Elsevier

Behaviour Research and Therapy

Volume 98, November 2017, Pages 39-57
Behaviour Research and Therapy

Regression-based statistical mediation and moderation analysis in clinical research: Observations, recommendations, and implementation

https://doi.org/10.1016/j.brat.2016.11.001Get rights and content

Highlights

  • Addresses the practice of mediation and moderation analysis using linear regression in the pages of Behaviour Research and Therapy.

  • Offer some observations and recommendations, debunks some popular myths, and describes some new advances.

  • Provides an example set of analyses using the PROCESS macro for SPSS and SAS.

  • Nudges clinical researchers away from historically significant but increasingly old school approaches toward modifications, revisions, and extensions that characterize more modern thinking about the analysis of the mechanisms and contingencies of effects.

Abstract

There have been numerous treatments in the clinical research literature about various design, analysis, and interpretation considerations when testing hypotheses about mechanisms and contingencies of effects, popularly known as mediation and moderation analysis. In this paper we address the practice of mediation and moderation analysis using linear regression in the pages of Behaviour Research and Therapy and offer some observations and recommendations, debunk some popular myths, describe some new advances, and provide an example of mediation, moderation, and their integration as conditional process analysis using the PROCESS macro for SPSS and SAS. Our goal is to nudge clinical researchers away from historically significant but increasingly old school approaches toward modifications, revisions, and extensions that characterize more modern thinking about the analysis of the mechanisms and contingencies of effects.

Section snippets

Statistical mediation analysis

Mediation analysis is used when a researcher seeks to test hypotheses about or better understand how an effect of X on Y operates. The causal antecedent X could be which of two forms of therapy a client receives, or it could be an individual difference measure such as exposure to various sources of trauma, or any other conceivable variable that has some kind of causal force on a consequent outcome variable. That consequent Y could be something like frequency or severity of symptoms of some

Moderation analysis

Whereas mediation analysis focuses on how a causal effect operates, moderation analysis is used to address, when, or under what circumstances, or for what types of people that effect exists or does not and in what magnitude. Fig. 3, panel A, graphically depicts the concept of moderation. In this figure, the arrow linking W to the effect of X on Y denotes that X's effect on Y depends in some way on W. More specifically, X's effect on Y is said to be moderated by W if the size or sign of X's

Integrating moderation and mediation: Conditional process analysis

We have seen that effects can operate indirectly through mediators, and that the size of effects can be dependent on other variables. As an indirect effect (mediation) is an effect, and effects can be contingent (moderation), it follows that an indirect effect can be contingent. In other words, the size of an indirect effect can be dependent on another variable—moderated mediation. For instance, a therapeutic method might indirectly influence later symptoms by changing people's cognitive

Summary

In this article, using nothing more complicated than the principles of ordinary least squares regression analysis, we have described the analysis of mediation and moderation effects and their integration as conditional process analysis. We also illustrated implementation using PROCESS, a freely available computational tool for SPSS and SAS that takes much of the computational burden off the researcher's shoulders. We discussed some new developments in thinking over the last several years,

References (86)

  • E.R. Lebowitz et al.

    Anxiety sensitivity moderates behavioral avoidance in anxious youth

    Behaviour Research and Therapy

    (2015)
  • C.P. McLean et al.

    Changes in negative cognitions mediate PTSD symptom reductions during client-centered therapy and prolonged exposure for adolescents

    Behaviour Research and Therapy

    (2015)
  • A.J. Morgan et al.

    Behavior change through automated e-mails: Mediation analysis of self-help strategy use for depressive symptoms

    Behaviour Research and Therapy

    (2013)
  • A. Pictet et al.

    When less could be more: Investigating the effects of a brief internet-based imagery cognitive bias modification intervention in depression

    Behaviour Research and Therapy

    (2016)
  • B. Sandin et al.

    Anxiety sensitivity, catastrophic misinterpretations and panic self-efficacy in the prediction of panic disorder severity: Towards a tripartite cognitive model of panic disorder

    Behaviour Research and Therapy

    (2015)
  • P. Smith et al.

    Computerized CBT for depressed adolescents: Randomized controlled trial

    Behaviour Research and Therapy

    (2015)
  • A.B.P. Staring et al.

    Self-esteem treatment in anxiety: A randomized controlled crossover trial of eye movement desensitization and reprocessing (EMDR) versus competitive memory training (COMET) in patients with anxiety disorders

    Behaviour Research and Therapy

    (2016)
  • H.R. Trompetter et al.

    Psychological flexibility and catastrophizing as associated change mechanisms during online acceptance & commitment therapy for chronic pain

    Behaviour Research and Therapy

    (2015)
  • N. Vincent et al.

    Hyperarousal, sleep scheduling, and time awake in bed as mediators of outcome in computerized cognitive-behavioral therapy (cCBT) for insomnia

    Behavioural Research and Therapy

    (2013)
  • B. Allen et al.

    Synchrony and desynchrony in the tripartite model of fear: Predicting treatment outcome in clinically phobic children

    Behavior Research and Therapy

    (2015)
  • R.M. Baron et al.

    The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations

    Journal of Personality and Social Psychology

    (1986)
  • D.J. Bauer et al.

    Probing interactions in fixed and multilevel regression: Inferential and graphical techniques

    Multivariate Behavioral Research

    (2005)
  • M. Beadman et al.

    A comparison of emotion regulation strategies in response to craving cognitions: Effects on smoking behaviour, craving, and affect in dependent smokers

    Behaviour Research and Therapy

    (2015)
  • K.A. Bollen

    Structural equation modeling with latent variables

    (1989)
  • C.P. Brandt et al.

    The examination of emotion dysregulation as a moderator of depression and HIV-relevant outcome relations among an HIV 1 sample

    Cognitive Behavioural Therapy

    (2015)
  • N.J.K. Breitborde et al.

    Mediators and moderatorsin early intervention research

    Early Interventions in Psychiatry

    (2010)
  • D.T. Campbell et al.

    A primer on regression artifacts

    (1999)
  • E. Cerin et al.

    A commentary on current practice in mediating variable analysis in behavioral nutrition and physical activity

    Public Health Nutrition

    (2009)
  • J. Cohen et al.

    Applied multiple regression/correlation analysis for the behavioral sciences

    (2003)
  • R.B. Darlington et al.

    Regression analysis and linear models: Concepts, applications, and implementation

    (2017)
  • T.A. Dennis-Tiwary et al.

    For whom the bell tolls: Neurocognitive individual differences in the acute stress-reduction effects of an attention bias modification game for anxiety

    Behavior Research and Therapy

    (2016)
  • J.R. Edwards et al.

    Methods for integrating moderation and mediation: A general analytical framework using moderated path analysis

    Psychological Methods

    (2007)
  • R. Emsley et al.

    Mediation and moderation of treatment effects in randomised controlled trials of complex interventions

    Statistical Methods in Medical Research

    (2010)
  • A.J. Fairchild et al.

    A general model for testing mediation and moderation effects

    Prevention Science

    (2009)
  • R.J. Friedrich

    In defense of multiplicative terms in multiple regression equations

    American Journal of Political Science

    (1982)
  • M.S. Fritz et al.

    Explanation of two anomalous results in statistical mediation analysis

    Multivariate Behavioral Research

    (2012)
  • J. Gaume et al.

    Under what conditions? Therapist and client characteristics moderate the role of change talk in brief motivational intervention

    Journal of Consulting and Clinical Psychology

    (2016)
  • A.F. Hayes

    Beyond Baron and Kenny: Statistical mediation analysis in the new millennium

    Communication Monographs

    (2009)
  • A.F. Hayes

    Introduction to mediation, moderation, and conditional process analysis: A regression-based approach

    (2013)
  • A.F. Hayes

    An index and test of linear moderated mediation

    Multivariate Behavioral Research

    (2015)
  • A.F. Hayes

    Partial, Conditional, and Moderated Mediation: Quantification, Inference, and Interpretation

    (2016)
  • A.F. Hayes et al.

    Computational procedures for probing interactions in OLS and logistic regression

    Behavior Research Methods

    (2009)
  • A.F. Hayes et al.

    A tutorial on testing, visualizing, and probing an interaction involving a multicategorical variable in linear regression analysis

    Communication Methods and Measures

    (2016)
  • Cited by (0)

    View full text