From jebo-skills
Organizes robustness checks for JEBO-style behavioral economics results, addressing demand effects, multiple comparisons, specification, and tuning for experiments, observational designs, and simulations.
How this skill is triggered — by the user, by Claude, or both
Slash command
/jebo-skills:jebo-robustnessThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
- An experiment has several treatment arms / outcomes and you have not corrected for multiplicity
At JEBO the right question is never "did we run enough checks?" but "for each way this could not be the behavioral mechanism, did we show it survives?" Build a threat → check map. The threats differ sharply across the four archetypes.
| Threat | Check JEBO referees expect |
|---|---|
| Experimenter demand | demand-treatment bounds (de Quidt-style), neutral framing, obfuscated objective |
| Multiple comparisons | pre-registered primary outcome; MHT correction (Romano–Wolf, List–Shaikh–Xu, BH) across arms/outcomes |
| Comprehension / confusion | results hold among subjects passing comprehension checks |
| Order / sequence effects | randomize order; show within-order stability |
| Subject pool / platform | replicate across pools (student vs. Prolific vs. field); attention screens on online samples |
| Bots / inattentive online subjects | attention checks, completion-time filters, duplicate-IP screening |
| Selection / attrition | balance among completers; Lee bounds if differential |
JEBO's distinctive robustness demand is mechanism robustness: even a real, replicable effect can be driven by a different behavioral channel than claimed. Where possible, add a check that separates your mechanism from the leading alternative (a moderation test the rival channel does not predict, a mediation analysis with the caveats stated, or a treatment that shuts the rival channel off).
【Archetype】experiment / observational / simulation
【Threat → check map】
- <threat 1> → <check>
- <threat 2> → <check>
【Multiplicity】primary outcome pre-registered? MHT method:
【Mechanism vs. alternative】<test separating claimed channel from rival>
【Headline stability】<range of estimate across specs + uncertainty>
【Next step】jebo-tables-figures
npx claudepluginhub brycewang-stanford/awesome-journal-skills --plugin jebo-skillsChecks robustness of risk/uncertainty parameters from Journal of Risk and Uncertainty results, organizing sensitivity analysis by threat type (functional form, elicitation device, stakes, multiple comparisons, etc.).
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.
Builds robustness suites for AEJ: Applied manuscripts to show headline estimates survive specification, sample, and inference choices.