Moderation Analysis

Moderation analysis tests whether the strength or direction of the relationship between a predictor and an outcome depends on a third variable, the moderator, usually through an interaction term in regression.

조절효과분석은 예측변수와 결과변수 사이 관계의 강도나 방향이 제3의 변수인 조절변수에 따라 달라지는지를 주로 회귀모형의 상호작용항으로 검정하는 방법입니다.

How do I test whether the effect of one variable on an outcome depends on another variable?

한 변수가 결과에 미치는 효과가 다른 변수에 따라 달라지는지 어떻게 검정하나요?

Moderation analysis adds a predictor-by-moderator interaction term to a regression model to test whether the predictor's effect on the outcome changes across levels of the moderator. A significant interaction is then probed with simple slopes or the Johnson-Neyman technique to show where the effect is present.

When to use Moderation Analysis

• Testing whether a treatment effect differs by gender, age, or another characteristic
• A continuous outcome with a focal predictor and a continuous or categorical moderator
• Examining boundary conditions of a theoretical relationship

When to use something else

• The third variable explains how the effect occurs: use mediation analysis
• Comparing latent paths across groups: use multi-group SEM

Assumptions and requirements

• Standard regression assumptions for the outcome model
• Moderator not caused by the focal predictor
• Adequate sample size, because interaction effects need more power
• Continuous variables optionally mean-centered to aid interpretation

What the analysis reports

• Interaction coefficient with standard error and p-value
• Change in R² from adding the interaction
• Simple slopes at chosen moderator values
• Johnson-Neyman significance region
• Interaction plot

How to report it

The interaction between [predictor] and [moderator] was significant, b = [value], SE = [value], p = [value], ΔR² = [value], indicating that the effect of [predictor] on [outcome] depended on [moderator].

Related methods

• Mediation Analysis
• Multiple Linear Regression
• Moderated Mediation

Run Moderation Analysis on Quorant