Statistical inferences based on rank estimation
Date
2013-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
A Rank estimation of a linear model is proposed by minimizing a dispersion function. A class of simple confidence intervals and simultaneous confidence intervals based on rank estimation is proposed for the multiple linear regressions and their small sample behavior is studied by comparing the coverage probabilities. Simulation studies show that simple and simultaneous confidence intervals based on bootstrap-t method is the most impressive method.
Description
Keywords
Rank estimation, Simultaneous confidence intervals, Bootstrap, Coverage probability, Family-wise error rate