---
name: brycewang-stanford/respond-to-referee
source: https://app.decimal.ai/s/brycewang-stanford-respond-to-referee@1/SKILL.md
source_sha256: 51ed9227ea51
---

# Respond to Referee

Structure a point-by-point response to referee reports for the sewage-house-prices manuscript.

**Input:** `$ARGUMENTS` — path to referee report file(s).

---

## Project-Specific Context

### Manuscript
- Main document: `docs/overleaf/_main.tex`
- Sections: `docs/overleaf/*.tex`
- Bibliography: `docs/overleaf/refs.bib`

### Analysis
- Scripts: `scripts/R/09_analysis/`
- Tables: `output/tables/`
- Figures: `output/figures/`

### Common Referee Concerns for This Paper
- Short EDM panel (2021-2024) — limited time variation
- LSOA FE may not capture neighbourhood-level sorting
- Spill count vs hours as treatment — which matters more?
- External validity beyond England
- Dry spill identification — rainfall threshold sensitivity
- Multiple testing across 6 radii and multiple specifications

---

## Workflow

### Step 1: Parse Inputs

1. Read referee report(s) from `$ARGUMENTS`
2. Read the manuscript (`docs/overleaf/_main.tex` and section files)
3. Scan existing analysis scripts to know what already exists

### Step 2: Classify Every Comment

Assign each referee point a classification:

| Class | Action |
|-------|--------|
| **NEW ANALYSIS** | Flag for user — requires new R script or specification |
| **CLARIFICATION** | Draft revised text for the relevant section |
| **REWRITE** | Draft structural revision of a section |
| **DISAGREE** | Draft diplomatic pushback — flag for user review (mandatory) |
| **MINOR** | Draft fix directly (typos, formatting, minor edits) |

### Step 3: Build Tracking Document

Save to `output/log/referee_response_tracker.md`:

```markdown
# Referee Response Tracker
**Date:** YYYY-MM-DD
**Journal:** [if known]
**Decision:** [R&R / Major / Minor]

## Summary
- Referee 1: N comments (X new analysis, Y clarification, Z disagree, W minor)
- Referee 2: N comments (...)
- Editor: N comments (...)
- **Total new analyses required:** X

## Action Items (Priority Order)

### HIGH: New Analysis Required
| # | Ref | Point | Status |
|---|-----|-------|--------|
| 1 | R1.3 | [Brief] | TODO |

### MEDIUM: Clarification / Rewriting
| # | Ref | Point | Status |
|---|-----|-------|--------|
| 1 | R1.1 | [Brief] | TODO |

### FLAGGED: Disagreements (require user review)
| # | Ref | Point | Draft Response |
|---|-----|-------|---------------|
| 1 | R2.5 | [Brief] | [Draft] |

### LOW: Minor Edits
- [ ] R1.7: Fix typo on p. 12
```

### Step 4: Draft Response Letter

Generate LaTeX response letter:

```latex
\documentclass[12pt]{article}
\usepackage[margin=1in]{geometry}
\usepackage{xcolor}
\definecolor{response}{RGB}{0,0,128}

\begin{document}

\title{Response to Referee Reports: Sewage in Our Waters}
\author{[Authors]}
\date{\today}
\maketitle

Dear Editor,

We thank the editor and referees for their careful and constructive comments.
We have revised the manuscript to address all points raised.

\bigskip

\textbf{Summary of major changes:}
\begin{enumerate}
\item [Major change 1 — addresses R1.3, R2.1]
\item [Major change 2 — addresses R1.5]
\end{enumerate}

\newpage
\section*{Response to Referee 1}

\subsection*{Comment 1.1}
\textit{[Exact quote of referee comment]}

\medskip
\textcolor{response}{%
\textbf{Response:} [Draft response]

\textbf{Paper change:} [Section X, page Y]
}

\end{document}
```

Save to `output/log/referee_response_[date].tex`.

### Step 5: Diplomatic Disagreement Protocol

When classification is **DISAGREE**:
1. Open with acknowledgment of the referee's point
2. Provide specific evidence (data, results, published work)
3. Offer partial concession where possible
4. **Never say:** "The referee is wrong" or "misunderstood"
5. **Instead say:** "We respectfully note...", "Upon reflection, we believe..."
6. **FLAG prominently** for user review

---

## Principles

- **The response letter is the authors' voice.** Match their academic tone.
- **Never fabricate results.** Mark NEW ANALYSIS items as TBD until implemented.
- **Flag all DISAGREE items.** These need human judgment — never resolve disagreements autonomously.
- **Prioritize editor's comments.** The editor signals which concerns matter most.
- **Track everything.** Every referee comment appears in both tracker and response letter.