In linear regression analysis, the assumption of independent variables is fundamental, and the ordinary least squares (OLS) estimator is generally considered as the best linear unbiased estimator (BLUE). However, when multicollinearity is present, the unique effects of individual variables become challenging to estimate, leading to incorrect statistical inferences. To address this issue, various alternative biasing estimators are discussed in the literature. Using a Monte Carlo study, this paper focuses on comparing different t-test statistics for testing the significance of linear regression coefficients, including OLS, ridge regression, the Liu estimator, the one-parameter modified Liu estimator, the James-Stein estimator, and the Kibria-Lukman one- and two-parameter biased estimators. The comparison focuses on empirical type I error and power properties, following established testing procedures. The simulation study revealed that some of the ridge, Liu, Kibria-Lukman, and Stein estimators exhibited superior power gains while maintaining a \(5\%\) nominal size.