Evaluation of Variance Inflation Factors in Regression Models Using Latent Variable Modeling Methods.

Area of Science:

• Statistics
• Econometrics
• Psychometrics

Background:

• Multicollinearity poses challenges in regression analysis, potentially inflating variance inflation factors (VIFs) and reducing tolerance indices.
• Accurate assessment of VIFs and tolerance is crucial for reliable interpretation of regression models.
• Existing methods may lack the ability for both point and interval estimation of these multicollinearity diagnostics.

Purpose of the Study:

• To present a novel procedure for evaluating variance inflation factors and tolerance indices in linear regression.
• To enable both point and interval estimation for these multicollinearity diagnostics.
• To facilitate more informed decisions regarding multicollinearity in empirical research.

Main Methods:

• The procedure utilizes latent variable modeling software for estimation.
• It allows for the estimation of VIFs and tolerance indices for potential explanatory variables.
• The method is demonstrated using the Mplus software and a simulation study.

Main Results:

• The procedure successfully provides point and interval estimates for VIFs and tolerance indices.
• The simulation study validates the capabilities and performance of the proposed method.
• Empirical illustration confirms the practical utility of the approach.

Conclusions:

• The developed procedure offers a robust approach to assessing multicollinearity in regression models.
• Utilizing latent variable modeling software enhances the evaluation of VIFs and tolerance.
• This method aids researchers in making more informed decisions when multicollinearity is present.