Eursr Rebuttal

R&R Rebuttal (eursr-rebuttal)

An ESR R&R is the normal road to publication for a promising paper — and it usually asks for substantial revision. Reviewers focus on the comparative design, measurement equivalence, and the modeling (level, clustering, few-cluster inference), so the response letter must satisfy a quantitative skeptic without breaking the paper, while the editor adjudicates.

When to trigger

• An R&R decision arrived and you are planning the revision + response letter
• Reviewers disagree (e.g., one wants more countries, another a different estimator)
• A reviewer requests analyses or measurement changes that would change the claims
• Writing the cover note to the editor summarizing the revision

Strategy

1. Read the editor's letter as the rubric. The editor flags which points are decisive — solve those first; the editor adjudicates conflicts among reviewers.
2. One point-by-point response, every comment addressed. Quote each comment, then respond. Never skip one — silence reads as non-compliance.
3. Concede or rebut explicitly, with evidence. For each: did what was asked (say where, with the new text/table number), or push back respectfully with a reason (theory, design, or measurement). A well-argued disagreement beats a capitulation that weakens the paper.
4. Answer modeling and measurement demands on their own terms. A request for invariance tests, df-aware SEs, a leave-one-country-out check, or an alternative harmonization should be run and reported honestly — these are ESR's core review currency.
5. Protect the contribution. Add robustness, invariance, and clarifications; resist changes that dilute the portable mechanism or the comparative leverage that earned the R&R.
6. Keep anonymity intact in the revised manuscript, and update the Data Availability Statement and replication package so any new tables/figures stay reproducible (see eursr-transparency-and-data).

Response-letter format

For each reviewer comment:

> [Quoted reviewer comment]

Response: [What we did / why we respectfully disagree].
Change: [Section/page/table-figure number where the revision appears].

Open with a short summary of the main changes to the editor; group by reviewer; end each entry with the location of the change so the editor can verify quickly.

Triage grid for an ESR R&R

Worked micro-example (illustrative)

A comparative attitudes paper gets an R&R where R1 doubts measurement equivalence and R2 thinks the macro effect rests on too few clusters.

R1: "Are the attitude scales comparable across countries?"
  Response: Added configural/metric/scalar invariance tests (Appendix Table A3); scalar fails for 3
  countries → re-estimated with partial scalar invariance; substantive conclusion unchanged. Change: §4.2.
R2: "24 countries can't support that macro claim."
  Response: Re-ran macro inference with a wild cluster bootstrap and a Bayesian two-level model
  (Table 5); interaction CI widens but excludes zero; macro claim tempered. Change: §5.1, Appendix B.

The letter concedes where evidence warrants, runs the modeling checks on their own terms, and protects the contribution by showing the cross-level result survives the stricter inference.

Referee pushback → ESR-specific fix

• "The revision didn't change anything." → End each entry with an exact location; visible change beats a thank-you.
• "You ignored the invariance issue." → Run the invariance tests, report the level reached, and state what partial invariance licenses.
• "You over-claim from few clusters." → Re-estimate with df-aware or Bayesian methods and temper the macro claim rather than defending the original SEs.

Calibration anchors

• The editor's letter is the rubric. At ESR the editor weights reviewers; solving the decisive points first converts an R&R.
• Run the modeling checks. Invariance, few-cluster inference, and leave-one-country-out are ESR's review currency — running them honestly is more persuasive than arguing them away.
• Substantial means substantial. Plan for a heavy revision; a light pass reads as non-engagement.

Anti-patterns

• Ignoring or merging away a comment without a visible response
• Capitulating to a request that breaks the paper's logic just to please a reviewer
• Dismissing a measurement-equivalence or few-cluster objection instead of addressing it
• "We thank the reviewer" with no actual change or argued reason
• New analyses that quietly contradict the original claim without acknowledgment
• Reintroducing identifying information into the revised (still anonymous) manuscript

Output format

【Editor's decisive points】addressed first? [list]
【Coverage】every reviewer comment answered? [Y/N]
【Concede vs rebut】each tagged with evidence + change location
【Modeling/measurement checks】invariance / few-cluster / LOO run and reported? [Y/N]
【Contribution protected】no dilution of mechanism / comparative leverage? [Y/N]
【Anonymity + DAS/package updated】[Y/N]
【Next】resubmit via ScholarOne

Supplementary resources

• ../../resources/external_tools.md — invariance, few-cluster, and Bayesian multilevel tooling
• ../../resources/official-source-map.md — ESR review model and ethics