Same anchor dataset as Module 01, one running question: tested patients come back within 30 days less often (9.85% vs 11.42%) — should the hospital push A1C testing on every admission? Every arc adds the next tool needed to answer that honestly: the central limit theorem gives a shape to the sampling wobble you already measured, confidence intervals turn one sample into an honest range, a hand-built t-test says whether a difference is real, correlation-vs-causation separates "moves together" from "causes," and the A/B-testing arc ties it all into one workflow — ending with a discovery the module's own research made: the headline gap is visit-level, not patient-level, and shrinks once that's fixed.
Builds forward from Module 01 — tidy data, descriptive statistics, basic visualization, and basic probability are assumed, not re-taught.
This module's Welch-t cross-check cells and the project starter use scipy, which wasn't in the
toolchain before this module. Re-run env/setup.sh once before starting — it installs the
newly pinned dependency and re-runs the smoke test.