Trigger
Use when the task involves:
- identifying an estimand or empirical strategy
- deciding between specifications or inference choices
- assessing threats to identification or robustness
- interpreting regression, event-study, IV, RD, or panel results
Workflow
- State the estimand, unit of observation, sample, and treatment timing.
- Spell out the identification strategy and the assumptions required for interpretation.
- Map variables, sample restrictions, specification choices, and inference choices.
- Report the main result, robustness evidence, and limitations separately.
Guardrails
- Do not confuse predictive fit with identification.
- Make sample restrictions, missing-data handling, and treatment coding explicit.
- Use inference choices that match the design, and say why they do.
- Separate the preferred specification from robustness or placebo specifications.