Random Effects: Variance Is the Spice of Life

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Covariates in regression analyses allow us to understand how independent variables of interest impact our dependent outcome variable. Often, we consider fixed effects covariates (e.g., gender or diabetes status) for which we examine subjects at each value of the covariate. We examine both men and women and, within each gender, examine both diabetic and nondiabetic patients. Occasionally, however, we consider random effects covariates for which we do not examine subjects at every value. For example, we examine patients from only a sample of hospitals and, within each hospital, examine both diabetic and nondiabetic patients. The random sampling of hospitals is in contrast to the complete coverage of all genders. In this column I explore the differences in meaning and analysis when thinking about fixed and random effects variables.

Original languageEnglish (US)
Pages (from-to)1343-1346
Number of pages4
JournalJournal of Foot and Ankle Surgery
Volume55
Issue number6
DOIs
StatePublished - Nov 1 2016

Fingerprint

Regression Analysis

Keywords

  • ANOVA
  • random effects
  • regression
  • t test
  • variance

ASJC Scopus subject areas

  • Surgery
  • Orthopedics and Sports Medicine

Cite this

Random Effects : Variance Is the Spice of Life. / Jupiter, Daniel.

In: Journal of Foot and Ankle Surgery, Vol. 55, No. 6, 01.11.2016, p. 1343-1346.

Research output: Contribution to journalArticle

@article{f5f0d0314e694378949a9ef8cd2417f4,
title = "Random Effects: Variance Is the Spice of Life",
abstract = "Covariates in regression analyses allow us to understand how independent variables of interest impact our dependent outcome variable. Often, we consider fixed effects covariates (e.g., gender or diabetes status) for which we examine subjects at each value of the covariate. We examine both men and women and, within each gender, examine both diabetic and nondiabetic patients. Occasionally, however, we consider random effects covariates for which we do not examine subjects at every value. For example, we examine patients from only a sample of hospitals and, within each hospital, examine both diabetic and nondiabetic patients. The random sampling of hospitals is in contrast to the complete coverage of all genders. In this column I explore the differences in meaning and analysis when thinking about fixed and random effects variables.",
keywords = "ANOVA, random effects, regression, t test, variance",
author = "Daniel Jupiter",
year = "2016",
month = "11",
day = "1",
doi = "10.1053/j.jfas.2016.09.003",
language = "English (US)",
volume = "55",
pages = "1343--1346",
journal = "Journal of Foot and Ankle Surgery",
issn = "1067-2516",
publisher = "Academic Press Inc.",
number = "6",

}

TY - JOUR

T1 - Random Effects

T2 - Variance Is the Spice of Life

AU - Jupiter, Daniel

PY - 2016/11/1

Y1 - 2016/11/1

N2 - Covariates in regression analyses allow us to understand how independent variables of interest impact our dependent outcome variable. Often, we consider fixed effects covariates (e.g., gender or diabetes status) for which we examine subjects at each value of the covariate. We examine both men and women and, within each gender, examine both diabetic and nondiabetic patients. Occasionally, however, we consider random effects covariates for which we do not examine subjects at every value. For example, we examine patients from only a sample of hospitals and, within each hospital, examine both diabetic and nondiabetic patients. The random sampling of hospitals is in contrast to the complete coverage of all genders. In this column I explore the differences in meaning and analysis when thinking about fixed and random effects variables.

AB - Covariates in regression analyses allow us to understand how independent variables of interest impact our dependent outcome variable. Often, we consider fixed effects covariates (e.g., gender or diabetes status) for which we examine subjects at each value of the covariate. We examine both men and women and, within each gender, examine both diabetic and nondiabetic patients. Occasionally, however, we consider random effects covariates for which we do not examine subjects at every value. For example, we examine patients from only a sample of hospitals and, within each hospital, examine both diabetic and nondiabetic patients. The random sampling of hospitals is in contrast to the complete coverage of all genders. In this column I explore the differences in meaning and analysis when thinking about fixed and random effects variables.

KW - ANOVA

KW - random effects

KW - regression

KW - t test

KW - variance

UR - http://www.scopus.com/inward/record.url?scp=84994691747&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84994691747&partnerID=8YFLogxK

U2 - 10.1053/j.jfas.2016.09.003

DO - 10.1053/j.jfas.2016.09.003

M3 - Article

C2 - 27769595

AN - SCOPUS:84994691747

VL - 55

SP - 1343

EP - 1346

JO - Journal of Foot and Ankle Surgery

JF - Journal of Foot and Ankle Surgery

SN - 1067-2516

IS - 6

ER -