Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →brycewang-stanford/aeja-robustness
Use when an American Economic Journal: Applied Economics (AEJ: Applied) manuscript's headline estimate must be shown to survive specification, sample, and inference choices before submission or in an R&R. Builds the robustness suite a sophisticated referee expects; it does not establish the primary identification (aeja-identification) or format the exhibits (aeja-tables-figures).
Passed 21 of 22 test cases with the skill, and 16 of 22 without it.
5 fixed16 passed either way1 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 5 test cases the skill fixed. Same model, same prompt, with and without the skill.
I have completed my primary instrumental variables estimation studying the wage returns of a job training program, with a baseline point estimate of 0.11 and standard error of 0.04 across 14 clusters. A reviewer is concerned about weak instruments, cluster bias, and potential selection on unobservables. Please conduct a comprehensive robustness evaluation formatted to show the primary specification details, a clear mapping of design threats to corresponding checks, inference stress tests including wild-cluster bootstrap, weak-IV sensitivity bounds, the range of point estimates across perturbations, and the suggested next workflow step.
| Checked | Without skill | With skill |
|---|---|---|
| Includes wild-cluster bootstrap to address inference with 14 clusters. | Passed | Passed |
| Designates aeja-tables-figures as the next workflow step. | Missed | Passed |
The instructions your model reads when the skill loads.
AEJ: Applied referees probe whether the headline number is stable, honestly inferred, and not the product of researcher degrees of freedom. Robustness here is not a wall of regressions — it is a targeted set of checks each tied to a specific threat to the design. Map every plausible objection to the one check that answers it, and report the checks so the reader sees the estimate barely moves.
| Threat to the result | The check that answers it |
|---|---|
| Omitted confounders | Oster δ / coefficient-stability bounds; added controls in steps |
| Specification search | a specification curve / multiverse; pre-registered primary spec |
| Functional form | levels vs logs, alternative outcome definitions, nonparametric version |
| Sample selection | drop influential units, alternative inclusion rules, balanced vs unbalanced panel |
| Inference too narrow | clustered SEs at the right level, wild-cluster bootstrap (few clusters), randomization inference |
| Design-specific fragility | DID: honest-DID bounds; RD: bandwidth/donut; IV: weak-IV-robust set |
| Multiple outcomes/subgroups | Romano–Wolf / List–Shaikh–Wooldridge MHT adjustment |
Run the battery, don't just enumerate it. Full map:
execution-with-mcp. AEJ: Applied is applied microeconomics — labor, health, education, and development field settings where a clean research design is the entry ticket.
romano_wolf (step-down FWER, accounts for
cross-test correlation) or benjamini_hochberg — report the adjusted threshold.oster_delta / sensemakr — the confounder strength that would
overturn the headline.test cases passed, vs 16 / 22 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.