|
Activity Number:
|
481
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Social Statistics Section
|
| Abstract - #305914 |
|
Title:
|
On the Use of Heywood Cases for Specification Testing in SEM
|
|
Author(s):
|
Stanislav Kolenikov*+ and Kenneth A. Bollen
|
|
Companies:
|
University of Missouri-Columbia and The University of North Carolina at Chapel Hill
|
|
Address:
|
146 Middlebush Hall, Columbia, MO, 65211,
|
|
Keywords:
|
latent variables ; structural equation models ; misspecification ; sandwich estimator ; estimation on the boundary
|
|
Abstract:
|
The situations when an error or a latent variable variance is estimated to be negative are referred to as Heywood cases in structural equation modeling. We shall discuss the reasons why Heywood cases emerge, and provide asymptotic theory framework for testing whether a Heywood case is indicative of gross model misspecification, as well as simulation results and empirical examples.
|
