Discussion Section That Goes Beyond Restating Results
Discussion is where peer reviewers catch overreach — and where authors most want to overreach. After a year of results, there's a natural urge to explain everything the data "imply." That urge is where most Discussion sections go wrong.
The most common failure isn't saying too little. It's saying the right things in the wrong order, at the wrong length, with the wrong confidence level.
What a Discussion Is Actually Doing
A Discussion has one job: place your findings in context without overclaiming them. Not "summarize results" — reviewers will flag that as redundant. Not "speculate freely" — reviewers will flag that as overreach. The target is a narrow band between those two.
Every functional Discussion answers four questions, in order:
- What did we find, and how does it compare to what was already known?
- What mechanism or explanation accounts for this finding?
- What limits how far these conclusions can travel?
- What should change as a result — and what shouldn't?
Four questions. Four paragraphs. In that order.
Paragraph 1: Principal Finding in Context
State the main finding in one sentence. Then immediately compare it to the prior literature. Not a bare restatement of the result — that belongs in the Abstract. The target is: "Our finding of X is consistent with [prior work] but diverges from [contradictory finding], which may reflect [one-sentence explanation]."
Too-far marker: If this paragraph exceeds 120 words, you're restating Results. Cut it.
Paragraph 2: Mechanism or Interpretation
This is the one paragraph where you're allowed to think out loud — carefully. Propose a mechanism for your finding. Cite support if it exists. If it doesn't, say so explicitly: "One possible explanation is X, though this is speculative and was not tested in the current study."
That hedge is not weakness. It's the sentence that keeps Reviewer 2 from writing "the authors overclaim in the Discussion."
Too-far marker: If you're proposing a mechanism with no citation and no explicit hedge, you've crossed from interpretation to overreach.
Paragraph 3: Limitations
Most limitations paragraphs come in two broken flavors: defensive (denying obvious weaknesses) or self-sabotaging (listing every concern a reviewer might raise). Neither works.
The structure that holds: three substantive limitations, ordered from most-to-least threatening to your study's validity, each with a one-sentence mitigation. Not "small sample size was a limitation" — "the sample of N=47 limits precision of the subgroup estimates; a larger multicentre cohort would be required to detect the differential effect hypothesized in [direction]."
Too-far marker: A limitation with no mitigation sentence leaves the criticism open for a reviewer to complete in the worst possible direction.
Paragraph 4: Implications
What changes as a result of this work — and what doesn't. Both parts matter. Implications should match the strength of your evidence. A pilot study doesn't change clinical guidelines; it raises a question worth testing in a larger trial. Say exactly that, and stop.
Too-far marker: If the word "should" appears without a direct data connection or a supporting citation, you've left interpretation and entered advocacy.
The Tone Problem
Discussion sections fail on hedging calibration as often as on structure. The failure modes are on both sides: under-hedged ("this proves that X") and over-hedged ("these data may potentially suggest, in certain contexts, that X might possibly..."). Both are rejectable.
The target is confident uncertainty. "These findings suggest X — consistent with [prior work] — though longitudinal data would be required to confirm [specific claim]." First-person helps: "We interpret this as..." is cleaner than "The data may be interpreted as indicating..." and less likely to trigger passive-voice comments.
If your Discussion reads like an annotated Results section, restructure around the four questions above. The test: if a reviewer read only your Discussion, would they know what question you answered, how your answer fits the field, what your honest caveats are, and what to do next?
One Process Note
Lock the Results before writing Discussion. Discussion written alongside still-changing Results drifts — you end up contextualizing findings that will be revised. Write Discussion last, after figures and Results are stable — this is the same sequencing this Discussion structure slots into within the 4-week manuscript writing sprint.
The Discussion Section Playbook ($7) at researchcraft.gumroad.com extends this to a 6-block architecture with before/after worked examples and a 9-item diagnostic checklist for the "superficial discussion" reviewer comment.