The first stage | Is there anything to recover? | One partition to report | Inference that admits the partition is uncertain | Which groupings are firm? | Checking the sampler | How much does the prior matter? | Watching the individual effects, not just the aggregate | The event-study reading | References

The design | The two dials | Precision under partial homogeneity | The separation boundary | Where the two estimators genuinely differ | Reproducing a published row | References