On power approximations and comparison of several asymptotic tests to detect a specified difference between two proportions

Julia N Soulakova, Ananya Roy

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

A large sample test is proposed for a problem of testing for a specified difference between two binomial proportions. The test is compared to the tests by Falk and Koch (1998), and Parmet and Schechtman (2007), and is shown to dominate in terms of the Type I error rate control. Asymptotic power is derived for each test and is shown to result in values quite proximate to the simulated power values. In addition, formulas to perform sample size estimation are provided. These methods are expected to be especially valuable in the design stage when obtaining the correct power/sample size estimation is essential.

Original languageEnglish (US)
Pages (from-to)376-390
Number of pages15
JournalCommunications in Statistics: Simulation and Computation
Volume41
Issue number3
DOIs
StatePublished - Mar 1 2012

Fingerprint

Asymptotic Test
Proportion
Approximation
Sample Size
Addition formula
Asymptotic Power
Type I Error Rate
Rate Control
Error Control
Testing

Keywords

  • Binomial tests
  • Large-sample test
  • Non-inferiority problem
  • Normal approximation

ASJC Scopus subject areas

  • Modeling and Simulation
  • Statistics and Probability

Cite this

On power approximations and comparison of several asymptotic tests to detect a specified difference between two proportions. / Soulakova, Julia N; Roy, Ananya.

In: Communications in Statistics: Simulation and Computation, Vol. 41, No. 3, 01.03.2012, p. 376-390.

Research output: Contribution to journalArticle

@article{0888eab1b49a428980a4232f26a4e260,
title = "On power approximations and comparison of several asymptotic tests to detect a specified difference between two proportions",
abstract = "A large sample test is proposed for a problem of testing for a specified difference between two binomial proportions. The test is compared to the tests by Falk and Koch (1998), and Parmet and Schechtman (2007), and is shown to dominate in terms of the Type I error rate control. Asymptotic power is derived for each test and is shown to result in values quite proximate to the simulated power values. In addition, formulas to perform sample size estimation are provided. These methods are expected to be especially valuable in the design stage when obtaining the correct power/sample size estimation is essential.",
keywords = "Binomial tests, Large-sample test, Non-inferiority problem, Normal approximation",
author = "Soulakova, {Julia N} and Ananya Roy",
year = "2012",
month = "3",
day = "1",
doi = "10.1080/03610918.2011.592247",
language = "English (US)",
volume = "41",
pages = "376--390",
journal = "Communications in Statistics Part B: Simulation and Computation",
issn = "0361-0918",
publisher = "Taylor and Francis Ltd.",
number = "3",

}

TY - JOUR

T1 - On power approximations and comparison of several asymptotic tests to detect a specified difference between two proportions

AU - Soulakova, Julia N

AU - Roy, Ananya

PY - 2012/3/1

Y1 - 2012/3/1

N2 - A large sample test is proposed for a problem of testing for a specified difference between two binomial proportions. The test is compared to the tests by Falk and Koch (1998), and Parmet and Schechtman (2007), and is shown to dominate in terms of the Type I error rate control. Asymptotic power is derived for each test and is shown to result in values quite proximate to the simulated power values. In addition, formulas to perform sample size estimation are provided. These methods are expected to be especially valuable in the design stage when obtaining the correct power/sample size estimation is essential.

AB - A large sample test is proposed for a problem of testing for a specified difference between two binomial proportions. The test is compared to the tests by Falk and Koch (1998), and Parmet and Schechtman (2007), and is shown to dominate in terms of the Type I error rate control. Asymptotic power is derived for each test and is shown to result in values quite proximate to the simulated power values. In addition, formulas to perform sample size estimation are provided. These methods are expected to be especially valuable in the design stage when obtaining the correct power/sample size estimation is essential.

KW - Binomial tests

KW - Large-sample test

KW - Non-inferiority problem

KW - Normal approximation

UR - http://www.scopus.com/inward/record.url?scp=84856248141&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84856248141&partnerID=8YFLogxK

U2 - 10.1080/03610918.2011.592247

DO - 10.1080/03610918.2011.592247

M3 - Article

VL - 41

SP - 376

EP - 390

JO - Communications in Statistics Part B: Simulation and Computation

JF - Communications in Statistics Part B: Simulation and Computation

SN - 0361-0918

IS - 3

ER -