A statistical method for synthesizing mediation analyses using the product of coefficient approach across multiple trials

Shi Huang, David Mackinnon, Tatiana Perrino, Carlos Gallo, Gracelyn Cruden, C. Hendricks Brown

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Mediation analysis often requires larger sample sizes than main effect analysis to achieve the same statistical power. Combining results across similar trials may be the only practical option for increasing statistical power for mediation analysis in some situations. In this paper, we propose a method to estimate: (1) marginal means for mediation path a, the relation of the independent variable to the mediator; (2) marginal means for path b, the relation of the mediator to the outcome, across multiple trials; and (3) the between-trial level variance–covariance matrix based on a bivariate normal distribution. We present the statistical theory and an R computer program to combine regression coefficients from multiple trials to estimate a combined mediated effect and confidence interval under a random effects model. Values of coefficients a and b, along with their standard errors from each trial are the input for the method. This marginal likelihood based approach with Monte Carlo confidence intervals provides more accurate inference than the standard meta-analytic approach. We discuss computational issues, apply the method to two real-data examples and make recommendations for the use of the method in different settings.

Original languageEnglish (US)
Pages (from-to)1-15
Number of pages15
JournalStatistical Methods and Applications
DOIs
StateAccepted/In press - Feb 25 2016

Fingerprint

Mediation
Statistical method
Coefficient
Statistical Power
Mediator
Confidence interval
Multiple Outcomes
Bivariate Normal Distribution
Marginal Likelihood
Variance-covariance Matrix
Path
Random Effects Model
Main Effect
Regression Coefficient
Standard error
Estimate
Recommendations
Sample Size
Coefficients
Statistical methods

Keywords

  • Data synthesis
  • Mediation
  • Multiple trials
  • R
  • Restricted maximum likelihood

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

A statistical method for synthesizing mediation analyses using the product of coefficient approach across multiple trials. / Huang, Shi; Mackinnon, David; Perrino, Tatiana; Gallo, Carlos; Cruden, Gracelyn; Brown, C. Hendricks.

In: Statistical Methods and Applications, 25.02.2016, p. 1-15.

Research output: Contribution to journalArticle

@article{87c06f5ca3e54adfb4260b7eb8bca16c,
title = "A statistical method for synthesizing mediation analyses using the product of coefficient approach across multiple trials",
abstract = "Mediation analysis often requires larger sample sizes than main effect analysis to achieve the same statistical power. Combining results across similar trials may be the only practical option for increasing statistical power for mediation analysis in some situations. In this paper, we propose a method to estimate: (1) marginal means for mediation path a, the relation of the independent variable to the mediator; (2) marginal means for path b, the relation of the mediator to the outcome, across multiple trials; and (3) the between-trial level variance–covariance matrix based on a bivariate normal distribution. We present the statistical theory and an R computer program to combine regression coefficients from multiple trials to estimate a combined mediated effect and confidence interval under a random effects model. Values of coefficients a and b, along with their standard errors from each trial are the input for the method. This marginal likelihood based approach with Monte Carlo confidence intervals provides more accurate inference than the standard meta-analytic approach. We discuss computational issues, apply the method to two real-data examples and make recommendations for the use of the method in different settings.",
keywords = "Data synthesis, Mediation, Multiple trials, R, Restricted maximum likelihood",
author = "Shi Huang and David Mackinnon and Tatiana Perrino and Carlos Gallo and Gracelyn Cruden and Brown, {C. Hendricks}",
year = "2016",
month = "2",
day = "25",
doi = "10.1007/s10260-016-0354-y",
language = "English (US)",
pages = "1--15",
journal = "Statistical Methods and Applications",
issn = "1618-2510",
publisher = "Physica-Verlag",

}

TY - JOUR

T1 - A statistical method for synthesizing mediation analyses using the product of coefficient approach across multiple trials

AU - Huang, Shi

AU - Mackinnon, David

AU - Perrino, Tatiana

AU - Gallo, Carlos

AU - Cruden, Gracelyn

AU - Brown, C. Hendricks

PY - 2016/2/25

Y1 - 2016/2/25

N2 - Mediation analysis often requires larger sample sizes than main effect analysis to achieve the same statistical power. Combining results across similar trials may be the only practical option for increasing statistical power for mediation analysis in some situations. In this paper, we propose a method to estimate: (1) marginal means for mediation path a, the relation of the independent variable to the mediator; (2) marginal means for path b, the relation of the mediator to the outcome, across multiple trials; and (3) the between-trial level variance–covariance matrix based on a bivariate normal distribution. We present the statistical theory and an R computer program to combine regression coefficients from multiple trials to estimate a combined mediated effect and confidence interval under a random effects model. Values of coefficients a and b, along with their standard errors from each trial are the input for the method. This marginal likelihood based approach with Monte Carlo confidence intervals provides more accurate inference than the standard meta-analytic approach. We discuss computational issues, apply the method to two real-data examples and make recommendations for the use of the method in different settings.

AB - Mediation analysis often requires larger sample sizes than main effect analysis to achieve the same statistical power. Combining results across similar trials may be the only practical option for increasing statistical power for mediation analysis in some situations. In this paper, we propose a method to estimate: (1) marginal means for mediation path a, the relation of the independent variable to the mediator; (2) marginal means for path b, the relation of the mediator to the outcome, across multiple trials; and (3) the between-trial level variance–covariance matrix based on a bivariate normal distribution. We present the statistical theory and an R computer program to combine regression coefficients from multiple trials to estimate a combined mediated effect and confidence interval under a random effects model. Values of coefficients a and b, along with their standard errors from each trial are the input for the method. This marginal likelihood based approach with Monte Carlo confidence intervals provides more accurate inference than the standard meta-analytic approach. We discuss computational issues, apply the method to two real-data examples and make recommendations for the use of the method in different settings.

KW - Data synthesis

KW - Mediation

KW - Multiple trials

KW - R

KW - Restricted maximum likelihood

UR - http://www.scopus.com/inward/record.url?scp=84975721135&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84975721135&partnerID=8YFLogxK

U2 - 10.1007/s10260-016-0354-y

DO - 10.1007/s10260-016-0354-y

M3 - Article

AN - SCOPUS:84975721135

SP - 1

EP - 15

JO - Statistical Methods and Applications

JF - Statistical Methods and Applications

SN - 1618-2510

ER -