Expecon Rebuttal

Rebuttal Strategy (expecon-rebuttal)

When to trigger

• An R&R or rejection arrived and you must decide what to concede, contest, or re-run
• A referee asks for a new treatment or more sessions — the costliest and most common ExpEcon request
• A reviewer raised a deception or power/inference-unit concern that could be decisive
• Two referees disagree, or the two editors' priorities differ, and you need a coherent plan

The ExpEcon revision reality: new data is on the table

Unlike journals where revisions are textual, an ExpEcon R&R often requires collecting new data — an added control treatment, more matching groups for power, or a robustness condition. Plan the response around that fact.

1. Triage every comment into: (a) new data needed (new treatment / more sessions), (b) re-analysis (correct inference unit, MHT correction, new test), (c) clarification (already in the paper, surface it), (d) respectful disagreement (with evidence). Most letters are mostly (b)–(c); isolate the (a) items because they set the timeline and budget.
2. Address the gate concerns first and decisively. A deception worry is existential: either prove the procedure is not deception under the ESA definition, or re-run without it — there is no middle path. A power/unit concern: re-analyze at the session/matching-group level and report MDE; if truly underpowered, run additional sessions rather than argue.
3. For each new-data request, decide and justify. Run it if it sharpens identification or power; if not, explain precisely why the existing design already answers the question (and offer a cheaper analysis that addresses the underlying worry). Never silently ignore a request for a treatment.
4. Re-deposit the package. New sessions mean updated data, instructions, and z-Tree/oTree code in the repository; keep the replication deposit in sync with the revised paper.

Triaging the common ExpEcon requests

Writing the response letter

• Point-by-point, quote-then-respond. Reproduce each comment, then state the change, the location (section/table/figure), and the new result. Editors and referees verify against the manuscript.
• Lead with what you did, not why the referee was wrong. Even when contesting, open with the action taken or the evidence, in a collegial register.
• Report new numbers honestly. If a new treatment weakened the effect, say so and interpret it; an experimentalist audience values the clean result over the convenient one.
• Separate primary from exploratory in everything you add, mirroring the pre-registration logic.
• Mind the two-editor structure. Where referees conflict, address both and let the response help the editors converge; do not pick one referee and ignore the other.

Worked vignette (illustrative)

A referee says a coordination-game result "might be a demand effect — subjects guessed the hypothesis." Three moves: (1) note the abstract/neutral framing already used (no leading language — quote the instructions); (2) report the post-experiment belief question showing subjects did not infer the hypothesis; (3) if still unconvinced, add a treatment with an obfuscated cover task and show the effect persists (illustrative: 0.7 SD, unchanged). The letter leads with these three actions, not with "the referee is mistaken," and updates the instructions and data in the repository to match.

Budgeting the new-data round

An added treatment is not just bench cost — it is a timeline the editors will weigh. Before committing, estimate: sessions needed for the new arm at the group-level power you must hit, lab availability, IRB amendment time, and re-deposit effort. If a requested treatment is genuinely infeasible, say so explicitly and offer the strongest available substitute (a re-analysis, a bound, or a focused robustness condition) rather than a vague promise. Editors respond far better to "we ran 10 of the 12 sessions you asked for; here is why 12 was infeasible and what the 10 show" than to silence or hand-waving.