The problem | A simulated panel | Comparing estimators | One estimator at a time | Per-period effects (event study) | Is the effect distinguishable from zero? A placebo test | Comparing standard-error methods | Recovering special cases | Penalty ablation | Adding covariates | A common counterfactual across methods | Bringing your own results | Designs and inference | Performance | Reference

Event study vs. effect matrix | A toy non-absorbing panel | One-step fit: nabs_effect_cells() | Comparing methods | Working from existing objects | Collapsing back to an event study | The cell schema | Why not PanelMatch?

1. Why use R for this at all? | 2. Getting your data in: nabs_read_dta() | 3. Rosetta stone: did_multiplegt_dyn → nabs_event_study() | What about csdid / did_imputation / xtevent? | 4. Stata-style argument aliases | 5. Getting results out: nabs_write_dta() | Reproducibility checklist

The 30-second version | A simulated panel | Three estimators, three lines of code | Or call the estimators directly | Naive TWFE reference | Plot | Plot styles | Schema