lecture8_s10 CFA

# lecture8_s10 CFA - Confirmatory Factor Analysis...

This preview shows pages 1–6. Sign up to view the full content.

Confirmatory Factor Analysis: Identification and estimation Psychology 588: Covariance structure and factor models Feb 19, 2010

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
Identification 2 • Covariance structure of measurement model: () xx δ =+ Σθ ΛΦΛ Θ where we can impose selective zero and equality constraints on Λ x and Φ ; and free selective off-diagonal elements in Θ δ , provided that the resulting model is identifiable • Though CFA has different parameter sets than the path model only with observed variables, the basic ideas hold: ¾ A model is identified if and only if every single free parameter has a unique solution ¾ A parameter is identified if it can be written as a function (or functions) of the data (variances/covariances)
Single-factor model 3 • As a simplistic case, consider only one factor measured by 2 or 3 indicators, with no correlated errors (see detailed algebraic derivation in pp. 240-242) ¾ 2 indicators --- one more constraint (in addition to the scaling constraint) needed for just identification; e.g., known reliability, tau-equivalent measures ¾ 3 indicators --- just identifiable with the scaling constraint ¾ Multiple orthogonal factors with a uni-factorial loading pattern can also be considered as separate single-factor models for identification, except that any non-zero covariance between indicators of different factors will also contribute to misfit

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
t-rule 4 • With no knowledge whatsoever about parameters to constrain, the factor model (for both EFA and CFA) has free parameters as many as: ¾ q × n factor loadings in Λ x ¾ n ( n +1) / 2 nonredundant elements in Φ ¾ q ( q + 1) / 2 nonredundant elements in Θ δ • Thus, constraints (including the scaling constraints) needed so as to satisfy ( ) 12 tq q + • As before, t-rule is necessary, not sufficient
Three indicators per factor 5

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

## This note was uploaded on 06/11/2010 for the course PSYC 588 taught by Professor Sunjinghong during the Spring '10 term at University of Illinois at Urbana–Champaign.

### Page1 / 14

lecture8_s10 CFA - Confirmatory Factor Analysis...

This preview shows document pages 1 - 6. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online