Related Experiment Videos
A two-stage estimation of structural equation models with continuous and polytomous variables
S Y Lee1, W Y Poon, P M Bentler
1Department of Statistics, Chinese University of Hong Kong, Shatin, NT, Hong Kong.
The British Journal of Mathematical and Statistical Psychology
|November 1, 1995
Summary
This study introduces an efficient computational method for analyzing structural equation models with mixed variable types. The new procedure enhances accuracy and robustness for complex statistical modeling.
Area of Science:
- Statistics
- Quantitative Psychology
- Econometrics
Background:
- Structural Equation Models (SEM) are widely used but analyzing models with mixed continuous and polytomous variables presents computational challenges.
- Existing methods may lack efficiency or robustness when dealing with heterogeneous data types within SEM.
Purpose of the Study:
- To develop a computationally efficient procedure for the analysis of structural equation models (SEM) involving both continuous and polytomous variables.
- To provide a robust estimation method for parameters in mixed-variable SEM.
Main Methods:
- A two-stage approach combining partition maximum likelihood (PML) for initial estimates and generalized least squares (GLS) for structural parameter estimation.
- Utilizes estimates of thresholds, polyserial, and polychoric correlations within an asymptotic distribution framework.
Main Results:
- The proposed procedure demonstrates computational efficiency for mixed-variable SEM.
- Asymptotic properties of the derived estimators are established, providing theoretical guarantees.
- Simulation studies indicate favorable empirical performance and robustness compared to existing methods.
Conclusions:
- The developed method offers an efficient and robust solution for analyzing structural equation models with continuous and polytomous variables.
- This approach advances the practical application of SEM in fields dealing with mixed data types.