Median Moments with Application to Location-Scale Family of Distributions | ||||
The Egyptian Statistical Journal | ||||
Article 6, Volume 49, Issue 2, December 2005, Page 181-198 | ||||
Document Type: Original Article | ||||
DOI: 10.21608/esju.2005.313565 | ||||
View on SCiNiTO | ||||
Abstract | ||||
Model fitting is one of the basic activities in statistical analysis. Classical estimation methods such as method of moments and least squares are not robust. They are sensitive to outliers, to other contamination in the data, or to incorrect assumptions about an appropriate population model. Consequently, the estimating of distribution parameters using methods which produce efficient and robust estimators are needed. In this article, we define population and sample median moments as a robust version of population and sample L-moments in which the expectation of the order statistics of a conceptual sample of population L-moments is replaced by median of the order statistics of a conceptual sample. The method is illustrated by several examples and compared with other methods. The results seem to indicate that the method possesses high efficiency and robustness compared to other competitor methods such as linear moments (L-moments) and Linear Quick Moment (LQ-moments). | ||||
Keywords | ||||
Moments; Median - Order Statistics; Outliers - Robust Estimation | ||||
Statistics Article View: 25 |
||||