Conditional Process Analysis (Moderated Mediation)
Also known as: moderated mediation, moderated mediation analysis, PROCESS model, Hayes PROCESS conditional process model, Koşullu Süreç Analizi (Moderated Mediation)
Conditional process analysis is Andrew F. Hayes's regression-based PROCESS framework (2018) that combines mediation and moderation in a single model, testing how an indirect effect changes across levels of a moderator. It quantifies conditional indirect and conditional direct effects and tests them with bootstrap confidence intervals.
Key highlights
- Combines mediation and moderation in one coherent regression framework, testing conditional mechanisms directly.
- Bootstrap confidence intervals make no normality assumption about the indirect effect, which is typically skewed.
- The index of moderated mediation gives a single, interpretable test of whether an indirect effect truly depends on the moderator.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use conditional process analysis when your theory says a mechanism (the indirect X→M→Y path) is itself conditional — stronger or weaker depending on a moderator — and you have at least about 100 observations to support bootstrap inference. It suits cross-sectional or longitudinal data with continuous, binary, or ordinal variables, and requires that the mediator and outcome equations be correctly specified. It is less appropriate for very small samples, where the bootstrap interval is biased, or when substantial measurement error calls for a latent-variable SEM instead.
Strengths & limitations
- Combines mediation and moderation in one coherent regression framework, testing conditional mechanisms directly.
- Bootstrap confidence intervals make no normality assumption about the indirect effect, which is typically skewed.
- The index of moderated mediation gives a single, interpretable test of whether an indirect effect truly depends on the moderator.
- Bootstrap confidence intervals are biased in small samples (n < 100), so the conditional indirect effect cannot be reliably estimated.
- Results are only as good as the specification: the mediator and outcome equations must be correctly modelled.
- Measurement error in observed mediators can bias paths and may force a move to a latent-variable SEM.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
What is the index of moderated mediation?
It is the product a₃·b, a single quantity that captures how much the indirect effect of X on Y through M changes for a one-unit change in the moderator W. If its bootstrap confidence interval excludes zero, the mediation is genuinely moderated.
Why use bootstrapping instead of a normal-theory test?
The indirect effect is a product of coefficients and is typically not normally distributed, so classical standard errors are inaccurate. Bootstrapping resamples the data to build an empirical confidence interval; Hayes recommends at least 5000 resamples.
How large a sample do I need?
At least about 100 observations. With smaller samples the bootstrap confidence interval becomes biased and the conditional indirect effect cannot be estimated reliably; a causal mediation approach may be preferable.
What if my variables have measurement error?
Observed-variable path models assume the mediator and outcome are measured without error. When measurement error is substantial, moving to a latent-variable structural equation model (including a Bayesian SEM) better protects the path estimates.
Sources
- 1.Hayes, A. F. (2018). Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach (2nd ed.). The Guilford Press.ISBN 978-1462534654
- 2.Preacher, K. J., Rucker, D. D., & Hayes, A. F. (2007). Addressing Moderated Mediation Hypotheses: Theory, Methods, and Prescriptions. Multivariate Behavioral Research, 42(1), 185-227.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 1). Conditional Process Analysis. ScholarGate. https://scholargate.app/causal-inference/conditional-process-analysis