synthetic-controllisted

# Synthetic control An opinionated synthetic-control workflow grounded in a read canon (references/canon.md, current as of 2026-07-28): Abadie's JEL survey, the practice manual by the method's originator, with the synthetic-DiD bridge and the DiD-vs-SC boundary supplied by the Arkhangelsky-Imbens panel survey (shared with the did skill). Deliverable: the recommendation with its citation, the R estimation and diagnostics code, and a methods paragraph. Refresh path: run litreview on the method since the canon date, then propose additions to references/canon.md as flagged addenda. ## Which precedent your design looks like | Design shape | Canonical case | Marketing analogue | What kills it | |---|---|---|---| | One treated region, outcome co-moves across regions | Basque terrorism (Abadie and Gardeazabal 2003) | a single-market launch | donors whose outcome does not co-move with the treated region | | One treated unit, a regulation, other units running their own versions of it | California Prop 99 (Abadie, Diamond, and Hainmueller 2010) | a regulation hitting one state's category sales | already-treated donors left in the pool | | One treated unit, pool restricted to genuine peers | German reunification (Abadie, Diamond, and Hainmueller 2015) | a platform policy change in one country | a pool wide enough that fit gets bought from dissimilar donors | | Mid-pack treated unit, smooth series, long pre-period | Texas prison construction (the Mixtape's data exercise) | a geo rollo