The Variance Estimation for Pareto Distribution: A Simulation Study

Abstract

This research proposes the variance estimation method of Pareto data and to compare three variance estimation methods. Those methods are simple method, the adjusted variance method with term (n - 1)/n, and the adjusted variance method with term (n - 2)/n. The research was considered by the absolute of bias (|Bias|), the mean absolute errors (MAE) and the mean square errors (MSE) of variance estimator. This research used Monte Carlo Simulation method. Results of this research show that the |Bias| of simple method is the lowest for almost all conditions. We can arrange the methods as follows: the adjusted variance method with term (n - 1)/n and the adjusted variance method with term (n - 2)/n, respectively. For all situations, the MAE and MSE the of adjusted variance method with term (n - 2)/n is the lowest. We can arrange the methods as follows: the adjusted variance method with term (n - 1)/n and simple method, respectively.