- Location: Hobbs Hall • 1818 Horton Ave • Nashville, TN 37212
- Room: 105
- Contact: Kris Preacher
- Email: firstname.lastname@example.org
- Audience: Free and Open to the Public
The use of the likelihood based method in quantifying uncertainty in structural equation models
Hao Wu, Department of Psychology, Boston College
The likelihood ratio test (LRT) has been a widely used method in psychometrics. When used as a difference test, it has the advantage of not being sensitive to the parametrization of the two models. A confidence set of parameters constructed by inversion of an LRT enjoys the same property. Such a method can be similarly applied to other problems in uncertainty quantification in SEMs. This talk will cover the use of the same idea in the calculation of fungible parameter sets of parameters, the calculation of bootstrap confidence intervals of model fit indices, and the calculation of CIs of parameters with a boundary.