Structural Equation Modeling

Lecture Notes
 Structural equation modeling (SEM)

Structural equation modeling (SEM)

is a comprehensive statistical modeling tool for
analyzing multivariate data involving complex
relationships between and among variables (Hoyle,
1995).
 Structural equation modeling (SEM)

SEM surpasses traditional regression models by

including multiple independent and dependent
variables to test associated hypothesizes about
relationships among observed and latent variables.

SEM explain why results occur while reducing

misleading results by submitting all variables in the
model to measurement error or uncontrolled
variation of the measured variables.
 Features of Structural equation modeling
(SEM)
According to Byrne (2010), Structural Equation
Modeling (SEM) is a powerful collection of
multivariate analysis techniques, which specifies the
relationships between variables through the use of
two main sets of equations:

1. Measurement equations
2. Structural Equations.
 Features of Structural equation modeling
(SEM)
Measurement equations - test the accuracy of
proposed measurements by assessing relationships
between latent variables and their respective
indicators

The Structural Equations - drive the assessment of the

hypothesized relationships between the latent
variables, which allow testing the statistical
hypotheses for the study.
 Features of Structural equation modeling
(SEM)

Remark: SEM considers the modeling of

interactions, nonlinearities, correlated
independents, measurement error, correlated
error terms, and multiple latent independents
each measured by multiple indicators.
Types of Variables Hypothesized in the SEM
Model

 Exogenous Variables - Variables that are not

influenced by other variables in the model.

 Endogenous Variables - Variable that is caused

by other variables in the model

 Indicator Variables - Variables that are directly

observed and measured (also known as
manifest variables in some circles).
 Types of Variables Hypothesized in the SEM
Model

 Latent Variables - Variables that are not directly

measured

 Measurement Model - This is a part of the entire

structural equation model diagram hypothesized for
the study including all observations that load onto
the latent variable, their relationships, variances,
and errors.
Types of Variables Hypothesized in the SEM
Model

 Structural Model - This is a part of the total

hypothesized structural equation model
diagram, which includes both latent and
indicator variables

 Structural Equation Model - This model

combines the structural model and the
measurement model, which includes everything
that has been measured and observed among
the variables examined.
 Difference Between SEM and Other
Conventional Methods of Statistical Analysis

The basic statistic in SEM is the covariance.

-covariance statistics convey more information
than a correlation (Hu & Bentler, 1999)

While conventional regression analysis

attempt to minimize differences between
observed and expected individual cases, SEM
aims to minimize differences between
observed and expected covariance matrices.
 Difference Between SEM and Other
Conventional Methods of Statistical Analysis

The basic statistic in SEM is the covariance.

-covariance statistics convey more information
than a correlation (Hu & Bentler, 1999)

While conventional regression analysis

attempt to minimize differences between
observed and expected individual cases, SEM
aims to minimize differences between
observed and expected covariance matrices.
 Difference Between SEM and Other
Conventional Methods of Statistical Analysis

Thatis, SEM, based on the covariance

statistic, attempts “to understand patterns
of correlations among a set of variables and
to explain as much of their variances”
(Kline, 1998, pp. 10-11).
 Difference Between SEM and Other
Conventional Methods of Statistical Analysis

Unlike conventional analysis, SEM allows the

inclusion of latent variables into the
analyses and it is not limited to relationships
among observed variables and constructs.
It allows the study to measure any
combination of relationships by examining a
series of dependent relationships
simultaneously while considering potential
errors of measurement among all variables.
 SEM has Several Advantages over
Conventional Analysis

1. Greater flexibility regarding assumptions

(particularly allowing interpretation even in
the face of multicollinearity).

2. Allows the use of confirmatory factor analysis

to reduce measurement error by testing multiple
indicators per latent variable while offering
superior model visualization through its graphical
modeling interface ( Hatcher, 2005; Joreskog,
1993; Kline, 2005).
 SEM has Several Advantages over
Conventional Analysis

3. Has the appealing capacity of testing

models overall rather than coefficients
individually.

4. Has the ability to test models with multiple

dependent variables, to include mediating
variables and to model error terms for all
indicator variables.
 SEM has Several Advantages over
Conventional Analysis

5.It considers potential errors of measurement

in all variables and when a hypothesized
structural model shows model fit indices that
are less than satisfactory, it allows
specification searches to find better fitting
models to the sample variance-covariance
matrix (Hu & Bentler, 1999; Kline, 2005;
Schumacker & Lomax, 2004).
 Structural Equation Modeling Process

1. First, it validates the measurement model in terms of

assessing the relationship between hypothetic latent
constructs and clusters of observed variables underlying
each construct.
-Validation of the measurement model is often
conducted by using Confirmatory Factor Analysis (CFA).

2. The second step centers around fitting the structural

model by measuring the significance of the relationship
between latent variables, which is often accomplished
through path analysis ( Hoyle, 1995; Kaplan, 2000).
 Hypothesized Diagram Model

Hypothesized diagram model in SEM is used

"to find a model that not only fits the data
well from a statistical point of view, but also
has the property that every parameter of the
model can be given a substantively meaningful
interpretation" (Joreskog, 1993, p. 295).
 Hypothesized Diagram Model

Hypothesized diagram model in SEM is

used "to find a model that not only fits the
data well from a statistical point of view, but
also has the property that every parameter of
the model can be given a substantively
meaningful interpretation" (Joreskog, 1993, p.
295).
 Hypothesized Diagram Model
Developing a Structural Equation Model

1. Model Specification - This is the process of formally stating a model by

determining which parameters are to be fixed or free.

2. Model Identification - This is the idea of having at least one unique

solution for each parameter estimate in the model from the observed data.

3. Model Estimation - This is the process in which start values of the free
parameters are chosen in order to generate an estimated population
covariance matrix ∑ (θ), from the model (Loehlin, 2004).

4. Testing Model Fit - This is the process of evaluating a structural equation

model with goodness-of-fit indices.
 Hypothesized Diagram Model

Developing a Structural Equation Model

1. Model Manipulation -This is the process of

making model adjustments through
specification searches. The process of drawing
hypothesized diagram model is can be done in
jasp open source software. (i.e., AMOS) and
requires compliance with some of the following
basic rules:
 Hypothesized Diagram Model

Developing a Structural Equation Model

1. Model Manipulation -This is the process of

making model adjustments through
specification searches. The process of drawing
hypothesized diagram model is can be done in
jasp open source software and with compliance
of some of the following basic rules:
 Hypothesized Diagram Model
Developing a Structural Equation Model

1. Model Manipulation –

 Latent variables are depicted with circles

 Indicator variables are represented with squares.
 Lines with arrows in one direction represent a hypothesized direct
relationship between the two variables.
 A curved line with arrows in both directions shows a covariance between
two variables.
 Only exogenous variables have covariance arrows.
 Endogenous variables should have a residual term. A residual term is
depicted by a circle with the letter E written in it, which stands for error
 Hypothesized Diagram Model

Developing a Structural Equation Model

1. Model Manipulation –
 The error term in the endogenous latent variable is called a disturbance and it is
depicted by a circle with a D written in it (Byrne, 2010).
 Parameters are the variances, regression coefficients and covariances among variables
 Variances are indicated by a two-headed arrow with both ends of the same arrow
pointing at the same variable.
 Regression coefficients are depicted along single-headed arrows indicating a
hypothesized pathway between two variables.
 Covariances are represented by double-headed, curved arrows between two variables
or error terms
Developing a Structural Equation Model
 Developing a Structural Equation Model

Measurement Model
-Confirmatory factor analysis (CFA) allows
assessment of the research questions by
determining whether the observed variables are
indeed good indicators of the latent variables.
Remark: A separate confirmatory factor models
should run for each set of observed variables
hypothesized to indicate their respective latent
variable.
 Developing a Structural Equation Model

Measurement Model
 Developing a Structural Equation Model

Structural Model
While the measurement model embodies
the relationships between measured variables
and latent variables, the structural model
represents the relationships between latent
variables only and it must be inferred from
measured variables.
Developing a Structural Equation Model
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
Political Democracy Modeling Using SEM
 References

 De Carvalho, J., & Chima, F. O. (2014). Applications of

structural equation modeling in social sciences
research. American International Journal of Contemporary
Research, 4(1), 6-11.
 Bollen (1979). Political Democracy and the Timing of
Development. American Sociological Review, 44: 572:587.
 https://www.youtube.com/watch?v=fno5LC5cKdM
 JASP Team (2024). JASP (Version 0.18. 3)[Computer
software].