TTU HomeTechAnnounce

TechAnnounce

Printer friendly format
Nonrecursive SEM Models and Parameter Estimation Issues
  Larry R. Price, Director and Professor of the Methodology, Measurement and Statistical Analysis (MMSA) at Texas State University will  discuss the increasing use of SEM models with feedback loops by a number of disciplines, several estimators available for these models, and present simulation results concerning parameter bias. This follows recent work Dr. Price reported in the journal Structural Equation Modeling: Performance of Nonrecursive Latent Variable Models Under Misspecification. Most SEM courses do not cover Nonrecursive (feedback loop) models, and there is very little literature on bias resulting from various aspects of model misspecification or choice of estimator. This work examined Maximum likelihood, Two-stage Least Squares, and Bayesian estimators under different degrees of misspecification, type of nonrecursive model and sample size. Recommendation for best practice are discusses when nonrecursive latent variable models are employed.
Posted:
3/15/2019

Originator:
Agha Hossein Sabzevari

Email:
roham.agha@ttu.edu

Department:
N/A

Event Information
Time: 4:00 PM - 5:30 PM
Event Date: 4/4/2019

Location:
NWI (Room number: 212)


Categories