Complex Regression Estimation Using Highly Correlated Auxiliary Variables

Key Points

• This research aims to address issues of instability and overestimation of variance in regression analysis using correlated auxiliary variables.
• Apply weighted principal component regression and weighted multiple regression methods.
• Conduct simulations to evaluate the effectiveness of these methods under different correlation levels.
• Analyze the impact of multicollinearity in regression estimation.
• Demonstrated that proposed methods reduce instability in estimates.
• Showed decreased overestimation of variance in complex sampling designs.
• Identified optimal estimation techniques for varying levels of correlations.