Abstract

In the linear regression model, the least-squares (LS) estimator is

commonly used to estimate regression parameters. However, LS becomes

unreliable and unfavorable when the model is affected by multicollinearity

and outliers simultaneously. Numerous authors have proposed various

estimators to address the challenges of multicollinearity and outliers in

linear regression models. This paper introduces an alternative robust

regression estimator, called the Two-Stage Robust Dawoud–Kibria

estimator, designed to address the two issues simultaneously. We

performed theoretical comparisons, conducted simulations under different

scenarios to illustrate the effectiveness of the proposed estimator.

Theoretical analysis and simulation results indicate that the proposed

estimator outperforms other regression estimators under certain conditions

when both multicollinearity and outlier issues are present, based on the

mean squared error criterion.

Key words: Two-Stage Robust Dawoud–Kibria estimator (MMDK),

Robust Dawoud–Kibria estimator (MDK), Robust Liu (M-Liu), Robust

Ridge (M-Ridge), Robust Özkale–Kaçiranlar (MOK), M-estimator (ME),

.Non-Robust estimators, multicollinearity, outliers, Mean squared errors(MSEs)