IMRaD (Introduction, Methods, Results, and Discussion) is the default shape a peer reviewer expects from an empirical paper, but it is not one fixed template applied identically everywhere. CASRAI’s IMRaD Structure entry defines what belongs in each of the four core sections, and the IMRaD Structure worked example guide walks through a single illustrative paper section by section. This guide takes a different angle: it maps how the skeleton itself bends, gets renamed, gets split, or gets dropped entirely as it moves across clinical medicine, engineering and computer science, the social sciences, evidence-synthesis research, and the humanities — so an author moving between fields, or a research-administration office supporting authors across several departments, can recognize which variant applies before drafting rather than assuming one universal rule.
Why IMRaD varies by discipline in the first place
The International Committee of Medical Journal Editors (ICMJE) describes IMRaD as “not an arbitrary publication format but a reflection of the process of scientific discovery” — each section corresponds to a stage of empirical research (why you started, what you did, what you found, what it means). That rationale holds cleanly for a single hypothesis-driven experiment. It holds less cleanly the moment a field’s actual research process doesn’t map one-to-one onto that four-stage sequence: a randomized controlled trial has regulatory reporting obligations a lab experiment doesn’t; a systematic review has no “participants” of its own, only studies it screens; a software-systems paper often has no single “result” separable from the system it built and evaluated; a humanities article may not report empirical findings at all. Each variant below is a genuine adaptation to a real difference in how that field produces and reports knowledge — not an arbitrary departure from a “correct” template.
Clinical and medical research: structured, checklist-driven variants
Biomedical IMRaD is the closest to the ICMJE’s canonical description, but clinical research layers reporting-guideline structure on top of the four base sections rather than replacing them:
- Randomized controlled trials. The CONSORT Statement doesn’t rename IMRaD’s sections, but it specifies detailed, checklist-mandated subheadings within them: Methods typically breaks into trial design, participants and eligibility criteria, interventions, outcomes, sample-size determination, randomization and allocation concealment, blinding, and statistical methods, each reported as its own labeled block rather than free-form prose. Results correspondingly reports participant flow (often via a CONSORT flow diagram), baseline characteristics, and outcomes in that structured order. A trial report that omits one of these named CONSORT subsections is missing a required element, not just writing informally.
- Case reports. A single-patient case report doesn’t have “Methods” or “Results” in the trial sense at all — the CARE guidelines structure a case report around patient information, clinical findings, timeline, diagnostic assessment, therapeutic intervention, and follow-up/outcomes instead, reflecting a narrative-clinical rather than experimental process.
- “Materials and Methods” naming. Many biomedical and life-science journals use “Materials and Methods” rather than “Methods” as the section label — the content and function are identical to standard IMRaD Methods; this is a naming convention, not a structural variant.
Engineering and computer science: methods-and-evaluation-heavy structures
Engineering, computer science, and related applied-technical fields commonly use a structure that shares IMRaD’s underlying logic but visibly diverges in section names and emphasis, closer to the conventions ACM- and IEEE-format conference and journal papers actually use in practice:
- A “Related Work” section — usually placed immediately after the Introduction, separate from it — situates the paper’s contribution against prior systems or approaches. Biomedical IMRaD folds this comparison into the Introduction’s narrowing-to-the-gap paragraph instead of giving it a dedicated section.
- “Methods” often becomes “Approach,” “System Design,” “Methodology,” or “Implementation” — and for a systems or software paper, this section is frequently the paper’s center of gravity: it describes what was built, not only how a study was conducted.
- “Results” often becomes “Evaluation” or “Experiments,” and commonly merges what biomedical IMRaD keeps separate — the experimental setup (which is procedurally a Methods concern), the metrics and results themselves, and some comparative interpretation — into one combined section, particularly in shorter conference-length papers where space is constrained.
- Discussion is frequently compressed or folded into the Evaluation/Results section, followed directly by a Conclusion (and often a distinct “Future Work” subsection) rather than a separate, expansive Discussion of the kind biomedical IMRaD expects.
The underlying stages IMRaD represents — motivation, method, outcome, interpretation — are usually all still present; what differs is where the section boundaries are drawn and what each is called. Authors moving from a biomedical to a CS/engineering venue (or vice versa) should check the target venue’s own author guidelines and recent published examples rather than assume section names transfer directly.
Social sciences: closest structural relative, with its own conventions
Psychology is, historically, one of IMRaD’s origin points in the social sciences: the American Psychological Association’s Publication Manual formalizes an Introduction–Method–Results–Discussion sequence that maps almost directly onto IMRaD, which is why psychology, much of education research, and other quantitative social-science fields follow a structure very close to the biomedical original. Two differences are still worth flagging:
- An explicit, often separately labeled Literature Review is more common as its own section (sometimes with its own subheadings) than in biomedical IMRaD, where the equivalent material is compressed into the Introduction’s narrowing paragraph.
- Qualitative social-science research frequently departs from IMRaD more substantially than quantitative work in the same discipline. CASRAI’s guide to writing the methodology section of a qualitative research paper covers this directly: qualitative papers often replace “Results” with “Findings,” organize that section thematically (by theme or code emerging from the analysis) rather than by hypothesis or outcome measure, and frequently merge Findings and Discussion into a single “Findings and Discussion” section because thematic interpretation is harder to cleanly separate from thematic reporting than statistical results are from their interpretation.
Evidence synthesis: systematic reviews and meta-analyses
A systematic review or meta-analysis doesn’t collect its own participants or run its own experiment — its “Methods” describe how studies were found and evaluated, not how data was generated. CASRAI’s guides to PRISMA and systematic review methodology and the PRISMA-P aligned systematic review protocol template cover this structure in depth; in outline, the PRISMA reporting standard replaces IMRaD’s Methods with sections covering eligibility criteria, information sources and search strategy, study selection process, data extraction, and risk-of-bias assessment, and replaces Results with study selection (often a PRISMA flow diagram), study characteristics, and synthesized results organized by outcome rather than by a single hypothesis. The Introduction and Discussion sections remain structurally closer to standard IMRaD.
Humanities: largely non-IMRaD by design
Most humanities scholarship — literary criticism, historical analysis, philosophy, much of qualitative cultural studies — does not use IMRaD at all, and this isn’t an omission to correct: IMRaD is built for reporting original empirical research with a discrete methods-generates-results-generates-interpretation process, and a thesis-driven argumentative essay doesn’t have that process. Humanities articles are typically organized thematically or chronologically around a sustained argument, with no separate “Methods” section (the analytical approach is usually described inline, where relevant, rather than isolated into its own section) and no “Results” section distinct from the argument itself. Authors and administrators supporting cross-disciplinary or humanities-adjacent programs should not expect — or impose — IMRaD conventions on this kind of scholarship; doing so mismatches the structure to work it wasn’t designed to organize.
Multi-study and mixed-methods papers
A paper reporting more than one study or combining quantitative and qualitative components often can’t use a single Results/Discussion pair at all. Two common adaptations: repeating a full Methods–Results block per study (Study 1 Methods, Study 1 Results, Study 2 Methods, Study 2 Results…) followed by a single integrative General Discussion, common in psychology; or, for mixed-methods designs, separate quantitative and qualitative Results subsections reported side by side before a combined Discussion that explicitly addresses how the two strands of evidence relate. Neither pattern is a deviation from IMRaD’s underlying logic so much as an extension of it to a research design IMRaD’s original four-section form wasn’t built to hold on its own.
How to decide which structure to use
The reliable answer is never “pick the version you prefer” — it’s the target venue’s own instructions for authors, plus recent issues of that specific journal or conference, read as the actual working template rather than a general guideline. In order of authority: (1) the target journal’s or conference’s current author guidelines, which frequently specify section names and required subheadings explicitly; (2) the reporting guideline that governs the study type, where one exists (CONSORT for randomized trials, PRISMA for systematic reviews, CARE for case reports, STROBE for observational studies); (3) 2-3 recently published papers of the same study type in that same venue, read specifically for section structure; (4) general biomedical IMRaD, as the fallback default when a field or venue gives no more specific guidance. Deviating from a venue’s expected structure without a stated reason is itself a signal a reviewer or editor notices — when a genuine reason exists (a study design the standard structure doesn’t fit well), it is worth naming briefly in a cover letter rather than leaving the reviewer to guess why the paper looks unusual.
Frequently asked questions
Does every empirical paper have to use IMRaD?
No. IMRaD is the default expectation for original empirical research in the biomedical and physical sciences and much of the quantitative social sciences, but it is a convention, not a universal requirement — reporting guidelines, venue-specific formats, and non-empirical scholarship (most humanities work) legitimately use other structures. What matters is meeting the actual expectations of the target venue and study type, not IMRaD specifically.
Why do computer science and engineering papers use “Related Work” as its own section instead of folding it into the Introduction?
Because positioning a technical contribution against prior systems or approaches is often substantial enough — and central enough to establishing novelty in a fast-moving field — to need its own space, rather than being compressed into the Introduction’s brief narrowing-to-the-gap paragraph that biomedical IMRaD relies on.
Is “Findings” the same thing as “Results”?
Functionally, in most qualitative research, yes — “Findings” is the qualitative-research equivalent of “Results,” reporting what the analysis produced. It’s typically organized by theme rather than by hypothesis or outcome measure, and is more often merged with Discussion than quantitative Results sections are, because thematic interpretation is harder to fully separate from thematic reporting.
Should a systematic review call its second section “Methods” or something else?
Most journals still label it “Methods,” but the content inside follows PRISMA’s structure (eligibility criteria, search strategy, study selection, data extraction, risk-of-bias assessment) rather than the participants/materials/procedure structure a primary-data study’s Methods section would contain. See CASRAI’s PRISMA and systematic review methodology guide for the full breakdown.
What’s the safest default if a target venue gives no specific structural guidance?
Standard biomedical IMRaD (Introduction, Methods, Results, Discussion) as described in the ICMJE Recommendations — it is the most widely recognized default across empirical research and the structure a reviewer is least likely to find unexpected, even outside biomedicine.
