Optimal and efficient crossover designs when subject effects are random

A. S. Hedayat, John Stufken, Min Yang

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

Most studies on optimal crossover designs are based on models that assume subject effects to be fixed effects. In this article we identify and study optimal and efficient designs for a model with random subject effects. With the number of periods not exceeding the number of treatments, we find that totally balanced designs are universally optimal for treatment effects in a large subclass of competing designs. However, in the entire class of designs, totally balanced designs are in general not optimal, and their efficiency depends on the ratio of the subject effects variance and the error variance. We develop tools to study the efficiency of totally balanced designs and to identify designs with higher efficiency.

Original languageEnglish (US)
Pages (from-to)1031-1038
Number of pages8
JournalJournal of the American Statistical Association
Volume101
Issue number475
DOIs
StatePublished - Sep 2006
Externally publishedYes

Fingerprint

Crossover Design
Balanced Design
Fixed Effects
Treatment Effects
Random Effects
High Efficiency
Entire
Design
Crossover
Model

Keywords

  • Fisher information matrix
  • Mixed-effects model
  • Totally balanced design
  • Universal optimality

ASJC Scopus subject areas

  • Mathematics(all)
  • Statistics and Probability

Cite this

Optimal and efficient crossover designs when subject effects are random. / Hedayat, A. S.; Stufken, John; Yang, Min.

In: Journal of the American Statistical Association, Vol. 101, No. 475, 09.2006, p. 1031-1038.

Research output: Contribution to journalArticle

Hedayat, A. S. ; Stufken, John ; Yang, Min. / Optimal and efficient crossover designs when subject effects are random. In: Journal of the American Statistical Association. 2006 ; Vol. 101, No. 475. pp. 1031-1038.
@article{9dc40a0dfb9341ca9cb2c216c1fee867,
title = "Optimal and efficient crossover designs when subject effects are random",
abstract = "Most studies on optimal crossover designs are based on models that assume subject effects to be fixed effects. In this article we identify and study optimal and efficient designs for a model with random subject effects. With the number of periods not exceeding the number of treatments, we find that totally balanced designs are universally optimal for treatment effects in a large subclass of competing designs. However, in the entire class of designs, totally balanced designs are in general not optimal, and their efficiency depends on the ratio of the subject effects variance and the error variance. We develop tools to study the efficiency of totally balanced designs and to identify designs with higher efficiency.",
keywords = "Fisher information matrix, Mixed-effects model, Totally balanced design, Universal optimality",
author = "Hedayat, {A. S.} and John Stufken and Min Yang",
year = "2006",
month = "9",
doi = "10.1198/016214505000001384",
language = "English (US)",
volume = "101",
pages = "1031--1038",
journal = "Journal of the American Statistical Association",
issn = "0162-1459",
publisher = "Taylor and Francis Ltd.",
number = "475",

}

TY - JOUR

T1 - Optimal and efficient crossover designs when subject effects are random

AU - Hedayat, A. S.

AU - Stufken, John

AU - Yang, Min

PY - 2006/9

Y1 - 2006/9

N2 - Most studies on optimal crossover designs are based on models that assume subject effects to be fixed effects. In this article we identify and study optimal and efficient designs for a model with random subject effects. With the number of periods not exceeding the number of treatments, we find that totally balanced designs are universally optimal for treatment effects in a large subclass of competing designs. However, in the entire class of designs, totally balanced designs are in general not optimal, and their efficiency depends on the ratio of the subject effects variance and the error variance. We develop tools to study the efficiency of totally balanced designs and to identify designs with higher efficiency.

AB - Most studies on optimal crossover designs are based on models that assume subject effects to be fixed effects. In this article we identify and study optimal and efficient designs for a model with random subject effects. With the number of periods not exceeding the number of treatments, we find that totally balanced designs are universally optimal for treatment effects in a large subclass of competing designs. However, in the entire class of designs, totally balanced designs are in general not optimal, and their efficiency depends on the ratio of the subject effects variance and the error variance. We develop tools to study the efficiency of totally balanced designs and to identify designs with higher efficiency.

KW - Fisher information matrix

KW - Mixed-effects model

KW - Totally balanced design

KW - Universal optimality

UR - http://www.scopus.com/inward/record.url?scp=33748861765&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33748861765&partnerID=8YFLogxK

U2 - 10.1198/016214505000001384

DO - 10.1198/016214505000001384

M3 - Article

AN - SCOPUS:33748861765

VL - 101

SP - 1031

EP - 1038

JO - Journal of the American Statistical Association

JF - Journal of the American Statistical Association

SN - 0162-1459

IS - 475

ER -