Causal Mediation in Educational Intervention Studies.

Roberts et al. (2018) wrote a how-to guide for researchers. They show step-by-step how to test if a third variable carries the real punch in an education study.

The authors use a self-regulation trial as the worked example. They walk readers through the code, the numbers, and the graphs.

The paper itself has no new data. Instead, it gives you a recipe to find the hidden engine inside your intervention.

Follow the recipe and you can say, for example, that gains in self-monitoring, not teacher attention, drove the math-score jump.

Whaling et al. (2025) sketched a caregiver-coaching Theory of Change packed with possible mediators like parent learning and motivation. Roberts et al. (2018) hands you the wrench to tighten those bolts in a real trial.

Kingston et al. (2010) already ran a mediation test. They showed experiential avoidance explains why trauma links to problem behavior. Their study is a live example of the method Greg et al. teach.

Zeiler (2006) showed that rehearsal works as a middle link in joint-control tasks. Greg et al. widen the lens, letting you test any mediator—response, cognitive, or social.

Stop guessing why your program works. Add a short mediator measure to your next group design—self-monitoring, parent stress, or peer attention. Run the free code Greg et al. share. If the mediator path is weak, you know where to boost training before the next cohort starts Monday.