Tests of Simple Slopes in Multiple Regression Models with an Interaction

Comparison of Four Approaches

Yu Liu, Stephen West, Roy Levy, Leona S. Aiken

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

In multiple regression researchers often follow up significant tests of the interaction between continuous predictors X and Z with tests of the simple slope of Y on X at different sample-estimated values of the moderator Z (e.g., ±1 SD from the mean of Z). We show analytically that when X and Z are randomly sampled from the population, the variance expression of the simple slope at sample-estimated values of Z differs from the traditional variance expression obtained when the values of X and Z are fixed. A simulation study using randomly sampled predictors compared four approaches: (a) the Aiken and West (1991) test of simple slopes at fixed population values of Z, (b) the Aiken and West test at sample-estimated values of Z, (c) a 95% percentile bootstrap confidence interval approach, and (d) a fully Bayesian approach with diffuse priors. The results showed that approach (b) led to inflated Type 1 error rates and 95% confidence intervals with inadequate coverage rates, whereas other approaches maintained acceptable Type 1 error rates and adequate coverage of confidence intervals. Approach (c) had asymmetric rejection rates at small sample sizes. We used an empirical data set to illustrate these approaches.

Original languageEnglish (US)
Pages (from-to)1-20
Number of pages20
JournalMultivariate Behavioral Research
DOIs
StateAccepted/In press - Apr 27 2017

Fingerprint

Multiple Regression
Multiple Models
Regression Model
Slope
Confidence Intervals
Interaction
Confidence interval
Error Rate
Predictors
Coverage
Bayes Theorem
Bootstrap Confidence Intervals
Sample Size
Population
Percentile
Small Sample Size
Rejection
Research Personnel
Bayesian Approach
Simulation Study

Keywords

  • Bayesian
  • bootstrap
  • interaction
  • multiple regression
  • simple slope

ASJC Scopus subject areas

  • Statistics and Probability
  • Experimental and Cognitive Psychology
  • Arts and Humanities (miscellaneous)

Cite this

@article{ee27447fdf12456cb35a42fe96b5b894,
title = "Tests of Simple Slopes in Multiple Regression Models with an Interaction: Comparison of Four Approaches",
abstract = "In multiple regression researchers often follow up significant tests of the interaction between continuous predictors X and Z with tests of the simple slope of Y on X at different sample-estimated values of the moderator Z (e.g., ±1 SD from the mean of Z). We show analytically that when X and Z are randomly sampled from the population, the variance expression of the simple slope at sample-estimated values of Z differs from the traditional variance expression obtained when the values of X and Z are fixed. A simulation study using randomly sampled predictors compared four approaches: (a) the Aiken and West (1991) test of simple slopes at fixed population values of Z, (b) the Aiken and West test at sample-estimated values of Z, (c) a 95{\%} percentile bootstrap confidence interval approach, and (d) a fully Bayesian approach with diffuse priors. The results showed that approach (b) led to inflated Type 1 error rates and 95{\%} confidence intervals with inadequate coverage rates, whereas other approaches maintained acceptable Type 1 error rates and adequate coverage of confidence intervals. Approach (c) had asymmetric rejection rates at small sample sizes. We used an empirical data set to illustrate these approaches.",
keywords = "Bayesian, bootstrap, interaction, multiple regression, simple slope",
author = "Yu Liu and Stephen West and Roy Levy and Aiken, {Leona S.}",
year = "2017",
month = "4",
day = "27",
doi = "10.1080/00273171.2017.1309261",
language = "English (US)",
pages = "1--20",
journal = "Multivariate Behavioral Research",
issn = "0027-3171",
publisher = "Psychology Press Ltd",

}

TY - JOUR

T1 - Tests of Simple Slopes in Multiple Regression Models with an Interaction

T2 - Comparison of Four Approaches

AU - Liu, Yu

AU - West, Stephen

AU - Levy, Roy

AU - Aiken, Leona S.

PY - 2017/4/27

Y1 - 2017/4/27

N2 - In multiple regression researchers often follow up significant tests of the interaction between continuous predictors X and Z with tests of the simple slope of Y on X at different sample-estimated values of the moderator Z (e.g., ±1 SD from the mean of Z). We show analytically that when X and Z are randomly sampled from the population, the variance expression of the simple slope at sample-estimated values of Z differs from the traditional variance expression obtained when the values of X and Z are fixed. A simulation study using randomly sampled predictors compared four approaches: (a) the Aiken and West (1991) test of simple slopes at fixed population values of Z, (b) the Aiken and West test at sample-estimated values of Z, (c) a 95% percentile bootstrap confidence interval approach, and (d) a fully Bayesian approach with diffuse priors. The results showed that approach (b) led to inflated Type 1 error rates and 95% confidence intervals with inadequate coverage rates, whereas other approaches maintained acceptable Type 1 error rates and adequate coverage of confidence intervals. Approach (c) had asymmetric rejection rates at small sample sizes. We used an empirical data set to illustrate these approaches.

AB - In multiple regression researchers often follow up significant tests of the interaction between continuous predictors X and Z with tests of the simple slope of Y on X at different sample-estimated values of the moderator Z (e.g., ±1 SD from the mean of Z). We show analytically that when X and Z are randomly sampled from the population, the variance expression of the simple slope at sample-estimated values of Z differs from the traditional variance expression obtained when the values of X and Z are fixed. A simulation study using randomly sampled predictors compared four approaches: (a) the Aiken and West (1991) test of simple slopes at fixed population values of Z, (b) the Aiken and West test at sample-estimated values of Z, (c) a 95% percentile bootstrap confidence interval approach, and (d) a fully Bayesian approach with diffuse priors. The results showed that approach (b) led to inflated Type 1 error rates and 95% confidence intervals with inadequate coverage rates, whereas other approaches maintained acceptable Type 1 error rates and adequate coverage of confidence intervals. Approach (c) had asymmetric rejection rates at small sample sizes. We used an empirical data set to illustrate these approaches.

KW - Bayesian

KW - bootstrap

KW - interaction

KW - multiple regression

KW - simple slope

UR - http://www.scopus.com/inward/record.url?scp=85018376286&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85018376286&partnerID=8YFLogxK

U2 - 10.1080/00273171.2017.1309261

DO - 10.1080/00273171.2017.1309261

M3 - Article

SP - 1

EP - 20

JO - Multivariate Behavioral Research

JF - Multivariate Behavioral Research

SN - 0027-3171

ER -