Tests of the three-path mediated effect

Aaron B. Taylor, David Mackinnon, Jenn-Yun Tein

Research output: Contribution to journalArticle

498 Citations (Scopus)

Abstract

In a three-path mediational model, two mediators intervene in a series between an independent and a dependent variable. Methods of testing for mediation in such a model are generalized from the more often used single-mediator model. Six such methods are introduced and compared in a Monte Carlo study in terms of their Type I error, power, and coverage. Based on its results, the joint significance test is preferred when only a hypothesis test is of interest. The percentile bootstrap and bias-corrected bootstrap are preferred when a confidence interval on the mediated effect is desired, with the latter having more power but also slightly inflated Type I error in some conditions.

Original languageEnglish (US)
Pages (from-to)241-269
Number of pages29
JournalOrganizational Research Methods
Volume11
Issue number2
DOIs
StatePublished - Apr 2008

Fingerprint

Testing
Type I error
Bootstrap
Mediator
Significance test
Confidence interval
Path model
Monte Carlo study
Mediation
Hypothesis test

Keywords

  • Bootstrapping
  • Mediation

ASJC Scopus subject areas

  • Management of Technology and Innovation
  • Strategy and Management
  • Decision Sciences(all)

Cite this

Tests of the three-path mediated effect. / Taylor, Aaron B.; Mackinnon, David; Tein, Jenn-Yun.

In: Organizational Research Methods, Vol. 11, No. 2, 04.2008, p. 241-269.

Research output: Contribution to journalArticle

@article{901e17a2be1f453bbdd369823416e3f7,
title = "Tests of the three-path mediated effect",
abstract = "In a three-path mediational model, two mediators intervene in a series between an independent and a dependent variable. Methods of testing for mediation in such a model are generalized from the more often used single-mediator model. Six such methods are introduced and compared in a Monte Carlo study in terms of their Type I error, power, and coverage. Based on its results, the joint significance test is preferred when only a hypothesis test is of interest. The percentile bootstrap and bias-corrected bootstrap are preferred when a confidence interval on the mediated effect is desired, with the latter having more power but also slightly inflated Type I error in some conditions.",
keywords = "Bootstrapping, Mediation",
author = "Taylor, {Aaron B.} and David Mackinnon and Jenn-Yun Tein",
year = "2008",
month = "4",
doi = "10.1177/1094428107300344",
language = "English (US)",
volume = "11",
pages = "241--269",
journal = "Organizational Research Methods",
issn = "1094-4281",
publisher = "SAGE Publications Inc.",
number = "2",

}

TY - JOUR

T1 - Tests of the three-path mediated effect

AU - Taylor, Aaron B.

AU - Mackinnon, David

AU - Tein, Jenn-Yun

PY - 2008/4

Y1 - 2008/4

N2 - In a three-path mediational model, two mediators intervene in a series between an independent and a dependent variable. Methods of testing for mediation in such a model are generalized from the more often used single-mediator model. Six such methods are introduced and compared in a Monte Carlo study in terms of their Type I error, power, and coverage. Based on its results, the joint significance test is preferred when only a hypothesis test is of interest. The percentile bootstrap and bias-corrected bootstrap are preferred when a confidence interval on the mediated effect is desired, with the latter having more power but also slightly inflated Type I error in some conditions.

AB - In a three-path mediational model, two mediators intervene in a series between an independent and a dependent variable. Methods of testing for mediation in such a model are generalized from the more often used single-mediator model. Six such methods are introduced and compared in a Monte Carlo study in terms of their Type I error, power, and coverage. Based on its results, the joint significance test is preferred when only a hypothesis test is of interest. The percentile bootstrap and bias-corrected bootstrap are preferred when a confidence interval on the mediated effect is desired, with the latter having more power but also slightly inflated Type I error in some conditions.

KW - Bootstrapping

KW - Mediation

UR - http://www.scopus.com/inward/record.url?scp=40249111525&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=40249111525&partnerID=8YFLogxK

U2 - 10.1177/1094428107300344

DO - 10.1177/1094428107300344

M3 - Article

AN - SCOPUS:40249111525

VL - 11

SP - 241

EP - 269

JO - Organizational Research Methods

JF - Organizational Research Methods

SN - 1094-4281

IS - 2

ER -