Testing multiple dispersion effects in unreplicated fractional factorial designs

Richard N. McGrath, Dennis K.J. Lin

Research output: Contribution to journalArticle

28 Citations (Scopus)

Abstract

In unreplicated 2k-p designs, the assumption of constant variance is commonly made. When the variance of the response differs between the two levels of a column in the effect matrix, that column produces a dispersion effect. In this article we show that two active dispersion effects may create a spurious dispersion effect in their interaction column. Most existing methods of dispersion-effect testing in unreplicated fractional designs are subject to these spurious effects. We propose a method of dispersion-effect testing based on geometric means of residual sample variance. We show through examples from the literature and simulations that the proposed test has many desirable properties that are lacking in other tests.

Original languageEnglish (US)
Pages (from-to)406-414
Number of pages9
JournalTechnometrics
Volume43
Issue number4
DOIs
StatePublished - Jan 1 2001

Fingerprint

Dispersion Effect
Fractional Factorial Design
Multiple Testing
Testing
Sample variance
Geometric mean
Fractional
Interaction
Simulation

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Modeling and Simulation
  • Applied Mathematics

Cite this

@article{5c8d11bc2b484c039c24f7f4c50378db,
title = "Testing multiple dispersion effects in unreplicated fractional factorial designs",
abstract = "In unreplicated 2k-p designs, the assumption of constant variance is commonly made. When the variance of the response differs between the two levels of a column in the effect matrix, that column produces a dispersion effect. In this article we show that two active dispersion effects may create a spurious dispersion effect in their interaction column. Most existing methods of dispersion-effect testing in unreplicated fractional designs are subject to these spurious effects. We propose a method of dispersion-effect testing based on geometric means of residual sample variance. We show through examples from the literature and simulations that the proposed test has many desirable properties that are lacking in other tests.",
author = "McGrath, {Richard N.} and Lin, {Dennis K.J.}",
year = "2001",
month = "1",
day = "1",
doi = "10.1198/00401700152672492",
language = "English (US)",
volume = "43",
pages = "406--414",
journal = "Technometrics",
issn = "0040-1706",
publisher = "American Statistical Association",
number = "4",

}

Testing multiple dispersion effects in unreplicated fractional factorial designs. / McGrath, Richard N.; Lin, Dennis K.J.

In: Technometrics, Vol. 43, No. 4, 01.01.2001, p. 406-414.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Testing multiple dispersion effects in unreplicated fractional factorial designs

AU - McGrath, Richard N.

AU - Lin, Dennis K.J.

PY - 2001/1/1

Y1 - 2001/1/1

N2 - In unreplicated 2k-p designs, the assumption of constant variance is commonly made. When the variance of the response differs between the two levels of a column in the effect matrix, that column produces a dispersion effect. In this article we show that two active dispersion effects may create a spurious dispersion effect in their interaction column. Most existing methods of dispersion-effect testing in unreplicated fractional designs are subject to these spurious effects. We propose a method of dispersion-effect testing based on geometric means of residual sample variance. We show through examples from the literature and simulations that the proposed test has many desirable properties that are lacking in other tests.

AB - In unreplicated 2k-p designs, the assumption of constant variance is commonly made. When the variance of the response differs between the two levels of a column in the effect matrix, that column produces a dispersion effect. In this article we show that two active dispersion effects may create a spurious dispersion effect in their interaction column. Most existing methods of dispersion-effect testing in unreplicated fractional designs are subject to these spurious effects. We propose a method of dispersion-effect testing based on geometric means of residual sample variance. We show through examples from the literature and simulations that the proposed test has many desirable properties that are lacking in other tests.

UR - http://www.scopus.com/inward/record.url?scp=0035516786&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035516786&partnerID=8YFLogxK

U2 - 10.1198/00401700152672492

DO - 10.1198/00401700152672492

M3 - Article

AN - SCOPUS:0035516786

VL - 43

SP - 406

EP - 414

JO - Technometrics

JF - Technometrics

SN - 0040-1706

IS - 4

ER -