
## 21 e9 S block block 3.97877
## 22 e10 S object object 5.37584
## 23 e11 S coding coding 8.16174
## 24 latcov1 S verbal perf 0.53322
## Std.Error A
## 1 0.200315
## 2 0.256354
## 3 0.173048
## 4 0.227320
## 5 0.200572
## 6 0.212373
## 7 0.270140
## 8 0.242338
## 9 0.225100
## 10 0.221685
## 11 0.233235
## 12 0.255009
## 13 0.509394
## 14 0.557177
## 15 0.422007
## 16 0.682585
## 17 0.507400
## 18 0.677590
## 19 0.755634
## 20 0.665637
## 21 0.636905
## 22 0.708170
## 23 0.874112
## 24 0.082033
##
## Model Statistics:
5.3. THIRD OPENMX MODEL 51

## | Parameters | Degrees of Freedom | Fit (-2lnL units)

## Model: 24 42 5435.7
## Saturated: 66 0 5375.1
## Independence: 11 55 5894.3
## Number of observations/statistics: 175/66
##
## chi-square: <U+03C7>² ( df=42 ) = 60.61, p = 0.031396
## Information Criteria:
## | df Penalty | Parameters Penalty | Sample-Size Adjusted
## AIC: -23.39 108.61 116.61
## BIC: -156.31 184.56 108.56
## CFI: 0.95991
## TLI: 0.9475 (also known as NNFI)
## RMSEA: 0.050319 [95% CI (0, 0.081389)]
## Prob(RMSEA <= 0.05): 0.4664
## timestamp: 2019-07-11 11:27:24
## Wall clock time: 0.0488 secs
## optimizer: CSOLNP
## OpenMx version number: 2.12.2
## Need help? See help(mxSummary)

What is in the model?

slotNames(cfa2c@output)

## NULL
names(cfa2c@output)

## [1] "matrices" "algebras"

## [3] "data" "SaturatedLikelihood"
## [5] "IndependenceLikelihood" "calculatedHessian"
## [7] "standardErrors" "gradient"
## [9] "hessian" "expectations"
## [11] "fit" "fitUnits"
## [13] "Minus2LogLikelihood" "maxRelativeOrdinalError"
## [15] "minimum" "estimate"
## [17] "infoDefinite" "conditionNumber"
## [19] "status" "iterations"
## [21] "evaluations" "mxVersion"
## [23] "frontendTime" "backendTime"
## [25] "independentTime" "wallTime"
## [27] "timestamp" "cpuTime"
52 CHAPTER 5. USING THE OPENMX PACKAGE FOR CFA

5.3.4 Draw the OpenMx model number 3

The semPlot package is very powerful and can recognize many ‘lm’ and SEM model objects. We can use
the identical code that we used in chapter 2 for the lavaan model.
# Note that the base plot, including standardized path coefficients plots positive coefficients green
# and negative coefficients red. Red-green colorblindness issues anyone?
# I redrew it here to choose a blue and red. But all the coefficients in this example are
# positive,so they are shown with the skyblue.
# more challenging to use colors other than red and green. not in this doc
semPaths(cfa2c, residuals=F,sizeMan=7,"std",
posCol=c("skyblue4", "red"),
#edge.color="skyblue4",
edge.label.cex=1.2,layout="circle2")

cdn inf

obj art
0.07 0.78
0.58 0.57
blc sml
0.67 0.69
prf 0.53 vrb

0.45 0.78

prn 0.61 0.40 vcb

0.30 0.50

pct dgt
cmp

# or we could draw the paths in such a way to include the residuals:

#semPaths(fit1, sizeMan=7,"std",edge.color="skyblue4",edge.label.cex=1,layout="circle2")
# the base path diagram can be drawn much more simply:
#semPaths(fit1)
# or
semPaths(cfa2c,"std")
5.4. COMPARE THE OPENMX MODELS 53

1.00 1.00

0.53

vrb prf

0.78 0.57 0.69 0.78 0.40 0.50 0.30 0.61 0.45 0.67 0.58 0.07

inf art sml vcb dgt cmp pct prn blc obj cdn

0.40 0.68 0.52 0.40 0.84 0.50 0.63 0.80 0.54 0.67 1.00

5.4 Compare the OpenMx models

In OpenMx a convenient function exists for model comparisons. This code compares the same two models
that we initially compared in the lavaan approach that used the anova function.
kable(mxCompare(cfa2c,cfa2b), booktabs=TRUE, format="markdown")

base comparison ep minus2LL df AIC diffLL diffdf p

TwoFac, Comp Common NA 24 5435.7 42 -23.390 NA NA NA
TwoFac, Comp Common Two Factor Model 23 5445.7 43 -15.392 9.998 1 0.00157
Chapter 6

Summary and Reproducibility

We have finished a nice book……..

Here is some information for for reproducibility:
sessionInfo()

## R version 3.5.2 (2018-12-20)

## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 7 x64 (build 7601) Service Pack 1
##
## Matrix products: default
##
## locale:
## [1] LC_COLLATE=English_United States.1252
## [2] LC_CTYPE=English_United States.1252
## [3] LC_MONETARY=English_United States.1252
## [4] LC_NUMERIC=C
## [5] LC_TIME=English_United States.1252
##
## attached base packages:
## [1] stats graphics grDevices utils datasets methods base
##
## other attached packages:
## [1] OpenMx_2.12.2 sem_3.1-9 lavaan_0.6-3 ggraph_1.0.2
## [5] ggplot2_3.1.0 corrplot_0.84 tidyr_0.8.2 magrittr_1.5
## [9] dplyr_0.8.0.1 MVN_5.6 kableExtra_1.0.1 knitr_1.21
## [13] psych_1.8.12 semPlot_1.1 car_3.0-2 carData_3.0-2
##
## loaded via a namespace (and not attached):
## [1] readxl_1.3.0 backports_1.1.3 Hmisc_4.2-0
## [4] BDgraph_2.55 plyr_1.8.4 igraph_1.2.4
## [7] lazyeval_0.2.2 sp_1.3-1 splines_3.5.2
## [10] digest_0.6.18 htmltools_0.3.6 viridis_0.5.1
## [13] matrixcalc_1.0-3 checkmate_1.9.1 lisrelToR_0.1.4
## [16] cluster_2.0.7-1 openxlsx_4.1.0 readr_1.3.1
## [19] jpeg_0.1-8 colorspace_1.4-1 ggrepel_0.8.0
## [22] rvest_0.3.2 rrcov_1.4-7 haven_2.1.0
## [25] xfun_0.4 crayon_1.3.4 lme4_1.1-20
## [28] zoo_1.8-4 survival_2.43-3 glue_1.3.1

54
 55

## [31] gtable_0.3.0 webshot_0.5.1 mi_1.0

## [34] kernlab_0.9-27 prabclus_2.2-7 DEoptimR_1.0-8
## [37] ggm_2.3 abind_1.4-5 VIM_4.8.0
## [40] scales_1.0.0 sgeostat_1.0-27 mvtnorm_1.0-8
## [43] GGally_1.4.0 sROC_0.1-2 Rcpp_1.0.1
## [46] viridisLite_0.3.0 xtable_1.8-3 laeken_0.5.0
## [49] htmlTable_1.13.1 units_0.6-2 foreign_0.8-71
## [52] mclust_5.4.2 Formula_1.2-3 stats4_3.5.2
## [55] truncnorm_1.0-8 vcd_1.4-4 htmlwidgets_1.3
## [58] httr_1.4.0 fpc_2.1-11.1 RColorBrewer_1.1-2
## [61] modeltools_0.2-22 acepack_1.4.1 farver_1.1.0
## [64] NADA_1.6-1 flexmix_2.3-15 pkgconfig_2.0.2
## [67] reshape_0.8.8 XML_3.98-1.17 nnet_7.3-12
## [70] kutils_1.60 tidyselect_0.2.5 rlang_0.3.1
## [73] reshape2_1.4.3 munsell_0.5.0 cellranger_1.1.0
## [76] tools_3.5.2 moments_0.14 ranger_0.11.1
## [79] pls_2.7-0 cvTools_0.3.2 fdrtool_1.2.15
## [82] evaluate_0.13 stringr_1.4.0 arm_1.10-1
## [85] yaml_2.2.0 zip_1.0.0 robustbase_0.93-3
## [88] purrr_0.3.2 glasso_1.10 pbapply_1.4-0
## [91] nlme_3.1-137 whisker_0.3-2 xml2_1.2.0
## [94] compiler_3.5.2 rstudioapi_0.9.0 curl_3.3
## [97] png_0.1-7 e1071_1.7-0.1 zCompositions_1.2.0
## [100] huge_1.2.7 tweenr_1.0.1 tibble_2.0.1
## [103] robCompositions_2.0.10 pbivnorm_0.6.0 pcaPP_1.9-73
## [106] stringi_1.3.1 highr_0.7 qgraph_1.6.1
## [109] rockchalk_1.8.140 forcats_0.4.0 trimcluster_0.1-2.1
## [112] lattice_0.20-38 Matrix_1.2-15 nloptr_1.2.1
## [115] pillar_1.3.1 lmtest_0.9-36 data.table_1.12.0
## [118] corpcor_1.6.9 R6_2.4.0 latticeExtra_0.6-28
## [121] bookdown_0.9 gridExtra_2.3 rio_0.5.16
## [124] boot_1.3-20 energy_1.7-5 MASS_7.3-51.1
## [127] gtools_3.8.1 assertthat_0.2.1 rjson_0.2.20
## [130] withr_2.1.2 nortest_1.0-4 mnormt_1.5-5
## [133] diptest_0.75-7 parallel_3.5.2 hms_0.4.2
## [136] grid_3.5.2 rpart_4.1-13 coda_0.19-2
## [139] class_7.3-15 minqa_1.2.4 rmarkdown_1.11
## [142] mvoutlier_2.0.9 d3Network_0.5.2.1 ggforce_0.1.3
## [145] numDeriv_2016.8-1 semTools_0.5-1 base64enc_0.1-3
Bibliography

Allaire, J., Xie, Y., McPherson, J., Luraschi, J., Ushey, K., Atkins, A., Wickham, H., Cheng, J., Chang, W.,
and Iannone, R. (2018). rmarkdown: Dynamic Documents for R. R package version 1.11.

Bache, S. M. and Wickham, H. (2014). magrittr: A Forward-Pipe Operator for R. R package version 1.5.

Boker, S. M., Neale, M. C., Maes, H. H., Spiegel, M., Brick, T. R., Estabrook, R., Bates, T. C., Gore, R. J.,
Hunter, M. D., Pritikin, J. N., Zahery, M., and Kirkpatrick, R. M. (2019). OpenMx: Extended Structural
Equation Modelling. R package version 2.12.1.

Brown, T. A. (2015). Confirmatory factor analysis for applied research. Methodology in the social sciences.
The Guilford Press, New York ; London, second edition. edition.

El-Sheikh, A. A., Abonazel, M. R., and Gamil, N. (2017). A review of software packages for structural
equation modeling: A comparative study. Applied Mathematics and Physics, 5(3):85–94.

Epskamp, S. and with contributions from Simon Stuber (2017). semPlot: Path Diagrams and Visual Analysis
of Various SEM Packages’ Output. R package version 1.1.

Fox, J., Nie, Z., and Byrnes, J. (2017). sem: Structural Equation Models. R package version 3.1-9.

Fox, J., Weisberg, S., and Price, B. (2018). car: Companion to Applied Regression. R package version 3.0-2.

Korkmaz, S., Goksuluk, D., and Zararsiz, G. (2018). MVN: Multivariate Normality Tests. R package version
5.5.

Narayanan, A. (2012). A review of eight software packages for structural equation modeling. The American
Statistician, 66(2):129–138.

R Core Team (2018). R: A Language and Environment for Statistical Computing. R Foundation for Statistical
Computing, Vienna, Austria.

Revelle, W. (2019). psych: Procedures for Psychological, Psychometric, and Personality Research. R package
version 1.8.12.

Rosseel, Y. (2012). Lavaan: An r package for structural equation modeling and more. version 0.5–12 (beta).
Journal of statistical software, 48(2):1–36.

Rosseel, Y. (2018). lavaan: Latent Variable Analysis. R package version 0.6-3.

RStudio Team (2015). RStudio: Integrated Development Environment for R. RStudio, Inc., Boston, MA.

Tabachnick, B. G., Fidell, L. S., and Ullman, J. B. (2019). Using multivariate statistics. Pearson, Boston,
seventh edition. edition.

Wei, T. and Simko, V. (2017). corrplot: Visualization of a Correlation Matrix. R package version 0.84.

Wickham, H., François, R., Henry, L., and Müller, K. (2018). dplyr: A Grammar of Data Manipulation. R
package version 0.7.8.

56
BIBLIOGRAPHY 57

Wickham, H. and Henry, L. (2018). tidyr: Easily Tidy Data with ’spread()’ and ’gather()’ Functions. R
package version 0.8.2.
Xie, Y. (2015). Dynamic Documents with R and knitr. Chapman and Hall/CRC, Boca Raton, Florida, 2nd
edition. ISBN 978-1498716963.

Xie, Y. (2018a). bookdown: Authoring Books and Technical Documents with R Markdown. R package version
0.9.
Xie, Y. (2018b). knitr: A General-Purpose Package for Dynamic Report Generation in R. R package version
1.21.
Zhu, H. (2019). kableExtra: Construct Complex Table with ’kable’ and Pipe Syntax. R package version 1.0.1.