Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach (3rd ed.)

The paper

Hayes (2022). Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach (3rd ed.). Guilford Press. ISBN 9781462549030.

Acclaimed for its thorough presentation of mediation, moderation, and conditional process analysis, this book has been updated to reflect the latest developments in PROCESS for SPSS, SAS, and, new to this edition, R. Using the principles of ordinary least squares regression, Andrew F. Hayes illustrates each step in an analysis using diverse examples from published studies, and displays SPSS, SAS, and R code for each example. Procedures are outlined for estimating and interpreting direct, indirect, and conditional effects; probing and visualizing interactions; testing hypotheses about the moderation of mechanisms; and reporting different types of analyses. Readers gain an understanding of the link between statistics and causality, as well as what the data are telling them. The companion website (www.afhayes.com) provides data for all the examples, plus the free PROCESS download. New to This Edition: Rewritten Appendix A, which provides the only documentation of PROCESS, including a discussion of the syntax structure of PROCESS for R compared to SPSS and SAS. Expanded discussion of effect scaling and the difference between unstandardized, completely standardized, and partially standardized effects. Discussion of the meaning of and how to generate the correlation between mediator residuals in a multiple-mediator model, using a new PROCESS option. Discussion of a method for comparing the strength of two specific indirect effects that are different in sign. Introduction of a bootstrap-based Johnson–Neyman-like approach for probing moderation of mediation in a conditional process model. Discussion of testing for interaction between a causal antecedent variable X and a mediator M in a mediation analysis, and how to test this assumption in a new PROCESS feature.

Contribution

new method

Data basis

none

Document

book

Review stage

peer reviewed

Disciplines

statistics, psychology, marketing

Access

closed

Existing implementations

Code the authors or the community already shipped. This is what decides how far up the ladder a work can climb — no implementation, no tools.

• otherPROCESS macroofficial
• pypiPyProcessMacroMITby Quentin Andréruns here
• githubQuentinAndre/pyprocessmacroMITby Quentin André

What it can do

• Fit processfit_process

  Fit one PROCESS model: every coefficient, direct, indirect and conditional effect.

• Spotlightspotlight

  The effect of X at chosen values of its moderators, with intervals.

• Floodlightfloodlight

  Johnson-Neyman regions: where on the moderator's observed range the effect is significant.

• Versionsversions

  Report the Python and package versions, their licences, the limits and the caveats.

Provenance

Each check stands on its own. There is no single “verified” state: whether an agent faithfully implements its paper is a curation judgement, not something this registry can mechanically prove.

Machine-readable: paper.jsonld · server.json · agent-card.json

Agent: org.agenticjournal/paper-hayes-2022-conditional-process · v0.2.0