Concept Architecture
Concept
Theoretically, Moderation Analysis is a statistical method used to determine whether the strength or direction of the relationship between an explanatory variable and an outcome varies according to the level of a third variable, known as the moderator. The moderator does not explain the causal pathway between variables but instead alters the magnitude or form of the association. Moderation analysis is founded on interaction modelling within regression analysis and is widely used to investigate treatment-effect heterogeneity and conditional relationships.
Mathematically, moderation is represented by including an interaction term between the explanatory variable and the moderator within a regression model. The coefficient of the interaction term quantifies the extent to which the effect of the explanatory variable changes as the moderator varies. Statistical inference is based on hypothesis testing of the interaction coefficient and estimation of conditional effects at different levels of the moderator.
In practice, moderation analysis is implemented using linear regression, logistic regression, mixed-effects models or structural equation models containing interaction terms. In health economics, it is used to investigate whether treatment effectiveness differs by age, disease severity, socioeconomic status, baseline risk or healthcare setting. The results support subgroup analyses, personalised healthcare strategies and evaluation of treatment-effect heterogeneity within economic evaluations.
Purpose
Used to evaluate whether the relationship between an exposure and an outcome depends on a third variable, identify treatment-effect heterogeneity and quantify interaction effects in health economic research.
Mathematical Formulae
Primary Formula
Moderation model:
Y = ?? + ??X + ??M + ??(X ? M) + �
where:
- Y = outcome
- X = explanatory variable
- M = moderator
- X ? M = interaction term
- ?? = moderation effect
- � = random error
Supporting Formulae
Conditional effect of X:
?Y / ?X = ?? + ??M
Hypothesis test:
H?: ?? = 0
Related Mathematical Methods
- Interaction Analysis
- Linear Regression
- Logistic Regression
- Mixed Effects Models
- Structural Equation Modelling
- Generalised Linear Models
- Subgroup Analysis
Example
A health economist evaluates whether the effect of a smoking cessation programme on annual healthcare costs differs according to patient age.
The fitted regression model is:
Cost = ?? ? 600(Treatment) + 15(Age) + 8(Treatment ? Age)
The interaction coefficient is:
?? = 8
This indicates that the treatment effect changes by �8 for each additional year of age.
For a patient aged 40 years:
Treatment effect = ?600 + (8 ? 40)
Treatment effect = ?�280
For a patient aged 70 years:
Treatment effect = ?600 + (8 ? 70)
Treatment effect = ?�40
The results suggest that the programme generates larger cost savings among younger patients than older patients.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| PRODUCT | =B2*C2 | Create the interaction term between the explanatory variable and moderator. |
| LINEST | =LINEST(D2:D501,B2:E501,TRUE,TRUE) | Estimate the moderation regression model. |
| IF | =IF(F2<0.05,"Significant Interaction","No Interaction") | Interpret statistical significance of the interaction effect. |
| SUMPRODUCT | =SUMPRODUCT(B2:E2,$H$2:$H$5) | Calculate predicted outcomes including the interaction term. |
| Solver | Estimate complex moderation models by optimising regression parameters. | Support interaction modelling in advanced analyses. |
VBA (Optional)
A VBA routine can automatically generate interaction variables, estimate moderation models across multiple candidate moderators and summarise conditional treatment effects.
Sources
- Aiken LS, West SG. Multiple Regression: Testing and Interpreting Interactions. Sage Publications.
- Hayes AF. Introduction to Mediation, Moderation, and Conditional Process Analysis. Guilford Press.
- Cohen J, Cohen P, West SG, Aiken LS. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Routledge.
- Dawson JF. Moderation in Management Research: What, Why, When and How. Journal of Business and Psychology. 2014.
- Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation. Oxford University Press.
- ISPOR Good Practice Reports.
Related Concepts (2)
Library
Publications
1
Statistical Analysis of Cost-Effectiveness Data — Willan & Briggs, 1st Edition ed., 2006 (John Wiley & Sons)
A synthesis of statistical methods for analysing cost-effectiveness data, including net-benefit regression, confidence intervals for the ICER, cost-effectiveness acceptability curves, and covariate adjustment. Part of the Wiley Statistics in Practice series.
BookView source →
Frequently Asked Questions (6)
What is moderation analysis?
A statistical approach examining whether an exposure-outcome relationship's strength or direction differs depending on the level of a third, moderating variable.
Source: Baron & Kenny 1986
What does moderation analysis reveal about when an effect holds?
Moderation analysis reveals whether the strength or direction of an exposure's effect on an outcome depends on a third variable, the moderator. It shows that an effect is not uniform, being larger in some groups than others, as when a drug helps younger patients more than older ones. Unlike mediation, which explains how an effect comes about, moderation identifies for whom or under what conditions it holds. Finding what an effect depends on is its purpose. VanderWeele (2015) discusses this idea.
Source: VanderWeele 2015
How does moderation analysis work?
Moderation analysis works by including an interaction term between the exposure and the moderator in a model, so that the coefficient on the interaction indicates whether the exposure's effect on the outcome changes across levels of the moderator. A significant interaction signals moderation. So moderation analysis works by testing the interaction between the exposure and the proposed moderator, which reveals whether the effect of the exposure differs depending on the moderator, and the results can be explored by examining the exposure's effect at different moderator values, showing how the relationship varies and thereby characterising the moderation.
Source: Baron & Kenny 1986
How does moderation differ from mediation?
Moderation concerns whether the strength or direction of an exposure's effect on an outcome depends on a third variable, the moderator, while mediation concerns whether the exposure's effect operates through an intermediate variable, the mediator. Moderation is about when or for whom an effect holds; mediation is about how or why it occurs. So moderation and mediation address different questions, with moderation examining variation in an effect across levels of another variable and mediation examining the pathway through which the effect is transmitted, and they are distinct though sometimes combined, since an effect may both work through a mediator and vary by a moderator.
Source: Baron & Kenny 1986
What is a moderator?
A moderator is a variable that affects the strength or direction of the relationship between an exposure and an outcome, so that the effect of the exposure differs across the levels of the moderator. For example, a treatment's effect might be larger in one age group than another, with age the moderator. So a moderator is a third variable that changes how an exposure relates to an outcome, identifying conditions or subgroups in which the effect is stronger, weaker, or reversed, and detecting moderators is the aim of moderation analysis, since knowing that an effect depends on a moderator helps target interventions and understand for whom they work best.
Source: Baron & Kenny 1986
Why is moderation analysis useful?
Moderation analysis is useful because effects often vary across individuals or conditions, and identifying moderators shows for whom or under what circumstances an exposure has a stronger, weaker, or different effect, which informs targeting and understanding of interventions. So moderation analysis is useful for uncovering variation in effects, since a single average effect may mask important differences between subgroups, and knowing the moderators helps tailor treatments to those most likely to benefit and clarifies the conditions under which an effect holds, which is valuable in research on heterogeneity of treatment effects and in personalising care.
Source: Baron & Kenny 1986
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 18 Dec 2025
Content version: 1.0.0
Canonical Identity
- Persistent URI
- https://healtheconomics.wiki/concept/moderation-analysis
- Term code
- HE-ES-SA-128
Stable URI · Machine-readable · Resolvable · CC BY 4.0