Related Experiment Videos
[Stagewise estimation for regression analysis when independent variables are latent variables]
1Department of Cognitive and Behavioral Science, Graduate School of Arts and Sciences, University of Tokyo, Komaba, Meguro-ku, Tokyo 153-8902.
Shinrigaku Kenkyu : the Japanese Journal of Psychology
|October 31, 2003
Summary
This study introduces a novel stagewise estimation method for psychological research involving latent variables. The method improves estimation efficiency and robustness compared to conventional approaches when analyzing observable dependent variables with independent latent variables.
Area of Science:
- Psychological research
- Econometrics
- Statistical modeling
Context:
- Psychological constructs are often unobservable, necessitating the use of latent variables (factors) as independent variables in regression analyses.
- Standard factor analysis methods can yield inaccurate factors and regression parameters when applied to combined sets of dependent and independent variables.
Purpose:
- To address the limitations of conventional factor analysis in psychological research when latent variables are used as predictors.
- To propose a stagewise estimation method for improved accuracy and robustness in structural equation modeling with latent variables.
Summary:
- A novel stagewise estimation method is proposed, involving separate estimation of measurement and structural equations.
- This method allows for the calculation of standard errors using the Bootstrapping technique.
- Numerical studies demonstrate superior estimation efficiency and robustness against model misspecification compared to traditional methods.
Impact:
- Provides a more accurate and reliable approach for analyzing psychological data with latent variables.
- Enhances the validity of regression parameters derived from models with unobservable constructs.
- Offers robust estimation solutions that are less sensitive to potential errors in model specification.