Writing a research paper is a longer, messier process than any single stage of it — outlining, drafting, revising, and preparing a manuscript for submission are distinct skills, and treating them as one undifferentiated task is a common source of stalled first drafts. This guide covers the writing process itself: the order in which experienced authors typically draft a manuscript’s sections, the style and voice conventions that mark academic prose, how to integrate citations honestly while drafting, how to revise a completed draft, and what "submission-ready" actually requires. It assumes a manuscript’s structure is already planned — for the IMRaD framework, section-by-section outlining, and how outlining differs across STEM, humanities, and systematic-review papers, see CASRAI’s dedicated guide to how to outline a research paper. This guide picks up from that outline and covers everything involved in turning it into a submission-ready manuscript.
Outlining vs. Writing: Two Different Tasks
An outline is a structural decision — what goes in which section, in what order. Writing is the sentence-level work of actually producing the prose that fills that structure, plus everything that happens after a full draft exists: revision, style polishing, citation integration, and preparing the manuscript for a specific journal’s requirements. Researchers who conflate the two often get stuck trying to write perfect prose and make structural decisions simultaneously, which is slower and more frustrating than doing them in sequence. If a working outline doesn’t exist yet, building one first (see the outlining guide linked above) makes everything covered here faster and less prone to major restructuring later.
From Outline to First Draft: What Order to Write Sections In
Most research papers are not written in the order they are eventually read. A widely taught approach in university writing centers and reflected in graduate-level scientific-writing courses is to draft sections roughly in this order, regardless of a paper’s final Introduction-Methods-Results-Discussion sequence:
- Methods first. This section describes what was actually done, and it is easiest to write accurately while procedural detail is fresh and requires the least interpretive judgment — it is closer to transcription than composition. Drafting it first also surfaces any missing methodological detail early, while it is still cheap to go back and check.
- Results second. With Methods drafted, the findings that Methods produced are the next most concrete thing to write down. A useful discipline here is to write the Results section from the outline’s ordered list of findings (see the outlining guide), describing what each table or figure shows without yet interpreting it.
- Introduction and Discussion third and fourth. These are the most interpretive sections — the Introduction has to justify why the study matters and what gap it fills, and the Discussion has to explain what the Results actually mean, acknowledge limitations, and situate the findings against prior work. Both benefit from being written after Results exists, because it is easier to justify a study’s importance and interpret its findings when the findings themselves are already down on paper rather than anticipated.
- Title and abstract last. A structured or unstructured abstract is a compressed summary of the whole paper; writing it before the paper exists means summarizing something that hasn’t been finalized yet, which routinely produces an abstract that no longer matches the final manuscript. Drafting the title and abstract after everything else is settled avoids that drift.
This order is a starting discipline, not a rule every author follows exactly, and it is not universal across genres — a humanities argumentative paper built around a thesis (see the outlining guide’s discipline-specific section) is often easier to draft roughly in reading order, since its Introduction states the thesis the rest of the paper exists to defend. For an IMRaD paper, though, drafting the most concrete, least interpretive sections first is a reliable way to build momentum and avoid staring at a blank Introduction.
Writing in the Register Academic Readers Expect
Academic prose has conventions that differ from other kinds of writing, and following them is part of what signals a manuscript is ready for editorial review, not a stylistic afterthought. The European Association of Science Editors’ EASE Guidelines for Authors and Translators of Scientific Articles — freely available in over two dozen languages and written specifically to help researchers, including non-native English speakers, avoid the structural and phrasing issues that most commonly trip up scientific writing — recommend that research articles be complete, concise, and clear: the tested hypothesis stated plainly in the Introduction, the abstract genuinely informative rather than vague, and the prose understandable without unnecessary jargon.
Some of the recurring, checkable conventions worth applying while drafting:
- Tense. Methods and Results are typically written in the past tense (what was done, what was found); statements of established, still-true fact — background context in the Introduction, or a general principle referenced in the Discussion — are typically written in the present tense. Mixing these inconsistently within a section is a common self-editing catch.
- Voice. Passive voice (“samples were collected”) is traditional in Methods sections and still common, but many journals and style guides, including the AMA Manual of Style, now favor active voice (“we collected samples”) where it doesn’t obscure who did what — check the target journal’s house style rather than assuming one convention applies everywhere.
- Precision over hedging. Vague qualifiers (“a significant amount,” “many studies”) read as imprecise to a reviewer where a specific number or citation is available. Conversely, overclaiming certainty a single study can’t support (using “prove” or “demonstrate conclusively” instead of “suggest” or “are consistent with”) is a common reviewer objection in the Discussion section specifically.
- Consistent terminology. Referring to the same variable, group, or concept by different names across sections (a habit that creeps in during revision, when a term gets changed in one place but not others) reads as sloppy and can genuinely confuse a reader trying to follow the argument.
- Numbers, units, and abbreviations. Define every abbreviation at first use, use consistent significant figures and units throughout, and follow the target journal’s specific style manual (commonly AMA Style in biomedical journals, APA Style in psychology and much of the social sciences) for exactly how these should be formatted — this is one of the more mechanical checks worth doing in a dedicated revision pass rather than while composing.
See CASRAI’s guide to et al. citation rules and citation styles and referencing formats for the reference-formatting side of this same house-style discipline.
Integrating Citations Honestly While Drafting
Citing sources while drafting is not just a mechanical formatting task — it is where originality problems most often originate if handled carelessly. Two distinct issues are worth keeping separate while writing:
- Plagiarism — presenting someone else’s words or ideas as one’s own, whether through direct copying or paraphrasing closely enough that a citation is still owed. See CASRAI’s dictionary entry on plagiarism for the operational definition.
- Text recycling, or self-plagiarism — reusing substantial passages from an author’s own previously published work without disclosure. The Committee on Publication Ethics (COPE) distinguishes acceptable reuse, such as boilerplate Methods language properly cited to an earlier paper, from problematic recycling that misleads readers about how novel the current work actually is. See CASRAI’s dictionary entry on self-plagiarism for the distinction in more depth.
A practical drafting discipline that avoids both: paraphrase in your own sentence structure while actively writing (rather than copying a source sentence and lightly editing it afterward), attach the citation the moment a claim or phrase is drafted rather than planning to “add citations later,” and keep a running note of which specific claims in your own prior publications, if any, are being restated so a disclosure decision can be made deliberately rather than by omission.
Revising a Completed Draft
A first draft is not a final manuscript, and treating revision as a distinct pass — rather than continuous small edits while composing — tends to produce a cleaner result faster. A few techniques worth applying in sequence once a full draft exists:
Structural revision first
Before line-editing prose, check whether the draft actually followed the outline. CASRAI’s outlining guide covers reverse-outlining in detail: going through the finished draft section by section and writing down, without re-reading closely, what each section actually accomplishes. Comparing that list against the original outline surfaces structural drift — a Discussion point with no corresponding Results finding, a section quietly doing two jobs at once — while it is still a cheap fix.
Then a full read-through for consistency
Read the whole manuscript start to finish in one sitting where possible, checking specifically for the tense/voice/terminology consistency issues described above, and for whether every claim in the Discussion actually traces back to something shown in Results. Reading a draft aloud, or having a text-to-speech tool read it back, is a low-cost technique many writing centers recommend for catching awkward phrasing and run-on sentences that are easy to skip over silently.
Get feedback before submission
Co-authors and, where available, a mentor or colleague outside the immediate project are worth involving before a manuscript goes to a journal, not just during peer review. A reader unfamiliar with the day-to-day project is often better positioned than the authors to spot an unexplained jump in logic or an assumption the authors no longer notice because they’re too close to the material. Circulate a draft with a specific, narrow ask (“does the Introduction’s gap statement make sense to someone outside this subfield?”) rather than an open-ended “any thoughts,” which tends to produce less actionable feedback.
Preparing the Manuscript for Submission
A revised draft still needs to be checked against the specific journal’s requirements before submission. This is a distinct step from writing the paper itself, but it commonly requires small rewrites — a title shortened to a journal’s character limit, an abstract restructured into headed sections, references reformatted to a required style — so it belongs in the same overall writing workflow. Worth checking before submitting:
- Author guidelines and formatting. Every journal publishes its own instructions for authors covering word limits, abstract structure, reference style, and file formats; these vary enough between journals that assuming one journal’s conventions apply to another is a common, avoidable source of a returned-without-review decision.
- Figures and tables. See CASRAI’s guide to scientific figure rules for journal submission for resolution, format, and integrity requirements specific to this step.
- Authorship and contributions. Confirm every listed author meets the target journal’s authorship criteria — most biomedical and many other journals reference the ICMJE’s four authorship criteria directly — and that contributions are documented, for example via a CRediT author statement or an author contribution statement.
- Conflict of interest and funding disclosure. See CASRAI’s guide to conflict of interest disclosure in scholarly publication.
- AI-assisted writing disclosure. If a generative AI tool was used to help draft, edit, or translate any part of the manuscript, most major publishers now require that use to be disclosed explicitly, and none permit listing an AI tool as an author. See CASRAI’s guide to journal and publisher policies on generative AI in manuscripts and can AI be listed as an author? for the current policy landscape.
- Journal choice and legitimacy. Before submitting, confirm the target journal is a legitimate venue for the field and check its impact factor and quartile ranking if that matters for the submission’s purpose; see CASRAI’s guide to identifying predatory journals and publishers if a venue’s legitimacy is in any doubt.
After Submission: Responding to Peer Review
Writing does not end at submission. Once a journal returns reviewer comments, composing the response is itself a distinct writing task with its own conventions. W.S. Noble’s widely cited Ten Simple Rules for Writing a Response to Reviewers (PLOS Computational Biology, 2017) lays out practical conventions many journals and editors expect: respond to every point a reviewer raises, even briefly; state plainly what was changed and where in the revised manuscript; and, where a reviewer’s suggestion is not adopted, explain the reasoning respectfully rather than simply declining. The Committee on Publication Ethics (COPE) provides complementary guidance — see CASRAI’s COPE guidelines explained and COPE flowcharts explained — on how authors, reviewers, and editors are expected to handle disagreements during this stage. A point-by-point response letter, keyed to each reviewer comment in order, with the corresponding change quoted or located in the revised manuscript, is the standard format most editors expect.
Common Pitfalls When Writing a Research Paper
- Drafting the Introduction first and getting stuck. Writing the most interpretive section before any findings are on paper is a common cause of stalled first drafts; see the section-ordering discussion above.
- An abstract that no longer matches the paper. Written early and never revised after later changes to Methods or Results, an abstract can quietly drift out of sync with the manuscript it’s supposed to summarize.
- Discussion claims that outrun the data. Asserting more certainty, generalizability, or causal strength than a single study’s Results actually support is one of the most common substantive reviewer objections.
- Inconsistent terminology introduced during revision. Renaming a variable, group, or condition in one section without updating it everywhere else.
- Treating citation as a final formatting pass instead of a drafting-time habit. Planning to “add citations later” increases the risk of both missed citations and inadvertent close paraphrasing that should have been quoted or cited more carefully; see the citation section above.
- Undisclosed AI-assisted drafting. Using a generative AI tool to draft or substantially edit text without disclosing it, where the target journal’s policy requires disclosure — see the submission-preparation section above.
- Skipping a structural revision pass. Line-editing prose before checking whether the draft still matches the outline means structural problems get caught later, when they’re more expensive to fix.
Frequently Asked Questions
What’s the difference between outlining a research paper and writing one?
Outlining is the structural planning stage — deciding what belongs in each section and in what order, before any full prose exists. Writing is producing that prose, plus everything downstream of a first draft: revision, style polishing, citation integration, and preparing the manuscript for a specific journal. See CASRAI’s outlining guide for the planning stage this guide picks up from.
Should I write the abstract first or last?
Last, in most cases. The abstract is a compressed summary of the whole paper, and writing it before the paper is finished risks summarizing something that changes before submission. See the section-ordering discussion above.
What order should I write the sections in?
A common, widely taught approach is Methods, then Results, then Introduction and Discussion, then title and abstract last — see the full explanation above. This isn’t a universal rule, especially for thesis-driven humanities papers, but it’s a reliable default for empirical IMRaD papers.
How do I avoid accidentally plagiarizing while writing?
Paraphrase in your own sentence structure at the moment of drafting rather than lightly editing a source sentence, attach citations as claims are written rather than planning to add them later, and be aware that reusing substantial passages from your own earlier publications without disclosure (text recycling) is a distinct but related issue. See the citation-integration section above and CASRAI’s dictionary entries on plagiarism and self-plagiarism.
Can I use AI writing tools to help write a research paper?
Most major publishers permit AI-assisted drafting or editing in some form but require it to be disclosed, and none currently permit listing an AI tool as an author. Policies differ by publisher and are changing quickly — see CASRAI’s guide to generative AI policies across publishers before assuming a specific use is permitted undisclosed.
How long does it typically take to write a research paper?
There’s no fixed timeline — it depends on the study’s complexity, how many co-authors need to review drafts, and how many revision cycles the target journal’s peer review requires. Drafting sections in the order described above (Methods and Results first) tends to produce a complete first draft faster than starting with the Introduction, but the revision and peer-review stages that follow are frequently longer than the initial drafting itself.
Related CASRAI Resources
- Scholarly Writing pillar — the full manuscript-preparation and publishing-craft cluster this guide belongs to
- How to Outline a Research Paper — the structural planning stage this guide picks up from
- Et Al. Citation Rules
- Citation Styles and Referencing Formats
- Scientific Figure Rules for Journal Submission
- ICMJE Authorship Criteria
- CRediT Author Statement Samples
- Journal and Publisher Policies on Generative AI in Manuscripts
- COPE Guidelines Explained
- How to Identify Predatory Journals and Publishers
- What Is a Dissertation? Structure, Defense, and Deposit
