Modeling and inference for an ordinal effect size measure

Euijung Ryu, Alan Agresti

Research output: Contribution to journalArticle

29 Citations (Scopus)

Abstract

An ordinal measure of effect size is a simple and useful way to describe the difference between two ordered categorical distributions. This measure summarizes the probability that an outcome from one distribution falls above an outcome from the other, adjusted for ties. We develop and compare confidence interval methods for the measure. Simulation studies show that with independent multinominal samples, confidence intervals based on inverting the score test and a pseudo-score-type test perform well. This score method also seems to work well with fully-ranked data, but for dependent samples a simple Wald interval on the logit scale can be better with small samples. We also explore how the ordinal effect size measure relates to an effect measure commonly used for normal distributions, and we consider a logit model for describing how it depends on explanatory variables. The methods are illustrated for a study comparing treatments for shoulder-tip pain.

Original languageEnglish (US)
Pages (from-to)1703-1717
Number of pages15
JournalStatistics in Medicine
Volume27
Issue number10
DOIs
StatePublished - May 10 2008

Fingerprint

Effect Size
Modeling
Confidence Intervals
Shoulder Pain
Normal Distribution
Confidence interval
Logistic Models
Interval Methods
Logit Model
Logit
Score Test
Pain
Tie
Small Sample
Categorical
Gaussian distribution
Simulation Study
Interval
Dependent
Therapeutics

Keywords

  • Confidence intervals
  • Logit models
  • Mann-Whitney statistic
  • Matched pairs
  • Multinomial distributions
  • Ordinal data

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability

Cite this

Modeling and inference for an ordinal effect size measure. / Ryu, Euijung; Agresti, Alan.

In: Statistics in Medicine, Vol. 27, No. 10, 10.05.2008, p. 1703-1717.

Research output: Contribution to journalArticle

Ryu, Euijung ; Agresti, Alan. / Modeling and inference for an ordinal effect size measure. In: Statistics in Medicine. 2008 ; Vol. 27, No. 10. pp. 1703-1717.
@article{01cf332f26e048f28357d8be282c69a1,
title = "Modeling and inference for an ordinal effect size measure",
abstract = "An ordinal measure of effect size is a simple and useful way to describe the difference between two ordered categorical distributions. This measure summarizes the probability that an outcome from one distribution falls above an outcome from the other, adjusted for ties. We develop and compare confidence interval methods for the measure. Simulation studies show that with independent multinominal samples, confidence intervals based on inverting the score test and a pseudo-score-type test perform well. This score method also seems to work well with fully-ranked data, but for dependent samples a simple Wald interval on the logit scale can be better with small samples. We also explore how the ordinal effect size measure relates to an effect measure commonly used for normal distributions, and we consider a logit model for describing how it depends on explanatory variables. The methods are illustrated for a study comparing treatments for shoulder-tip pain.",
keywords = "Confidence intervals, Logit models, Mann-Whitney statistic, Matched pairs, Multinomial distributions, Ordinal data",
author = "Euijung Ryu and Alan Agresti",
year = "2008",
month = "5",
day = "10",
doi = "10.1002/sim.3079",
language = "English (US)",
volume = "27",
pages = "1703--1717",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",
number = "10",

}

TY - JOUR

T1 - Modeling and inference for an ordinal effect size measure

AU - Ryu, Euijung

AU - Agresti, Alan

PY - 2008/5/10

Y1 - 2008/5/10

N2 - An ordinal measure of effect size is a simple and useful way to describe the difference between two ordered categorical distributions. This measure summarizes the probability that an outcome from one distribution falls above an outcome from the other, adjusted for ties. We develop and compare confidence interval methods for the measure. Simulation studies show that with independent multinominal samples, confidence intervals based on inverting the score test and a pseudo-score-type test perform well. This score method also seems to work well with fully-ranked data, but for dependent samples a simple Wald interval on the logit scale can be better with small samples. We also explore how the ordinal effect size measure relates to an effect measure commonly used for normal distributions, and we consider a logit model for describing how it depends on explanatory variables. The methods are illustrated for a study comparing treatments for shoulder-tip pain.

AB - An ordinal measure of effect size is a simple and useful way to describe the difference between two ordered categorical distributions. This measure summarizes the probability that an outcome from one distribution falls above an outcome from the other, adjusted for ties. We develop and compare confidence interval methods for the measure. Simulation studies show that with independent multinominal samples, confidence intervals based on inverting the score test and a pseudo-score-type test perform well. This score method also seems to work well with fully-ranked data, but for dependent samples a simple Wald interval on the logit scale can be better with small samples. We also explore how the ordinal effect size measure relates to an effect measure commonly used for normal distributions, and we consider a logit model for describing how it depends on explanatory variables. The methods are illustrated for a study comparing treatments for shoulder-tip pain.

KW - Confidence intervals

KW - Logit models

KW - Mann-Whitney statistic

KW - Matched pairs

KW - Multinomial distributions

KW - Ordinal data

UR - http://www.scopus.com/inward/record.url?scp=42449146735&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=42449146735&partnerID=8YFLogxK

U2 - 10.1002/sim.3079

DO - 10.1002/sim.3079

M3 - Article

C2 - 17918752

AN - SCOPUS:42449146735

VL - 27

SP - 1703

EP - 1717

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 10

ER -