Last updated June 27, 2026

Academic writing: a researcher’s guide to manuscripts that get published

Every manuscript that lands on a reviewer’s desk carries the same burden: it must communicate what you did, why it matters, and how reliably you did it — in prose clear enough that a specialist in an adjacent subfield can follow the reasoning without effort. That is a high bar, and it is not simply a matter of grammatical correctness. Academic writing is an argument, assembled in a recognized structure, with every sentence doing a defined job. This guide covers that structure, the choices that shape it, and the habits that separate manuscripts reviewers recommend from those they decline.

The anatomy of a research paper: IMRaD

Most empirical journals organize manuscripts around the IMRaD framework — Introduction, Methods, Results, and Discussion — with an Abstract at the front and (sometimes) a Conclusion at the end. Understanding what each section is for tells you what to put in it and what to leave out.

Abstract

The abstract is not a preview. It is a compressed, self-contained argument: background and problem, what you did, what you found, and what it means. A reader who only reads the abstract should be able to decide whether the paper is relevant to their work and whether your central claim is plausible.

Write the abstract last. Every word carries weight because the abstract is how your paper will be indexed, surfaced in searches, and assessed in the first thirty seconds by a reviewer who has not agreed to accept it yet. Aim for precision over generality: “We found a 12-point reduction in symptom severity” is more useful than “We observed improvements.”

Structured abstracts (with subheadings) are standard in clinical and health-science journals. Check the target journal’s guide for authors before you draft yours.

Introduction

The introduction does three things: it establishes why the topic matters, it shows what is already known and where the gap lies, and it states clearly what your study does to address that gap. This is sometimes called the “problem-solution” or “known-gap-present” structure, and it should tighten as it goes — beginning broad, focusing progressively, ending with your research question or hypothesis.

A common failure mode is the literature dump: three paragraphs cataloguing prior work without building a coherent argument about why your study is needed. Every study you cite should earn its place by either defining the terrain your work builds on or identifying the limitation your work corrects.

For detailed guidance on how to frame existing scholarship, see our guide to writing a literature review.

Methods

The methods section serves one primary reader: the person trying to reproduce your study. Write for reproducibility, not for brevity. A reader should be able to reconstruct your protocol from your description alone — including participant selection criteria, instruments used, how data were collected, and any analytical decisions made before you looked at the results.

Reporting clarity here is a common site of reviewer criticism. “Data were analysed using standard methods” does not tell a reviewer which methods, and it cannot. Name the specific tests. Describe how you handled missing data, outliers, and any violations of distributional assumptions. If you pre-registered your study, say so and link the registration.

Whether you are working with qualitative or quantitative data, the methods section needs to communicate that your choices were principled, not post-hoc. This question — which approach fits which research question — is explored in depth in our guide to qualitative vs quantitative research.

If your analysis involves statistical modelling, be explicit about model specification, software versions, and any corrections for multiple comparisons. Our biostatistics resource covers the most common reporting requirements.

Results

The results section reports findings without interpretation. Reserve explanation and meaning for the discussion. A results section that begins editorializing — “Surprisingly, we found…” or “This remarkable result suggests…” — signals that the author has not kept the sections cleanly separated.

Figures and tables should be able to stand alone, with informative captions. The prose in your results section should guide the reader through the figures, not repeat every number in them. Highlight what is notable; let the figure carry the data.

Report effect sizes and confidence intervals alongside p-values. A result that is statistically significant may be trivially small in practical terms; one that misses conventional significance thresholds may still be meaningful given the effect size and sample. Journals increasingly require you to present the full picture.

Discussion

The discussion is where you interpret your findings in light of the existing literature, address limitations honestly, and explain what your results mean for the field. It is also where many manuscripts lose reviewer confidence, because writers conflate what they found with what they hoped to find.

A strong discussion:

The discussion should not introduce new data. If you find yourself wanting to report a result here that did not appear in your results section, move it there.

Conclusion

Not all journals require a separate conclusion section. When they do, keep it short: one to three paragraphs summarizing the key finding and its main implication. Avoid repeating the abstract verbatim. The conclusion is an opportunity to articulate the “so what” with economy — it should leave the reader with a single clear takeaway, not a catalogue of everything the paper covered.

Structuring a literature review and an argument

A literature review is not a chronological account of who discovered what. It is a synthesis that maps the intellectual landscape: what is established, what is debated, and where knowledge runs out. If the review reads like a list of summaries — “Jones (2019) found X. Smith (2021) found Y. Patel (2022) found Z” — it is cataloguing rather than synthesizing.

Group the literature thematically or by the logic of the problem. Show how studies relate to one another: do they converge on a finding, or does the evidence conflict? If it conflicts, explain why — different methods, different populations, different operationalizations of a key construct. That explanatory work is the analytical contribution of the review itself.

For a standalone review paper or a dissertation chapter, the structure becomes even more important. See our dedicated resource on what a literature review is for guidance on organizing a full stand-alone review.

On the question of sources: peer-reviewed empirical studies carry more evidential weight than opinion pieces, textbooks, or news articles. Knowing when each type of source is appropriate — and how to handle the distinction between original research and accounts of it — is covered in our guide to primary vs secondary sources.

Qualitative and quantitative research: reporting what you did clearly

The distinction between qualitative and quantitative research is not simply one of numbers versus words. It reflects a deeper choice about what kind of knowledge claim you are making. Quantitative studies test hypotheses about relationships between measurable variables in populations; qualitative studies explore how people make meaning of their experience, typically in contexts where measurement would impoverish understanding.

Both are legitimate, and both have specific reporting conventions. Quantitative methods demand explicit sample sizes, tests, effect measures, and confidence intervals. Qualitative methods demand thick description of the analytical process — how you collected data, how you developed your coding scheme, how you moved from raw data to themes, and how you assured trustworthiness (member checking, thick description, reflexivity, or other techniques standard in your tradition).

Mixed-methods papers carry the burden of both. If you mix methods, explain the rationale for the integration — why the combination is more useful than either approach alone.

Regardless of approach, the guiding principle is this: a reader should never have to wonder what you actually did. Vagueness in the methods section is not humility; it is a gap that reviewers will flag.

Clarity, concision, and academic style

Active vs passive voice

Academic writing has a long association with the passive voice, partly for historical reasons and partly because the passive genuinely fits some contexts — it foregrounds the action rather than the agent, which is appropriate when the agent is obvious or irrelevant (“Blood samples were collected at baseline”). But overuse of the passive can make writing opaque, particularly when it conceals who made decisions or how they were made.

Many journals and style guides now encourage the active voice in methods and discussion sections. “We analysed the data using…” is cleaner and more transparent than “The data were analysed using…” — especially for a reviewer trying to assess your analytical choices.

This is not a rule to follow mechanically. The question is always: does the passive voice serve clarity here, or does it obscure the actor? Our full guide on active vs passive voice in academic writing works through the tradeoffs section by section.

Hedging

Hedging is not evasion — it is epistemic accuracy. When your findings support a conclusion but do not prove it, language should reflect that. “These results suggest…” is honest; “These results prove…” is overclaiming. Reviewers who specialize in your area will notice when you overstate, and it damages the credibility of everything else you write.

Common hedges include modal verbs (may, might, could, would), adverbs (possibly, likely, arguably), and explicit markers of uncertainty (these findings are consistent with, one interpretation is). Use them purposefully, and do not hedge so much that you obscure your actual claim.

Signposting

Signposting is the practice of telling readers where they are in your argument. Transitions between sections and paragraphs should make the logical structure explicit: “Having established that X, we now turn to Y” or “While the previous section examined A, B raises a different concern.” Readers should never lose track of why a given paragraph exists.

Concision

Academic writing tends toward the verbose because writers mistake length for thoroughness. The goal is to say what you mean in as few words as it takes to say it clearly — no fewer, no more. Common sources of avoidable length include:

Edit for concision in a dedicated pass. Reading your prose aloud is the fastest way to hear where sentences have bloated or where a simpler word would do the job of a complex one.

Common writing mistakes that trigger reviewer criticism

Reviewers evaluate both the science and the writing. These are the most frequent writing-level objections:

Mismatch between claim and evidence. Conclusions in the abstract or discussion that go beyond what the results actually show. This is the single most common reason for major revisions.

Unclear or missing research question. If a reviewer cannot identify your specific question by the end of the introduction, the rest of the paper has no anchor.

Imprecise terminology. Using a technical term inconsistently, or using lay synonyms interchangeably with technical terms, signals imprecision in the underlying thinking.

Passive structure obscuring decisions. “Participants were excluded if…” without explaining who made the exclusion decision or on what basis.

Results mixed with interpretation. Editorializing in the results section where neutrality is required.

Underpowered discussion of limitations. A single formulaic sentence — “This study has several limitations” — followed by brief acknowledgments does not satisfy reviewers who know the literature. Engage specifically with how each limitation affects the interpretation of your findings.

Inconsistent tense. Most style guides use past tense for what you did and found, and present tense for what is generally true or what your study demonstrates. Switching mid-section is jarring and signals hasty writing.

If you are preparing to submit and want systematic feedback on these issues before the paper goes out, start a free review — PerfectPaper reads your manuscript the way a peer reviewer would and surfaces these problems before they reach a journal.

A revision workflow that works

First drafts are for getting the argument down. Revision is where the paper becomes good. A repeatable revision workflow prevents the common trap of cycling through the same issues without resolving them.

Pass 1 — argument check. Read only for logic and structure. Does each section do its job? Is the research question clearly stated? Does the discussion address what the results actually show? Fix structural problems before polishing sentences.

Pass 2 — evidence and claims. Does every assertion have appropriate support? Are citations used to support claims, or just to appear thorough? Are all claims proportionate to the evidence?

Pass 3 — clarity and concision. Sentence-level editing: cut deadwood, replace nominalizations with verbs, clarify antecedents, check for passive constructions that obscure agency.

Pass 4 — feedback loop. Give the manuscript to a colleague or co-author who was not involved in writing it. Ask them to flag anything they had to re-read to understand. Do not ask whether they liked it; ask what they found unclear.

Pass 5 — final polish. Format references, check journal style requirements, read aloud for flow, verify that tables and figures match text descriptions.

Getting substantive feedback at pass 4 is the step writers most often skip under deadline pressure — and most often regret skipping. The peer-review process itself (see our guide to peer review) is less painful when your manuscript has already been stress-tested by an outside reader.

If you are preparing a dissertation chapter rather than a journal article, the revision stakes are similar but the structural conventions differ — our guide to dissertation proofreading covers those distinctions. And when the manuscript is ready to submit, our journal submission guide walks through the targeting and submission process.

Frequently asked questions

What is the IMRaD structure and do all journals use it?

IMRaD — Introduction, Methods, Results, and Discussion — is the organizing framework for most empirical research papers in the natural sciences, social sciences, and health fields. It reflects the logical sequence of a research study: why it was done, how it was done, what was found, and what it means. Review papers, theoretical papers, and humanities articles often use different structures (thematic sections, a single extended argument), so check the conventions of your target journal before drafting.

How long should each section of a research paper be?

There is no universal ratio. In general, the methods and results sections are usually the longest in empirical papers, because they carry the primary evidence. The introduction is typically shorter — enough to frame the question, not a full review of the literature. A common failure mode is an introduction that is longer than the methods, which signals that the author is more comfortable with background than with their own study. Follow your target journal’s word limits and look at recent published papers in that journal for a sense of proportion.

How do I write a literature review that synthesizes rather than summarizes?

A synthesizing literature review groups studies by theme, finding, or method — not by publication date or author — and explains how they relate to one another. For each cluster of studies, ask: what do these studies agree on, where do they diverge, and what does the pattern of agreement and divergence tell us? The explanatory move — connecting individual studies into a coherent picture — is the analytical work that distinguishes synthesis from summary. For a step-by-step approach, see our literature review guide.

When should I use the active voice and when is the passive appropriate?

Use the active voice when the identity of the agent matters for interpretation — particularly for decisions you made during data collection or analysis. Use the passive voice when the agent is obvious, unimportant, or when foregrounding the action rather than the actor improves clarity. In practice, methods sections benefit from both: active voice for decisions (“We excluded participants who…”) and passive for procedures where the agent is irrelevant (“Samples were stored at −80°C”). Our active vs passive voice guide goes deeper on this.

What do reviewers mean when they say my argument is “unclear”?

Usually one of three things: the research question is not stated explicitly, the connection between the introduction and the rest of the paper is not made clear, or the discussion draws conclusions that the results do not support. Re-read your introduction asking: could a stranger identify my specific research question from this? Then read the discussion asking: do my conclusions follow directly from the findings I reported, or am I inferring further than the data allow? Both issues are structural, and fixing them requires revision at the section level, not sentence editing.

How is writing a journal article different from writing a dissertation chapter?

A journal article is compressed — typically 5,000 to 8,000 words — and assumes a specialist audience. Every claim must be grounded quickly, and the literature review is selective. A dissertation chapter develops an argument at greater length, situates work within a broader intellectual tradition, and often needs to justify methodological choices to an examining committee that may include generalists. The revision process for dissertations also typically involves an advisor review cycle before any external submission. See our dissertation proofreading guide for the specific considerations that apply.

When your draft is as strong as you can make it through self-editing, the next step is a reader who will engage with the argument the way a reviewer will. Start a free review on PerfectPaper and get structured, substantive feedback on your manuscript before it reaches a journal.