What attracted me to this paper (69 page PDF) was the assertion in the abstract that "Many interventions that 'work' in small-scale trials often fail at scale, highlighting the centrality of effective scaling for realizing the promise of evidence-based policy." So that's what done here. But also, it's noteworthy that they change the application of the personalized adaptive learning (PAL) software from a take-home implementation to one that happens in-school (though "substituting classroom teaching with computer-aided instruction has typically not been effective"). So they do realize learning gains, but it's interesting that "while absolute treatment effects are similar across students, gains relative to the counterfactual are higher for weaker students." However, "because students were often so far behind, even meaningful increases in learning from this low base are unlikely to be captured by grade-level school exams." Also worth noting is that scaling is less about fidelity to the original implementation and "often requires adapting the intervention design in ways that build on the insights and principles illustrated by initial efficacy trials, but may lead to substantially different implementation protocols."
Today: Total: [Direct link] [Share]
