 2
 Hollander, M. and D. A. Wolfe. 1973.Nonparametric Statistical Methods. New York: John Wiley & Sons.
Total intext references: 1
 Intext reference with the coordinate start=168302
 Prefix

:[0,4]
Methods and Formulas
The Hodges–Lehmann method is an extension of the Wilcoxon–Mann–Whitney test to the problem of estimating the
shift parameter (by a median unbiased estimate and a confidence interval). The Hodges–Lehmann estimates were developed by
Hodges and Lehmann (1963) and are described in detail in
 Exact

Hollander and Wolfe (1973, 75–82)
 Suffix

–82), and in Lehmann (1975, 81–95).
Suppose, that we have two samples of data. Sample 1 consists ofmobservations,x1;:::;xmdrawn from the distributionF(x),
and sample 2 consists ofnobservations,y1;:::;yndrawn from the distributionG(y).
 3
 Lehmann, E. L. 1975.Nonparametrics: Statistical Methods Based on Ranks. San Francisco: Holden–Day.
Total intext references: 2
 Intext reference with the coordinate start=165763
 Prefix

Upper Saddle River, NJ: Prentice–Hall.
sg123Hodges–Lehmann estimation of a shift in location between two populations
Duolao Wang, London School of Hygiene and Tropical Medicine,UK, Duolao.Wang@lshtm.ac.uk
Introduction
The Hodges–Lehmann method (Hodges and Lehmann 1963;
 Exact

Lehmann 1975)
 Suffix

is a nonparametric procedure that extends the
Wilcoxon–Mann–Whitney test to the problem of estimating the shift parameter between two populations. This method gives
both a point estimate and a confidence interval for the shift parameter.
 Intext reference with the coordinate start=168344
 Prefix

Methods and Formulas
The Hodges–Lehmann method is an extension of the Wilcoxon–Mann–Whitney test to the problem of estimating the
shift parameter (by a median unbiased estimate and a confidence interval). The Hodges–Lehmann estimates were developed by
Hodges and Lehmann (1963) and are described in detail in Hollander and Wolfe (1973, 75–82), and in
 Exact

Lehmann (1975, 81–95)
 Suffix

–95).
Suppose, that we have two samples of data. Sample 1 consists ofmobservations,x1;:::;xmdrawn from the distributionF(x),
and sample 2 consists ofnobservations,y1;:::;yndrawn from the distributionG(y).