Chiang Mai Journal of Science

     Count data are prevalent across various disciplines, and the Poisson regression model (PRM) is often employed to analyze such data due to its widespread popularity. The model’s parameters are typically estimated using the maximum likelihood estimator (MLE). However, when multicollinearity exists among the explanatory variables, MLE may lead to unstable and unreliable parameter estimates. This is because multicollinearity can lead to inflated variances, increased prediction errors, incorrect parameter signs, and a higher mean squared error (MSE). To address the issue of multicollinearity in Poisson regression models, this study introduces a new general class of ridge-type Kibria–Lukman estimators designed to address multicollinearity in PRM. We examine the theoretical foundations of this estimator and its practical uses. We conduct theoretical comparisons with existing estimators and do a Monte Carlo simulation study across several situations to evaluate the efficacy of our proposed estimator. Ultimately, we demonstrate the superior efficacy of our estimator in mitigating multicollinearity in PRM via real-world data that validate our simulation results and theoretical analyses. Providing a powerful approach to data analysis and obtaining stable and reliable parameters.