narrative-frameworks

Narrative frameworks for scientific writing

Last updated: 2026-05-17

Synthesis

Scientific writing has acquired, over the last two decades, a small canon of narrative frameworks that practitioners (and an increasing number of editors) treat as common ground. The most widely cited are Joshua Schimel’s OCAR / LDR / LD structures (the spine of Writing Science), Randy Olson’s ABT (“And, But, Therefore”) for compressed framing, and the inherited IMRaD skeleton that medical and life-sciences journals have used since the 1970s. Stephen Heard’s The Scientist’s Guide to Writing sits alongside these as the most-recommended trade book for early-career researchers, and George Whitesides’s 2004 Advanced Materials essay is the canonical statement of the idea that the outline is the experiment — that writing precedes (and shapes) the research itself.

These frameworks are best understood as complementary lenses, not competitors. OCAR/ABT operate at the document scale (how a paper opens, where the tension lives, how it resolves). IMRaD is a structural convention for the finished artifact. Whitesides’s outline-first method is a process recommendation about when narrative work should happen relative to the bench work. A drafting or critique skill that wants to ground its advice should name which lens it is invoking and at which scale.

The evidence base under these frameworks is uneven. IMRaD’s adoption history is well documented (Sollaci & Pereira 2004). Schimel, Heard, and Whitesides are widely loved but lightly evaluated — there are few controlled studies of whether OCAR-trained writers produce better-cited papers. ABT is closer to a communication heuristic than a research finding. reader-expectation-approach covers the one framework with deeper empirical roots.

Evidence and frameworks

Schimel: OCAR, LDR, LD (Writing Science, 2012)

Joshua Schimel’s Writing Science: How to Write Papers That Get Cited and Proposals That Get Funded (Oxford University Press, 2012; ISBN-13 978-0-19-976024-4 paperback, 978-0-19-976023-7 hardcover) [1] is the most influential single book on scientific writing of the last 15 years. Its central claim is that a paper is a story, and that a story has a problem–solution structure that the reader is biologically wired to follow.

Schimel proposes three nested structures of decreasing patience:

• OCAR — Opening, Challenge, Action, Resolution. Used for full papers where the reader has bought into the contract. The opening establishes characters and setting; the challenge frames the specific question; the action describes the work; the resolution shows what changed.
• LDR — Lead, Development, Resolution. A more compressed structure for abstracts, mid-length pieces, or sections where the reader will not wait.
• LD — Lead, Development. The “newspaper” form. The lead carries the punchline; the development supplies enough evidence to be credible. Used for the very compressed forms (titles, single-sentence summaries, talk abstracts).

OCAR maps cleanly onto the rhetorical Aristotelian arc (set-up, complication, climax, resolution), which is part of why it feels natural; Schimel’s contribution is the explicit mapping to the elements of an empirical paper and the corresponding diagnostic moves (“what is your challenge sentence?”, “where is the resolution actually delivered?”).

Schimel also imports from George Gopen the topic / stress position discipline at the sentence level (see reader-expectation-approach), making Writing Science a useful one-stop reference even though Gopen’s own work is more rigorous on the sentence-level claims.

Heard: The Scientist’s Guide to Writing (2016, 2022)

Stephen Heard’s The Scientist’s Guide to Writing: How to Write More Easily and Effectively throughout Your Scientific Career (Princeton University Press; 1st ed. 2016, ISBN 978-0-691-17022-0; 2nd ed. 2022, ISBN 978-0-691-21920-2) [2] is the most-assigned trade book in graduate writing seminars. Where Schimel is opinionated about narrative form, Heard is practical and process-oriented: how to start, how to break through stuck drafts, how to handle coauthors, how to read peer-review comments without falling apart. The 2nd edition adds chapters on effective reading, journal selection, and preprints.

Heard does not propose a competing structural framework; he largely accepts IMRaD and emphasizes the workflow around the draft. This is why scriptorium’s transformation skills (compression, redundancy-removal) borrow more from Schimel’s structural vocabulary, while drafting and review-handling skills borrow more from Heard’s process advice.

Olson: ABT (Houston, We Have a Narrative, 2015)

Randy Olson’s “And, But, Therefore” framework — published in book form as Houston, We Have a Narrative: Why Science Needs Story (University of Chicago Press, 2015; ISBN 978-0-226-27084-5) [3] — compresses narrative to three connectives. Setup (“And” = context the audience already shares), tension (“But” = the problem or contradiction), resolution (“Therefore” = consequence). Olson traces the form to screenwriter Frank Daniel (via a TV-writers’ remark by Trey Parker and Matt Stone), not to empirical work on scientific communication.

ABT is best treated as a forcing function for compressed framing: it collapses any abstract, talk opening, tweet, or grant specific aim down to a three-clause stress test. If the “But” clause is missing or weak, the framing is descriptive rather than argumentative. ABT does not provide guidance on what to do once the reader is engaged; for that the OCAR/LDR machinery is more useful.

ABT is a marketing-tier framework — widely adopted in science communication (NIH, NSF, AAAS workshops cite it) but lightly evaluated in peer-reviewed literature.

IMRaD and its critics (Sollaci & Pereira, 2004)

The Introduction–Methods–Results–and–Discussion structure is now the dominant convention in biomedical and many life-sciences journals. Sollaci and Pereira’s 50-year survey of four leading internal-medicine journals (BMJ, JAMA, The Lancet, NEJM) showed IMRaD entered use in the 1940s, passed 80% adoption by the 1970s, and was effectively universal by the 1980s [4].

IMRaD is structural, not narrative — it tells you where to put what, but not what story to tell within each section. A common failure mode (named explicitly by Schimel) is the IMRaD-shaped paper that has no story: the Introduction lists prior work without staking a challenge; the Discussion recapitulates Results without naming a resolution. Treating IMRaD as a story constraint rather than as a filing cabinet is the difference between a publishable and a citable paper.

Several disciplines have explicitly broken from strict IMRaD — Cell-family journals invert results-before-methods; many high-impact CS conferences use their own forms; preregistered reports place Methods before Results are known. Reporting guidelines (see reporting-guidelines) often presuppose IMRaD but do not require it.

Whitesides: outline as research plan (Adv. Mater., 2004)

George Whitesides’s “Whitesides’ Group: Writing a Paper,” Advanced Materials 16(15): 1375–1377 (2004), DOI: 10.1002/adma.200400767 [5], is short, dense, and unusual. Its core move is to invert the temporal relationship between writing and research. Rather than writing the paper after the experiments, Whitesides argues you should outline the paper at the start of the project — figures, tables, claims and all — and let the outline tell you which experiments are worth doing.

“The paper, and the ideas that structure it, are the plan for the research. The data come later.”

In Whitesides’s group this took the literal form of pre-experimental outlines that evolved as data accumulated. The reason this is relevant to an agentic writing system: the outline is not a downstream consumer of research output, it is an upstream control surface. A scriptorium that treats outlines as living state (rather than as throwaway scaffolding) is closer to Whitesides’s model than to the conventional “draft after submission-ready results” model.

How this informs scriptorium

• argumentative-flow (current) inherits Schimel’s diagnostic vocabulary — opening, challenge, action, resolution — and uses it to ask whether each paragraph and section delivers the move it claims. The skill should reference OCAR as the structural template it edits toward.
• reviewer-simulation (current) should test the ABT compression on abstracts and specific aims: if the simulated reviewer can’t extract the But clause, the framing is descriptive.
• specific-aims (planned, v0.4) should ground in ABT plus Schimel’s challenge-sentence diagnostic. NIH specific aims pages are the canonical ABT artifact in biomedicine.
• results-narrative / discussion-drafting (planned) should ground in OCAR’s resolution criterion: does the discussion resolve the challenge stated in the introduction, or only restate the results?
• Outline-as-state. The Whitesides idea — outline as living research artifact — argues that MANUSCRIPT_STATE.yaml should be able to hold a full document outline, not just claim summaries. This is a roadmap item.

Open questions / weak evidence

• There are no controlled trials of OCAR vs. unstructured writing on citation outcomes. The evidence for Schimel’s framework is the testimonial weight of editors and trainees, not RCTs.
• ABT’s empirical base is essentially anecdotal. It is useful because it is cheap to apply, not because it has been shown to predict success.
• IMRaD’s universality has not been re-surveyed since Sollaci & Pereira (2004); the rise of preprints, registered reports, and computational-method papers likely shows IMRaD-divergent forms gaining ground in some fields.
• Whitesides’s outline-first method has obvious selection bias: his group was elite and well-funded. Whether it transfers to less-resourced labs has not been studied.

References

1. Schimel J. Writing Science: How to Write Papers That Get Cited and Proposals That Get Funded. Oxford University Press; 2012. ISBN-13 978-0-19-976024-4 (paperback), 978-0-19-976023-7 (hardcover).
2. Heard SB. The Scientist’s Guide to Writing: How to Write More Easily and Effectively throughout Your Scientific Career. 1st ed. Princeton University Press; 2016. ISBN-13 978-0-691-17022-0. 2nd ed. 2022, ISBN-13 978-0-691-21920-2.
3. Olson R. Houston, We Have a Narrative: Why Science Needs Story. University of Chicago Press; 2015. ISBN-13 978-0-226-27084-5.
4. Sollaci LB, Pereira MG. The introduction, methods, results, and discussion (IMRAD) structure: a fifty-year survey. J Med Libr Assoc. 2004;92(3):364–367. PMID: 15243643. PMCID: PMC442179.
5. Whitesides GM. Whitesides’ Group: Writing a Paper. Adv Mater. 2004;16(15):1375–1377. DOI: 10.1002/adma.200400767.