Paper Structure

---
name: paper-structure
description: Architecting a scientific paper — IMRaD logic, storyline design, and paragraph-level structure that reviewers follow.
category: scientific
---

## Overview

A paper is a persuasive document with a fixed architecture: it must
establish a gap, fill it credibly, and state what changed. This skill
covers designing the storyline before writing (the "story arc"),
structuring each IMRaD section to do its job, paragraph-level
organization (topic sentences, evidence, interpretation), and the
revision passes that turn a draft into a submission.

## When to use

- Planning a paper: turning results into a coherent storyline
- Structuring Introduction, Methods, Results, Discussion for maximum clarity
- Diagnosing why a draft feels muddled or gets "unclear" reviews
- Outlining before writing to avoid the blank-page stall
- Revising: systematic passes for logic, evidence, and language

## Core concepts

- **The story arc:** status quo → gap/complication → what you did → key findings → what it means. Every section serves this arc; material that doesn't advance it gets cut or moved to supplement.
- **IMRaD as contract:** Introduction (why this matters, what's unknown, what you did — ending with the paper's claims), Methods (enough to reproduce), Results (what you found, no interpretation), Discussion (what it means, limitations, what's next). Reviewers check each contract.
- **Paragraph discipline:** one idea per paragraph; topic sentence first, evidence middle, interpretation/linking sentence last. A paragraph whose first and last sentences don't connect is two paragraphs.
- **Claim–evidence pairing:** every substantive claim in Introduction/Discussion must be traceable to a cited source or your own result — unanchored claims are what reviewers attack.
- **The "need to know" filter:** readers need motivation, method essentials, results, and meaning — in that order. Interesting-but-irrelevant material (however hard-won) goes to the supplement or out.
- **Title and headings as signposts:** headings should telegraph content ("X increases Y under Z", not "Results"); the title names the finding or the question, not the topic area.

- **The "hourglass" shape:** broad context → narrow methods/results → broad implications — the classic IMRaD arc; each section's breadth should match its position in the hourglass.
- **Signposting:** explicit roadmaps ("We first show X, then demonstrate Y, finally establish Z") at section openings — readers navigate long papers via signposts, not memory.
- **The title as a claim:** the strongest titles state the finding, not the topic ("X drives Y via Z" beats "Studies on X") — the title is the paper's narrowest, sharpest summary.

## Practical workflow

### 1. Design the story before drafting

1. Write the 5-sentence story: (1) big-picture context, (2) specific gap, (3) what you did, (4) key result, (5) implication. If you can't, you're not ready to write.
2. List your 3–5 key findings in order; each becomes a Results subsection and ideally a figure.
3. Draft the figures first — the paper's skeleton; write text to serve the figures, not vice versa.

### 2. Draft section by section

1. **Results first:** describe each figure in order — what was done (briefly), what was observed, with numbers. No interpretation beyond what's needed to follow.
2. **Methods:** full reproducibility detail; move routine protocols to supplement but keep the novel/decision-critical parts in main text.
3. **Introduction:** funnel from broad to specific — field context (1–2 paragraphs), the gap with citations (the "however"/"unknown" paragraph), your approach and claims (final paragraph maps to Results order).
4. **Discussion:** restate key findings (briefly), interpret each (mechanism, comparison with literature — agreement and disagreement explained), limitations (stated plainly, with why they don't sink the conclusions), broader implications and next steps.

### 3. Revise in passes

1. **Logic pass:** does every paragraph advance the arc? Cut or relocate the rest. Check claim–evidence pairing throughout.
2. **Reviewer pass:** read as a skeptical reviewer — where would you object? Preempt with data, caveats, or citations.
3. **Language pass:** active voice where the actor matters, precise verbs, no hedging stacks ("might possibly suggest"); consistent terminology (pick one term per concept).
4. **Fresh-eyes pass:** after a break, read only topic sentences — the paper's argument should be fully visible in them.

### 4. Final checks

1. Every figure/table cited in order; every citation in the reference list and vice versa.
2. Numbers consistent between text, figures, tables, and abstract.
3. Authorship, contributions, data/code availability statements complete.

### 5. Structure the Discussion to persuade

1. Open with the 2–3 sentence takeaway — what the results establish, plainly stated before any caveats.
2. Interpret each key result: mechanism, comparison with literature (agreement and disagreement, both explained), and what rules out alternatives.
3. State limitations as boundary conditions ("our conclusions hold for X; beyond that..."), then close with the forward look — what this enables next.

### 6. Quick-reference checklist

- [ ] Hourglass shape maintained (broad → narrow → broad)
- [ ] Title states the finding, not the topic
- [ ] Every Introduction claim is delivered by the Results/Discussion
- [ ] Signposts at each section opening
- [ ] Methods contain everything a replicator needs
- [ ] Discussion argues and interprets, not just summarizes
- [ ] Limitations stated as boundary conditions
- [ ] Target journal's format checked before final layout

## Common pitfalls

- **Writing before the story exists:** drafting as a substitute for thinking — produces a data dump, not a paper.
- **Results–Discussion bleed:** interpreting in Results and re-reporting in Discussion — keep the contract; cross-reference instead.
- **The "and then" structure:** chronological lab-diary ordering instead of logical ordering — organize by finding, not by when you did it.
- **Gap inflation:** overselling novelty ("for the first time ever") — reviewers know the literature; precise positioning beats hype.
- **Limitation amnesia:** omitting weaknesses reviewers will spot — stated limitations disarm; hidden ones detonate.
- **Abstract–paper mismatch:** claims in the abstract that the body doesn't support — the fastest route to rejection.
- **Introduction promises the Discussion doesn't keep:** claims previewed up front that results never deliver — align the two before submission; reviewers check.
- **Methods in the Results:** procedural detail interrupting the findings narrative — move it to Methods, keep only what's needed to interpret the result.