From aer-skills
Adds robustness, heterogeneity, mechanism, and placebo checks to empirical economics manuscripts. Use after identification and before writing the introduction.
How this skill is triggered — by the user, by Claude, or both
Slash command
/aer-skills:aer-robustnessThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A modern AER referee report contains three predictable demands:
A modern AER referee report contains three predictable demands:
This skill anticipates all three so that the referee finds the answer already in the paper. Skipping this step turns a referee report into a 6-month delay.
Every empirical AER paper should report, at minimum:
Report heterogeneity that the theory predicts, not heterogeneity discovered by mining:
Distinguish two purposes:
State both explicitly in the manuscript. Do not let the reader infer.
If the result is contested or counterintuitive, present a specification curve (Simonsohn-Simmons-Nelson 2020) showing the estimate across all reasonable analytic choices. This converts "you chose your specification to get this result" into "the result holds across the entire reasonable choice set."
For any empirical paper, predict and pre-empt:
| Comment | Pre-emption |
|---|---|
| "The result may be driven by [omitted variable]" | Include it as a control; show robustness without it |
| "Standard errors are not clustered correctly" | Report 2-3 clustering schemes; wild bootstrap if needed |
| "Pre-trends look suspect" | Formal joint test + honest DiD bounds |
| "This is a mechanical effect from [other channel]" | Direct placebo or sample restriction excluding that channel |
| "Effect size is implausibly large/small" | Sanity-check against existing magnitudes in the literature |
A null result is publishable at AER if and only if:
Always report 95% CIs alongside p-values. Many AER editors explicitly prefer effect-size reporting over significance stars.
Keep main-text robustness to one table with each row a different specification. Push the deep robustness into the appendix in this order:
When working from the AER-skills repository or plugin bundle, load only the resource needed for the paper's design:
docs/methods-reference.mdtemplates/stata/04_robustness.do, templates/r/04_robustness.R, or templates/python/robustness.pytemplates/stata/05_heterogeneity.do, templates/r/05_heterogeneity.R, or templates/python/heterogeneity.pyThe empirical section is referee-ready only when all are present and theory-motivated (not mined):
ROBUSTNESS COVERAGE: <spec / sample / outcome / cluster / estimator>
HETEROGENEITY: <pre-specified / exploratory>
MECHANISM EVIDENCE: <channel / ruling-out / both>
PLACEBO TESTS: <list>
ANTICIPATED REFEREE COMMENTS PRE-EMPTED: <count>
NEXT SKILL: <aer-paper-body | aer-tables-figures>
npx claudepluginhub brycewang-stanford/aer-skills --plugin aer-skillsHardens empirical economics results against REStud referee scrutiny by adding robustness, placebo, heterogeneity, and mechanism checks. Targets missing or fragile specifications before publication.
Builds robustness suites for AEJ: Applied manuscripts to show headline estimates survive specification, sample, and inference choices.
Organizes robustness checks for JEEA manuscripts around threats a general-interest referee would raise, ensuring headline results are stable to specification, sample, inference, and assumption perturbations.