SEM

Question 1

Structural Equation Modeling

Answer 1

○ Regression with latent (unobservable) variable that we can’t directly measure
○ Allows for multiple predictors and outcomes/dependent variables
○ Measurement model: identifies # of latent constructs via confirmatory factor analysis
○ Structural model: identifies causal relationships via regression paths
SEM fits the hypothesized model to the observed data: fit indices evaluate how likely it is that a given model gave rise to the observed data

Question 2

Types of Constructs

Answer 2

1. Reflective: construct is cause of the measure, is reflected by indicators (observable measures)
2. Formative: measures cause the construct (Ex: SES)

Question 3

Reflective Construct

Answer 3

□ Indicators are called “effect indicators” because reflective construct influences indicators
-Must be correlated
□ Items should show strong internal consistency reliability (correlated, because all reflection of underlying construct)
□ Factor loading: represented by lambda (y) in diagrams
-FL = relative weight that the item is given in the estimation of the latent factor
-Need not include all facets of a construct, as long as its unidimensional; can be interchanged which allows for parallel-forms
□ If a person’s level on the construct changes, their score on each measure should change accordingly

Question 4

Identified Model

Answer 4

• # datapoints = # parameters
• df = 0
• Remember, df = #datapoints - #parameters

Question 5

Under-identified Model

Answer 5

• # datapoints < # parameters
• df<0
• Remember, df = #datapoints - #parameters
• Requires you to add assumptions & variances

Question 6

Over-identified model

Answer 6

• # datapoints > # parameters
• df > 0
• Remember, df = #datapoints - #parameters

Question 7

Latent Class Model

Answer 7

• SEM for categorical models

- Ex: diagnosis

Question 8

Mixture Model

Answer 8

-SEM for combo of categorical constructs and continuous constructs

Question 9

Benefits of SEM

Answer 9

○ Allows for correlated errors (unlike CTT)
○ Handles multiple dependent variables simultaneously (unlike multiple regression)
○ Uses all available info/data using Full Info Maximum Likelihood (FIML) if pts are missing data (instead of deleting entire pt)
○ Can be used to account for different forms of measurement error (method bias)