Moderated APIM
Dyad-level moderators of actor and partner effects
The full APIM workflow: a dyad-level moderator (e.g. presence of children, relationship length, time spent together) shapes one or more of the actor and partner effects. The focal question is typically whether the partner crossover effect is moderated — is the partner’s influence on me stronger or weaker in some dyads than in others?
Tutorial 08 — Moderated APIM: Hahn, Binnewies, & Dormann (2014)
Replicates the theoretical model of Hahn, Binnewies, & Dormann (2014, Journal of Vocational Behavior). The presence of children moderates the partner WNC crossover effect. Both MLM and SEM specifications.
MLM + SEM Advanced
When to use this specification
A moderated APIM is appropriate when:
- You have a dyad-level variable (no within-dyad variance) that you hypothesised moderates one or more of the actor or partner effects. Examples: presence of children, relationship length, cultural context, study site.
- Your focal question is whether the partner crossover effect is itself moderated. This is the most common form of the moderated APIM in the literature.
- You want to test the moderation with both MLM (direct interaction terms) and SEM (manual product indicators) so that you can compare the two.
The two ways to fit a moderated APIM
| Approach | Spec | When to use |
|---|---|---|
| MLM | Add predictor:moderator to the lmer formula |
Quick, well-powered, default approach |
| SEM wide | Create a product indicator in the data, add to the lavaan model |
Useful when you want fit indices or simple slopes from the SEM, or for visualisation |
The two approaches should give you nearly identical point estimates for the moderation. The MLM is the workhorse; the SEM is most useful when you already have a wide-format SEM in your pipeline and want to extend it.