|
Activity Number:
|
390
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Social Statistics Section
|
| Abstract - #308825 |
|
Title:
|
Comparing Factor Loadings in Exploratory Factor Analysis: Comparison of Asymptotic and Randomization Tests
|
|
Author(s):
|
Holmes Finch*+ and Brian French
|
|
Companies:
|
Ball State University and Purdue University
|
|
Address:
|
Department of Educational Psychology, Muncie, IN, 47306,
|
|
Keywords:
|
multigroup factor analysis ; invariance testing ; exploratory factor analysis ; randomization test
|
|
Abstract:
|
Multigroup Confirmatory Factor Analysis (MCFA) is useful for testing factorial invariance. MCFA requires model identification often by setting a factor loading equal across groups. A non-invariant referent loading can result in inaccurate tests and parameter estimation. A search procedure has been suggested for identifying an invariant referent loading, but is quite difficult to use. Exploratory Factor Analysis (EFA) also has been recommended for identifying invariant factor loadings through visually comparing loadings. However, this approach does not account for sampling variation. This study introduces two new tests for comparing loadings across groups: (1) an asymptotic test statistic using Maximum Likelihood EFA SE's, and (2) a randomization test statistic. The paper presents both methods, compares their performance in a simulation study, and provides an applied example.
|
