Lecture 5 - CFA and SEM Flashcards
What is the definition of Structural Equation Modelling (SEM)?
A broad and powerful method for fitting networks of constructs to data.
What types of analyses can SEM perform?
Logistic regression, moderation and mediation, ANOVA, multilevel modelling, and CFA.
What are the components of SEM?
Measurement model and Structural model.
What do observed variables represent in SEM figures?
The data, represented by boxes.
What do latent variables represent in SEM figures?
Hypothesised constructs, represented by circles.
What is the purpose of Confirmatory Factor Analysis (CFA)?
To identify latent psychological constructs and test hypotheses about factor structure.
What is the difference between Exploratory Factor Analysis (EFA) and CFA?
EFA is data-driven and identifies the structure of a dataset, while CFA is theory-driven and tests specific hypotheses about factor structure.
What are the steps in the CFA process?
Preliminaries, Evaluate Model Fit, Evaluate Parameter Estimates, Evaluate Alternative Models.
What does the χ2 test evaluate in CFA?
The discrepancy between predicted and empirical variance-covariance matrices.
What are some alternative indices used to evaluate model fit?
NFI, NNFI, IFI, CFI, GFI, AGFI.
What are the residual fit indices used to evaluate model fit?
SRMR and RMSEA.
When should you use EFA?
When exploring variable structures, especially with new variables.
When should you use CFA?
When testing a priori hypotheses based on theory or previous research.
Why is it problematic to conduct CFA on the same sample as EFA?
Because it involves generating and testing hypotheses on the same sample, which can lead to biased results.
What is the importance of evaluating alternative models in CFA?
To ensure that the hypothesized model is not only fitting well but is also the best model among alternatives.
What is the implication of a good model fit in CFA?
It indicates that the model is consistent with the data’s correlational structure, but it does not confirm the model’s correctness.
What is the implication of a poor model fit in CFA?
The model does not adequately account for the data, indicating the need for adjustments.
What does the measurement model in SEM represent?
The relations between latent variables and their indicators (measured variables).
What does the structural model in SEM represent?
The hypothesised causal relations between latent variables.
What is the main takeaway of CFA compared to EFA?
CFA is more restrictive but allows for stronger conclusions due to its theory-driven approach.
