Scientific Manuscript Review

Scientific Manuscript Review

Table of Contents

• Core Principles
• Workflow
• Section-by-Section Review
• Language Guidelines
• Guardrails
• Quick Reference

Related skills: Grant proposals → grant-proposal-assistant | Recommendation letters → academic-letter-architect | Emails → scientific-email-polishing

Core Principles

Seven foundational beliefs guiding manuscript review:

1. Clarity over cleverness: Scientific clarity is more important than stylistic elegance
2. Narrative shapes comprehension: Structure and story arc determine reader understanding
3. Audience dictates tone: Expert vs. general audience requires different depth and framing
4. Format signals credibility: Professional formatting reflects scientific rigor
5. Claims require evidence: Strong assertions need strong data and appropriate hedging
6. Each section has a job: Introduction sells the problem, Results show the data, Discussion interprets
7. Constraints shape structure: Word limits and journal guidelines determine emphasis

Workflow

Copy this checklist and track your progress:

Manuscript Review Progress:
- [ ] Step 1: Identify manuscript type and extract core message
- [ ] Step 2: Structural pass - map and evaluate overall organization
- [ ] Step 3: Introduction review - gap statement, focus, hypothesis
- [ ] Step 4: Results review - question, approach, finding, interpretation
- [ ] Step 5: Discussion review - synthesis, context, limitations
- [ ] Step 6: Scientific clarity check - claims, controls, hedging
- [ ] Step 7: Language polish - terminology, voice, jargon
- [ ] Step 8: Formatting check - journal compliance

Step 1: Identify Manuscript Type and Core Message

Determine document type (research article, review, perspective, short communication). Extract the ONE finding or message readers must remember. Ask: "If readers remember only one thing, what should it be?" See resources/methodology.md for extraction techniques.

Step 2: Structural Pass

Map overall organization against standard IMRaD (Introduction, Methods, Results, Discussion) or review structure. Check logical sequencing - does each section flow into the next? Identify unclear transitions or missing context. See resources/methodology.md for structure evaluation.

Step 3: Introduction Review

Evaluate using the Introduction Arc: Broad context → Narrow focus → Knowledge gap → Hypothesis/Objective. Check that gap statement is explicit and compelling. Verify ending with clear hypothesis or objective. See resources/template.md for template.

Step 4: Results Review

For each figure/table/experiment: Question addressed? → Approach used? → Key finding (with statistics)? → Interpretation (what it means)? Flag data-dump writing that lacks interpretation. Ensure findings build toward core message. See resources/template.md for results structure.

Step 5: Discussion Review

Verify structure: Revisit hypothesis → Interpret findings in field context → Place in broader literature → Acknowledge limitations → Suggest future directions. Check for overclaiming (speculation presented as fact). Ensure clear separation of data interpretation vs. speculation. See resources/methodology.md for discussion framework.

Step 6: Scientific Clarity Check

Run the clarity checklist: Claims supported by data? Quantitative details present (statistics, n values)? Controls adequately described? Interpretations appropriately hedged? Mechanistic explanations where needed? See resources/template.md for full checklist.

Step 7: Language Polish

Ensure terminology consistency throughout. Remove or define jargon on first use. Prefer active voice when it aids clarity. Standardize abbreviations. Check for hedging language ("suggests" vs "proves"). See resources/methodology.md for specific guidance.

Step 8: Formatting Check

Verify compliance with target journal guidelines (word limits, reference format, figure requirements). Check section headings match journal requirements. Ensure abstract follows structured/unstructured requirement. Validate using resources/evaluators/rubric_scientific_manuscript.json. Minimum standard: Average score ≥ 3.5.

Section-by-Section Review

Introduction Structure

Goal: Convince readers the problem matters and your approach is sound

The Funnel Structure:

[Broad context - establish field importance, 1-2 sentences]
        ↓
[Narrow to specific area - what's been done]
        ↓
[Knowledge gap - what's missing, why it matters]
        ↓
[Your hypothesis/objective - what you will address]

Common problems:

• Gap statement buried or implicit (make it explicit: "However, X remains unknown")
• Too broad opening (readers don't need history of the universe)
• No clear hypothesis at end (readers don't know what to expect)
• Overlong literature review (move details to Discussion)

Results Structure

Goal: Present data clearly with interpretation, not just numbers

Per-paragraph/figure structure:

[Question this experiment addresses]
[Approach/method used]
[Key finding - with quantification]
[Brief interpretation - what this means]

Common problems:

• Data dump (listing results without interpretation)
• Missing statistics (p-values, n values, confidence intervals)
• Vague descriptions ("we found differences" vs "we found 3-fold increase")
• Figures not referenced in logical order
• Key findings buried in text (highlight important results)

Discussion Structure

Goal: Interpret findings and place in broader context

Standard flow:

[Restate main finding and hypothesis status]
        ↓
[Interpret key results in field context]
        ↓
[Compare to prior literature - agreements/disagreements]
        ↓
[Mechanistic implications (if applicable)]
        ↓
[Limitations - honest acknowledgment]
        ↓
[Future directions - what comes next]
        ↓
[Concluding statement - big picture significance]

Common problems:

• Overclaiming (data doesn't support conclusions)
• Repeating Results section (discuss, don't recapitulate)
• Missing limitations (reviewers will note them anyway)
• Speculation unmarked (clearly label "we speculate that...")
• No connection to field (discuss in isolation)

Language Guidelines

Active vs. Passive Voice:

• Use active for clarity: "We measured" not "Measurements were made"
• Use passive when agent is obvious or unimportant: "Samples were incubated at 37°C"
• Avoid dangling modifiers: Not "Having analyzed the data, the conclusion was..." but "Having analyzed the data, we concluded..."

Hedging Language:

• Strong data: "demonstrates", "shows", "establishes"
• Moderate confidence: "suggests", "indicates", "supports"
• Speculation: "may", "might", "could potentially"
• Match hedge strength to evidence strength

Jargon Management:

• Define on first use: "polymerase chain reaction (PCR)"
• Avoid unnecessary jargon when plain language works
• Field-standard terms don't need definition (DNA, protein, cell)
• Reader-appropriate: more definition for broad audience journals

Terminology Consistency:

• Pick one term and stick with it (don't alternate between "subjects", "participants", "patients")
• Create terminology table for complex manuscripts
• Check abbreviations defined before use

Guardrails

Key requirements: