MMSR (MAN-MMA032A)
Difference between EFA, CFA, SEM
CFA (Confirmatory Factor Analysis), SEM (Structural Equation Modeling), and EFA (Exploratory Factor
Analysis) are interrelated techniques in multivariate statistics, each serving a distinct role within the
analysis of latent variables. Here's how they relate to one another and how they differ:
1. Purpose and Sequence of Use
EFA (Exploratory Factor Analysis):
o Purpose: Explore the underlying factor structure of a dataset without prior
assumptions about the number of factors or which variables load onto them.
o When to Use: Typically used early in research to identify potential latent constructs
and determine their structure.
o Exploratory in Nature: It is data-driven, and decisions about the number of factors or
their relationships are made based on statistical criteria (e.g., eigenvalues, scree
plot).
CFA (Confirmatory Factor Analysis):
o Purpose: Test a hypothesized factor structure, typically based on prior research or
theory.
o When to Use: After EFA has identified a factor structure or when testing a pre-
specified measurement model.
o Confirmatory in Nature: It is theory-driven, where the researcher specifies which
observed variables load onto which latent constructs, and tests the model fit.
SEM (Structural Equation Modeling):
o Purpose: Combine both the measurement model (validated using CFA) and the
structural model (relationships between latent variables).
o When to Use: After the measurement model (via CFA) has been validated, to
examine causal pathways and more complex relationships between latent variables.
o Integration of EFA and CFA: SEM can incorporate EFA-derived factor structures (if no
prior theory exists) but is often used after a CFA confirms the factor structure.
2. Relationship
EFA → CFA → SEM
1. EFA informs CFA:
EFA is often the first step when no clear theory exists about the factor
structure. It helps identify the number of factors and the items associated
with each factor.
2. CFA refines EFA results:
CFA tests the factor structure suggested by EFA or based on theoretical
reasoning. It provides statistical evidence about how well the observed
variables measure the latent constructs.
3. CFA is a component of SEM:
SEM uses the measurement model confirmed by CFA as a foundation and
adds structural pathways to examine relationships between constructs.
Difference between EFA, CFA, SEM
CFA (Confirmatory Factor Analysis), SEM (Structural Equation Modeling), and EFA (Exploratory Factor
Analysis) are interrelated techniques in multivariate statistics, each serving a distinct role within the
analysis of latent variables. Here's how they relate to one another and how they differ:
1. Purpose and Sequence of Use
EFA (Exploratory Factor Analysis):
o Purpose: Explore the underlying factor structure of a dataset without prior
assumptions about the number of factors or which variables load onto them.
o When to Use: Typically used early in research to identify potential latent constructs
and determine their structure.
o Exploratory in Nature: It is data-driven, and decisions about the number of factors or
their relationships are made based on statistical criteria (e.g., eigenvalues, scree
plot).
CFA (Confirmatory Factor Analysis):
o Purpose: Test a hypothesized factor structure, typically based on prior research or
theory.
o When to Use: After EFA has identified a factor structure or when testing a pre-
specified measurement model.
o Confirmatory in Nature: It is theory-driven, where the researcher specifies which
observed variables load onto which latent constructs, and tests the model fit.
SEM (Structural Equation Modeling):
o Purpose: Combine both the measurement model (validated using CFA) and the
structural model (relationships between latent variables).
o When to Use: After the measurement model (via CFA) has been validated, to
examine causal pathways and more complex relationships between latent variables.
o Integration of EFA and CFA: SEM can incorporate EFA-derived factor structures (if no
prior theory exists) but is often used after a CFA confirms the factor structure.
2. Relationship
EFA → CFA → SEM
1. EFA informs CFA:
EFA is often the first step when no clear theory exists about the factor
structure. It helps identify the number of factors and the items associated
with each factor.
2. CFA refines EFA results:
CFA tests the factor structure suggested by EFA or based on theoretical
reasoning. It provides statistical evidence about how well the observed
variables measure the latent constructs.
3. CFA is a component of SEM:
SEM uses the measurement model confirmed by CFA as a foundation and
adds structural pathways to examine relationships between constructs.
