|
Activity Number:
|
510
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #309921 |
|
Title:
|
Confirmatory Factor Analysis and Measurement Error: A Monte Carlo Simulation
|
|
Author(s):
|
Bruce Brown*+ and Tyson T. Phalp and Petr Blahus
|
|
Companies:
|
Brigham Young University and Myriad Genetics and Charles University
|
|
Address:
|
1098 SWKT, Provo, UT, 84602,
|
|
Keywords:
|
confirmatory factor analysis ; structural validity ; monte carlo simulation
|
|
Abstract:
|
It has been demonstrated by Lauritzen et al (2007) that exploratory factor analysis cannot produce structurally valid results unless a minimum reliability of r = .50 is met, and very few factor analyses in published studies meet these reliability requirements. The present study examines the operating characteristics of confirmatory factor analyses using the same Monte Carlo simulation methods. A variety of "manifest data" configurations are created from known underlying latent variable structures. Comparisons are drawn between the structural validity results for confirmatory factor analyses and those for exploratory factor analyses. It is demonstrated that when one knows how many factors are present in the latent variable structure, confirmatory factor analysis is somewhat more effective than exploratory factor analysis in producing structurally valid results.
|
