February 2006 #1 Comparing Estimators of Quartiles Under Various Models by Steven T. Garren


by Steven T. Garren.

Abstract: Some authors define the first quartile of a data set to be the median of the ordered data strictly to the left of the overall median. Other authors define the first quartile to be the median of the ordered data to the left of and including the overall median. We propose a third estimator (of the population first quartile) which has minimum mean squared error under normality, among weighted averages of two order statistics, and robustness against nonnormal distributions is examined. Likewise, estimators of the third quartile may be defined. The three estimators of first and third quartiles are compared in terms of mean squared error under normality, a t-distribution with 3 degrees of freedom, a uniform distribution, and an exponential distribution. The preferred estimator depends on the distribution and the sample size modulus 4, for sample sizes no larger than 30.

Key Words: Exponential distribution, Normal distribution, t-distribution, Uniform distribution

Steven T. Garren, garrenst@jmu.edu

Editor: Richard Graf,rgraf@sunstroke.sdsu.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (179403 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 4821 times since July 24, 2006.

Return to the InterStat Home Page.