Key Takeaways — Chapter 35: Writing for Science

The summary card. Use it to re-ground before the quiz, or to review weeks later.

The one idea

Publishing is a negotiation with named human gatekeepers, not a submission to an impartial standard. The science has to be right — Chapter 14 saw to that — but being right was never sufficient. An editor decides whether reviewers ever see your paper; reviewers decide whether it's accepted; and the conventions you follow, the cover letter you write, and the response letter you send are all moves in that negotiation. The same result, framed and defended by someone who understands this, gets published; framed by someone who thinks the science should speak for itself, gets rejected.

🚪 Threshold concept: Publishing is a negotiation with named humans. Before you cross this, the editor and reviewers feel like an exam board grading a fixed answer, and a critical review feels like a failing mark to argue against. After, you see reviewers as collaborators-by-friction whose buy-in you negotiate, the editor as a person you persuade, and the cover and response letters as the instruments of that persuasion — not paperwork around the "real" science.

There is no "science voice" — only field voices

  • Conventions are local, not universal. Chemistry Methods still favor the passive ("the sample was heated"); molecular and cell biology increasingly favor the active first-person ("we treated the cells"), and several major journals encourage it.
  • You don't learn your field's voice from a rule — you learn it by reading three recent papers in your exact target journal and matching them. The convention is on the page.
  • Use the active voice for your own reasoning and choices (Introduction, Discussion) even in a passive-leaning field. The passive belongs to the recipe (Methods), where the actor is genuinely irrelevant — that's the one condition Chapter 3 said justifies it.

Report statistics so a skeptic can check you

  • "Significant" or "better" without a number is not a result — it's the statistical cousin of "the results are promising."
  • The grammar of a reported test: statistic(degrees of freedom) = value, exact p, effect size. Example: t(38) = 9.4, p < .001, d = 2.1.
  • The formatting tells that mark a pro: italicize statistics (t, F, p, r, M, SD, d) not the numbers; lowercase p; report the exact p-value; and no leading zero on p-values or correlations (p = .045, not 0.045) because they can't exceed 1.
  • Always pair a p-value with an effect size (Cohen's d, η², r, an odds ratio, or a confidence interval). The p-value says an effect is detectable; the effect size says whether it's big enough to matter — and with a large sample, even a trivial difference is "significant."

The apparatus around the paper

  • Supplementary materials: the main text gets what the argument needs to stand; the supplement gets what a skeptic needs to verify it (full data, replicates, robustness checks). Never move an inconvenient result or a real limitation to the SI to keep the paper clean — that's hiding, not supporting. Write the supplement as carefully as the paper (S-prefixed callouts, stand-alone captions).
  • Preprint = about timing: a complete manuscript posted publicly (arXiv, bioRxiv, ChemRxiv) before/during review — readable and citable early, but explicitly not peer-reviewed (label it). Check your journal's preprint policy first.
  • Open access = about the published version: gold OA (free to read; author pays an APC), green OA (self-archive a copy for free), or subscription. Many funders now mandate OA; green OA is the free route when an APC is out of reach. A fee alone is not a predatory tell — Chapter 14's tells (no real review + flattery + speed) still apply.
  • Author-contribution statements (often the CRediT taxonomy) make credit precise, localize responsibility, and force the authorship conversation early. The ICMJE bar: authorship = substantial intellectual contribution + drafting/revising + approval + accountability. Funding or supervision alone doesn't qualify.

The two letters that decide a paper's fate

  • Cover letter (gets you past the editor's desk): state the specific contribution with its headline number, make the significance case for this journal, handle the housekeeping, and suggest/exclude reviewers. Don't restate the title; don't flatter; don't oversell.
  • Response to reviewers (gets you past the reviewers): address every comment; concede what's right, gladly and first; disagree with evidence, not attitude. Mechanics: quote each comment, respond beneath it, state exactly what changed and where. Concede first, defend narrowly — conceding valid points buys the credibility that makes your firm points land.
  • Different journal audiences = different documents. Broad journals (Nature, Science) need near-zero jargon, broad significance, and method in the SI; specialist journals expect field vocabulary, within-field significance, and method in the main text. Choose the venue before you write — it shapes the first sentence.

Themes this chapter surfaced: #2 audience-is-everything (field voices, broad vs. specialist journals, the editor and reviewers as readers) · #3 clarity-is-not-the-enemy-of-precision (statistical reporting; jargon helps a shared audience and blocks a broad one) · #4 revision-is-where-the-writing-happens (the response letter) · #5 structure-serves-the-reader (main text vs. supplement).

Threshold concept: Publishing is a negotiation with named humans, not a submission to an impartial standard.

Back to: Chapter 35 · Exercises · Quiz · Further Reading