Lecture 5: Confirmatory Factor Analysis

Question 1

What does CFA stand for?

Answer 1

Confirmatory Factor Analysis

Question 2

What broader statistical framework is CFA a part of?

Answer 2

Structural Equation Modelling (SEM)

Question 3

What does SEM model?

Answer 3

Networks of observed and latent variables

Question 4

What shape represents latent variables in SEM diagrams?

Answer 4

Circles or ovals

Question 5

What shape represents observed variables in SEM diagrams?

Answer 5

Boxes

Question 6

What do single-headed arrows indicate in SEM diagrams?

Answer 6

Causal paths

Question 7

What do double-headed arrows indicate in SEM diagrams?

Answer 7

Correlations

Question 8

What part of SEM specifies how indicators reflect latent variables?

Answer 8

Measurement model

Question 9

What part of SEM specifies relationships between latent variables?

Answer 9

Structural model

Question 10

Is CFA data-driven or theory-driven?

Answer 10

Theory-driven

Question 11

What does EFA freely estimate?

Answer 11

All factor loadings

Question 12

What does CFA often constrain to zero?

Answer 12

Cross-loadings or weak factor loadings

Question 13

What advantage does CFA have over EFA?

Answer 13

Greater parsimony and theoretical precision

Question 14

What does specifying a CFA model involve?

Answer 14

Defining factors, indicators, and correlations

Question 15

What is a free parameter in CFA?

Answer 15

A value estimated from data

Question 16

What is a fixed parameter in CFA?

Answer 16

A value constrained based on theory

Question 17

What test compares predicted and observed covariance matrices?

Answer 17

Chi-square test

Question 18

What is the preferred chi-square/df ratio?

Answer 18

Less than 2

Question 19

Name one absolute or relative fit index used in CFA.

Answer 19

Comparative Fit Index (CFI)

Question 20

What is a good CFI value?

Answer 20

Greater than .95

Question 21

What is the RMSEA cutoff for good model fit?

Answer 21

Less than .06

Question 22

What is the SRMR cutoff for good model fit?

Answer 22

Less than .08

Question 23

What does parsimony in CFA mean?

Answer 23

Explaining data simply without unnecessary complexity

Question 24

What is used to compare nested models?

Answer 24

Chi-square difference test

Question 25

What metric is used to compare non-nested models?

Answer 25

Akaike’s Information Criterion (AIC)

Question 26

What kind of AIC value indicates a better model?

Answer 26

Lower AIC

Question 27

What might poor model fit indicate?

Answer 27

Incorrect theoretical constraints

Question 28

What is HARKing?

Answer 28

Hypothesising after results are known

Question 29

What is the recommended minimum participants per parameter in CFA?

Answer 29

5 to 10

Question 30

What does CFA estimate?

Answer 30

Factor loadings and inter-factor correlations

Question 31

What type of testing does CFA allow?

Answer 31

Hypothesis testing about factor structure

Question 32

What does SEM allow that CFA does not?

Answer 32

Modelling of structural relationships

Question 33

Name one application of full SEM.

Answer 33

Longitudinal mediation modelling

Question 34

Name one major advantage of SEM.

Answer 34

Reduces measurement error

Question 35

Name one major limitation of SEM.

Answer 35

Requires substantial expertise

Question 36

What does CFA offer in psychological research?

Answer 36

A rigorous, theory-based framework for latent constructs