Confirmatory Factor Analysis Flashcards
(19 cards)
What is Confirmatory Factor Analysis (CFA)?
A statistical technique used to test whether data fit a hypothesised measurement model, based on prior theory and involving both observed and latent variables.
How does CFA differ from Exploratory Factor Analysis (EFA)?
EFA explores data without pre-defined structure; CFA tests if a pre-specified model (based on theory) fits the data.
When is it appropriate to use CFA?
- When testing if a specific model fits your data.
- When the model is based on existing theory/research.
- When both observed and latent variables are included.
What is a latent variable?
A theoretical construct not directly measured, inferred from observed variables (e.g., wellbeing, intelligence).
What is an observed variable?
A variable that can be directly measured, like survey responses, age, or test scores.
What are the main steps of CFA?
“1. Model Specification
2. Model Identification
3. Model Estimation
4. Model Evaluation”
What is model specification in CFA?
Drawing the model (usually in AMOS), defining which observed variables load onto which latent variables.
What is model identification in CFA?
Checking if the model has enough knowns to estimate the unknowns (i.e., positive degrees of freedom).
What symbol represents latent variables in CFA diagrams?
Circles (also used for error terms).
How do you check model identification?
Look at the degrees of freedom under “Chi Square” in AMOS → Notes for Model. Positive df means identified.
What is model estimation in CFA?
Estimating parameters (e.g., factor loadings) using methods like Maximum Likelihood Estimation.
What does *** mean in AMOS output?
*** = p < .001, indicating very strong statistical significance.
What should be reported in model estimation (APA-style)?
”- Range of standardised beta values
- Whether they are significant
- Screenshot of AMOS model with betas”
What is model evaluation in CFA?
Assessing how well the theoretical model fits the observed data, usually using chi-square tests.
What does a significant chi-square test in CFA mean?
It indicates poor model fit — the data do not match the proposed theoretical model well.
Why is the chi-square test often significant in CFA even if the model is good?
Large sample sizes make chi-square tests sensitive to small deviations, often resulting in significance.
What is Maximum Likelihood Estimation (MLE)?
A method to find the parameter values that make the observed data most probable under the specified model.
What software is used for CFA in this module?
AMOS, which must be used with SPSS on a PC (not Mac compatible).
