choose-statistical-testlisted

# Skill: choose-statistical-test > **Invoked by:** `applied-statistician` (primary). Also consulted by `experiment-analysis` and `statistical-qa-of-metrics` when they need to name the underlying test. > > **When to invoke:** "which test should I use?"; "is the difference between these groups significant?"; any comparison of 2+ groups or relationship between 2 variables. > > **Output:** the recommended test + the assumption checks that gate it + the nonparametric fallback + a short runnable snippet (method-before-library) + the effect-size/CI to report alongside. ## Procedure 1. **Restate the question** in the tree's terms: is it a *group difference*, a *relationship*, or a *prediction*? 2. **Traverse the decision tree** in [`../../knowledge/test-selection-decision-tree.md`](../../knowledge/test-selection-decision-tree.md) against: - outcome **data type** (continuous / categorical / ordinal / count / time-to-event), - **number of groups** (1 / 2 / 3+), - **paired vs independent**, - the **assumption gate** (normality, equal variance, independence). 3. **Name the method first, the library second.** Resolve to the parametric leaf only if the assumption gate passes; otherwise take the nonparametric fallback. 4. **Emit a ≤10-line snippet** (Tier-1 tooling — `scipy.stats` / `pingouin` / `statsmodels`; see [`../../knowledge/statistics-tooling-2026.md`](../../knowledge/statistics-tooling-2026.md)) that (a) checks the assumption, (b) runs the test, (c) reports **effect