Effect of regression to the mean in multivariate distributions

Varghese George, William D. Johnson

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Estimating treatment effects in the presence of regression to the mean is a problem arising in truncated distributions that is being recognized with increasing interest in recent literature. As noted in a previous communication by the authors (1991), any extraneous source of variability such as within-subject variability and measurement errors can contribute to the magnitude of regression toward the mean. The main focus of this paper is consideration of a model for estimating treatment effects when truncation and regression to the mean occur on more than one random variable. This situation occurs often in investigations where subjects are selected for study because measurements on two characteristics of interest both exceed specified values.

Original languageEnglish (US)
Pages (from-to)333-350
Number of pages18
JournalCommunications in Statistics - Theory and Methods
Volume21
Issue number2
DOIs
StatePublished - Jan 1 1992
Externally publishedYes

Fingerprint

Multivariate Distribution
Regression
Treatment Effects
Truncated Distributions
Measurement Error
Truncation
Exceed
Random variable
Model

Keywords

  • bivariate normal distribution
  • measurement error
  • regression to the mean
  • repeated measurements
  • replicate measurements
  • truncation
  • within-subject effect

ASJC Scopus subject areas

  • Statistics and Probability

Cite this

Effect of regression to the mean in multivariate distributions. / George, Varghese; Johnson, William D.

In: Communications in Statistics - Theory and Methods, Vol. 21, No. 2, 01.01.1992, p. 333-350.

Research output: Contribution to journalArticle

@article{901a7af47ed84cc2b9910ca0c2c1c593,
title = "Effect of regression to the mean in multivariate distributions",
abstract = "Estimating treatment effects in the presence of regression to the mean is a problem arising in truncated distributions that is being recognized with increasing interest in recent literature. As noted in a previous communication by the authors (1991), any extraneous source of variability such as within-subject variability and measurement errors can contribute to the magnitude of regression toward the mean. The main focus of this paper is consideration of a model for estimating treatment effects when truncation and regression to the mean occur on more than one random variable. This situation occurs often in investigations where subjects are selected for study because measurements on two characteristics of interest both exceed specified values.",
keywords = "bivariate normal distribution, measurement error, regression to the mean, repeated measurements, replicate measurements, truncation, within-subject effect",
author = "Varghese George and Johnson, {William D.}",
year = "1992",
month = "1",
day = "1",
doi = "10.1080/03610929208830782",
language = "English (US)",
volume = "21",
pages = "333--350",
journal = "Communications in Statistics - Theory and Methods",
issn = "0361-0926",
publisher = "Taylor and Francis Ltd.",
number = "2",

}

TY - JOUR

T1 - Effect of regression to the mean in multivariate distributions

AU - George, Varghese

AU - Johnson, William D.

PY - 1992/1/1

Y1 - 1992/1/1

N2 - Estimating treatment effects in the presence of regression to the mean is a problem arising in truncated distributions that is being recognized with increasing interest in recent literature. As noted in a previous communication by the authors (1991), any extraneous source of variability such as within-subject variability and measurement errors can contribute to the magnitude of regression toward the mean. The main focus of this paper is consideration of a model for estimating treatment effects when truncation and regression to the mean occur on more than one random variable. This situation occurs often in investigations where subjects are selected for study because measurements on two characteristics of interest both exceed specified values.

AB - Estimating treatment effects in the presence of regression to the mean is a problem arising in truncated distributions that is being recognized with increasing interest in recent literature. As noted in a previous communication by the authors (1991), any extraneous source of variability such as within-subject variability and measurement errors can contribute to the magnitude of regression toward the mean. The main focus of this paper is consideration of a model for estimating treatment effects when truncation and regression to the mean occur on more than one random variable. This situation occurs often in investigations where subjects are selected for study because measurements on two characteristics of interest both exceed specified values.

KW - bivariate normal distribution

KW - measurement error

KW - regression to the mean

KW - repeated measurements

KW - replicate measurements

KW - truncation

KW - within-subject effect

UR - http://www.scopus.com/inward/record.url?scp=0342478144&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0342478144&partnerID=8YFLogxK

U2 - 10.1080/03610929208830782

DO - 10.1080/03610929208830782

M3 - Article

AN - SCOPUS:0342478144

VL - 21

SP - 333

EP - 350

JO - Communications in Statistics - Theory and Methods

JF - Communications in Statistics - Theory and Methods

SN - 0361-0926

IS - 2

ER -