The effects of misconceptions on the properties of friedman's test

Roy St Laurent, Philip Turk

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Friedman's test is a widely used rank-based alternative to the analysis of variance (ANOVA) F-test for identifying treatment differences in a randomized complete block design. Many texts provide incomplete or misleading information about when Friedman's test may be appropriately applied. We discuss the assumptions needed for the test and common misconceptions. We show via simulation that when the variance or skew of the treatment distributions differ, application of Friedman's test to detect differences in treatment location can result in Type I error probabilities larger than the nominal α, and even when α is unaffected, the power of the test can be less than expected.

Original languageEnglish (US)
Pages (from-to)1596-1615
Number of pages20
JournalCommunications in Statistics: Simulation and Computation
Volume42
Issue number7
DOIs
StatePublished - Aug 1 2013

Fingerprint

Misconceptions
Analysis of variance (ANOVA)
F Test
Block Design
Type I error
Analysis of variance
Error Probability
Skew
Categorical or nominal
Error probability
Alternatives
Simulation

Keywords

  • Analysis of variance
  • Heteroskedasticity
  • Nonparametric test
  • Power
  • Size of test
  • Skew

ASJC Scopus subject areas

  • Modeling and Simulation
  • Statistics and Probability

Cite this

The effects of misconceptions on the properties of friedman's test. / Laurent, Roy St; Turk, Philip.

In: Communications in Statistics: Simulation and Computation, Vol. 42, No. 7, 01.08.2013, p. 1596-1615.

Research output: Contribution to journalArticle

@article{8f7d44a1a64a4768871f2e36a791928d,
title = "The effects of misconceptions on the properties of friedman's test",
abstract = "Friedman's test is a widely used rank-based alternative to the analysis of variance (ANOVA) F-test for identifying treatment differences in a randomized complete block design. Many texts provide incomplete or misleading information about when Friedman's test may be appropriately applied. We discuss the assumptions needed for the test and common misconceptions. We show via simulation that when the variance or skew of the treatment distributions differ, application of Friedman's test to detect differences in treatment location can result in Type I error probabilities larger than the nominal α, and even when α is unaffected, the power of the test can be less than expected.",
keywords = "Analysis of variance, Heteroskedasticity, Nonparametric test, Power, Size of test, Skew",
author = "Laurent, {Roy St} and Philip Turk",
year = "2013",
month = "8",
day = "1",
doi = "10.1080/03610918.2012.671874",
language = "English (US)",
volume = "42",
pages = "1596--1615",
journal = "Communications in Statistics Part B: Simulation and Computation",
issn = "0361-0918",
publisher = "Taylor and Francis Ltd.",
number = "7",

}

TY - JOUR

T1 - The effects of misconceptions on the properties of friedman's test

AU - Laurent, Roy St

AU - Turk, Philip

PY - 2013/8/1

Y1 - 2013/8/1

N2 - Friedman's test is a widely used rank-based alternative to the analysis of variance (ANOVA) F-test for identifying treatment differences in a randomized complete block design. Many texts provide incomplete or misleading information about when Friedman's test may be appropriately applied. We discuss the assumptions needed for the test and common misconceptions. We show via simulation that when the variance or skew of the treatment distributions differ, application of Friedman's test to detect differences in treatment location can result in Type I error probabilities larger than the nominal α, and even when α is unaffected, the power of the test can be less than expected.

AB - Friedman's test is a widely used rank-based alternative to the analysis of variance (ANOVA) F-test for identifying treatment differences in a randomized complete block design. Many texts provide incomplete or misleading information about when Friedman's test may be appropriately applied. We discuss the assumptions needed for the test and common misconceptions. We show via simulation that when the variance or skew of the treatment distributions differ, application of Friedman's test to detect differences in treatment location can result in Type I error probabilities larger than the nominal α, and even when α is unaffected, the power of the test can be less than expected.

KW - Analysis of variance

KW - Heteroskedasticity

KW - Nonparametric test

KW - Power

KW - Size of test

KW - Skew

UR - http://www.scopus.com/inward/record.url?scp=84873411207&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84873411207&partnerID=8YFLogxK

U2 - 10.1080/03610918.2012.671874

DO - 10.1080/03610918.2012.671874

M3 - Article

AN - SCOPUS:84873411207

VL - 42

SP - 1596

EP - 1615

JO - Communications in Statistics Part B: Simulation and Computation

JF - Communications in Statistics Part B: Simulation and Computation

SN - 0361-0918

IS - 7

ER -