A large number of experiments in agronomic sciences use variables that may give rise to problems of multicollinearity. About the applicability of regression models, the problem of multicollinearity results mainly in increased standard error, thus, the Student's t-value is reduced, affecting the inferential results. Many actions are proposed in the literature to solve the problems of multicollinearity, however, the performance of these measurements are subject to the degree that multicollinearity of the variables may present, as well as the sample size. To address this problem, this paper aims to evaluate some ridge estimators using the Monte Carlo's simulation and demonstrate their application using real data from an entomological experiment. The ridge estimators evaluated were effective, in comparison with the least squares estimator. The results showed that the ridge estimators evaluated can be applied to experimenst that consider the variables with different degrees of multicollinearity, for samples greater than n=50.
multicollinearity; sample size; regression models
