Moderated Mediation in SPSS with PROCESS Model 7

Moderated mediation tests whether an indirect effect changes depending on a moderator. In SPSS, PROCESS Model 7 handles the most common case: the moderator W acts on the first path, from X to the mediator M. The result you need is the index of moderated mediation. If its bootstrap confidence interval excludes zero, the indirect effect depends on W. Here is how to run it and explain it clearly in a paper.

The logic of Model 7

Suppose perceived value (X) increases purchase intention (Y) through attitude (M), as shown in our Model 4 mediation guide. Now imagine that the effect of perceived value on attitude is weaker for price-sensitive consumers (W). If that first path changes with W, the whole indirect effect changes too. That is moderated mediation, also called a conditional indirect effect.

Choose the model that matches your hypotheses, not the one that gives the best results.

Run Model 7 step by step

1. Go to Analyze > Regression > PROCESS.
2. Y variable: purchase intention (PI). X variable: perceived value (PV). Mediator(s) M: attitude (ATT). Moderator variable W: price sensitivity (PS).
3. Set Model number to 7.
4. Set bootstrap samples to 5000.
5. In Options, choose Only continuous variables that define products for mean centering, pick your conditioning values (16th, 50th, 84th percentiles or -1SD, mean, +1SD) and tick Generate code for visualizing interactions.
6. Click OK.

Read the output in four blocks

1. Outcome: ATT (first stage)

The interaction is significant: the effect of perceived value on attitude weakens as price sensitivity increases.

2. Outcome: PI (second stage)

This block gives b (ATT → PI) = 0.48, p < .001, and the direct effect c' (PV → PI) = 0.21, p < .001. In Model 7, W does not act here.

3. Conditional indirect effects

The indirect effect is significant at all three levels, but it becomes smaller as price sensitivity increases.

4. Index of moderated mediation

This is the formal test (Hayes, 2015). The index equals a3 × b (here -0.12 × 0.48 ≈ -0.06). Its interval excludes zero, so moderated mediation is supported. For each one-point increase in price sensitivity, the indirect effect decreases by about 0.06.

Common mistakes to avoid

• Concluding from the conditional effects alone. Two indirect effects can both be significant and still not differ significantly. Only the index tests the difference.
• Relying on a significant a3 only. A significant interaction on the first path is necessary but not sufficient. Report the index.
• Choosing the model after seeing the data. The location of W must come from your hypotheses.

How to report moderated mediation

A moderated mediation model was tested with PROCESS Model 7 (Hayes, 2022) and 5,000 bootstrap samples. Price sensitivity moderated the effect of perceived value on attitude (a3 = -0.12, p = .003). The index of moderated mediation was significant (index = -0.06, 95% CI [-0.10, -0.02]), indicating that the indirect effect of perceived value on purchase intention through attitude weakened as price sensitivity increased (low: 0.31, 95% CI [0.21, 0.41]; high: 0.19, 95% CI [0.11, 0.28]).

Add a conceptual model figure and, if possible, a plot of the first-stage interaction.

FAQ

What if the index is not significant but one conditional effect is?

Then you cannot claim moderated mediation. You can report simple mediation (Model 4) and note that the indirect effect did not vary with W.

Can I include covariates?

Yes. Add them in the Covariate(s) box. By default PROCESS includes them in both the mediator and outcome equations.

Can W be categorical?

Yes. A 0/1 variable works directly; the conditional indirect effects are then given for each group.

Is Model 7 the same as a multigroup analysis?

The idea is close when W is a group variable, but PROCESS tests the difference directly through the index, which is simpler and more rigorous.

References

• Hayes, A. F. (2015). An index and test of linear moderated mediation. Multivariate Behavioral Research, 50(1), 1–22. https://doi.org/10.1080/00273171.2014.962683
• Hayes, A. F. (2022). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (3rd ed.). Guilford Press.
• 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. https://doi.org/10.1080/00273170701341316