Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/13059
Title: Statistical inference for measures of stochastic ordering in comparative studies
Authors: ZHAO YUDONG
Keywords: extended logistic family; boundary-respecting confidence interval; Mann-Whitney measure;Mann-Whitney statistic; stochastic ordering; ROC analysis
Issue Date: 8-Feb-2008
Source: ZHAO YUDONG (2008-02-08). Statistical inference for measures of stochastic ordering in comparative studies. ScholarBank@NUS Repository.
Abstract: A simple semi-parametric method is introduced for constructing boundary-respecting confidence intervals for the two-sample Mann-Whitney measure.We show that stochastic ordering is equivalent to monotone transformation of location shift. Then consideration of a number of location shift families indicates a suitable class of shapes to model the asymptotic variance, leading to a confidence interval method based on roots of quadratics. This proposed scheme is also adapted to paired data. We show that a stochastically positive random variable and its negative can be transformed to a symmetric location shift model. We further give a central place to the logistic location shift model in the boundary-respecting interval procedure through a variance-controlling transformation, simplifying the calculations. We also generalize a distribution family from the logistic distribution, thus covering a wide range of symmetric distribution shapes. Based on this extended logistic family, we develop rank procedures which can always retain high efficiency for one-sample location problems.
URI: http://scholarbank.nus.edu.sg/handle/10635/13059
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
ZhaoYudong_Thesis.pdf557.29 kBAdobe PDF

OPEN

NoneView/Download

Page view(s)

248
checked on Dec 18, 2017

Download(s)

198
checked on Dec 18, 2017

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.