Manuscript Writing

How to Write a Discussion Section That Adds Theoretical Value

The discussion is where most papers earn their place in a journal — or quietly lose it. Here is the move-by-move structure reviewers expect, and how to lift each paragraph from describing your result a second time to making a claim that extends the field.

Research Ramp May 2026 7 min read

You have your results. The tables are clean, the figures are tidy, the analysis holds up. Now you sit down to write the discussion section — and something strange happens. After two pages, you realise you have just described your results again, in slightly more emotional language. You have used the word interesting three times. You have not actually said anything new.

This is the single most common failure pattern in submitted manuscripts. Editors see it constantly. Reviewers flag it with a sentence that quietly ends careers in a particular journal: "the discussion does not move beyond the results."

Learning how to write discussion section material that adds theoretical value is the test of whether your paper deserves to be in a Scopus or SSCI journal. Below is what reviewers are looking for, and a paragraph-level structure you can apply tomorrow.

What the discussion section is actually for

The results section answers the question what did we find? The discussion answers a different one: what does this finding mean for the field? Those are not the same question. One is descriptive. The other is interpretive, comparative, and theoretical.

If you find yourself writing sentences in the discussion that begin "Table 3 shows that…" or "The mean score for Group A was…", stop. Those belong in the results section, not here. The discussion's job is to add an analytical layer on top of those numbers.

62% Of desk rejections cite "limited contribution to theory" or "no clear interpretation" as a reason. Most of that failure is in the discussion.
~3 Distinct theoretical claims is the minimum reviewers expect in a discussion of a full research paper.
1:1 Rough word-count parity between results and discussion in well-balanced manuscripts. If your discussion is shorter, something is missing.

The 5-move discussion paragraph

Strong discussions are not written as one long flowing argument — they are built from disciplined paragraphs, each addressing one finding. Each of those paragraphs follows roughly the same five-move structure. Once you internalise it, your drafting speed roughly doubles.

The structure reviewers expect

One paragraph per major finding. Each paragraph runs these moves in order.

1

Re-state the finding

One sentence. Not a re-description of the table. "Participants in the autonomy condition reported higher engagement than the controlled condition."

2

Interpret what it means

Translate the statistical or thematic pattern into a substantive claim. What does this finding tell us about the phenomenon, not the numbers?

3

Anchor in prior literature

Compare to at least one prior study. Confirm, extend, contradict, or refine — and say which. Vague phrases like "in line with previous work" are not enough; name the studies and the relationship.

4

Explain the mechanism

Offer a plausible why. This is where theory enters. A finding without a proposed mechanism is data; a finding with one is a contribution.

5

State the boundary or scope

Acknowledge where the claim holds — and where it might not. "This relationship may hold more strongly in collectivist contexts than individualist ones," for instance.

A paragraph that runs all five moves is roughly 180–240 words. Most discussion failures we see in client manuscripts stop at move 1 or 2 — restating the finding, interpreting it briefly, then moving on. The theoretical value lives in moves 3, 4, and 5.

What "adding theoretical value" actually looks like

The phrase theoretical contribution sounds enormous. It is not. A contribution is usually one of four small things: extending a known relationship to a new context; refining a construct with finer-grained evidence; identifying a boundary condition; or proposing a mechanism the prior literature had assumed rather than demonstrated.

The contrast below shows the same finding written two ways. Both are accurate. Only one adds value.

Restating the result

"The results show that teacher autonomy support was positively related to student engagement (β = 0.42, p < 0.001). This is an important finding for educators. Teachers who provide more autonomy support tend to have more engaged students."

No mechanism, no comparison, no boundary. The reader learns nothing they could not learn from the regression table.

Adding theoretical value

"Autonomy support predicted engagement (β = 0.42), extending Reeve's (2016) findings from Western contexts into Chinese vocational classrooms — a setting where collectivist norms might have predicted attenuation. The effect held even after controlling for class size, suggesting the mechanism operates at the dyad level rather than the group level. This refines self-determination theory's assumption that autonomy effects are weaker under high-power-distance conditions."

Same finding. Confirms-and-extends, proposes a mechanism, refines a theoretical assumption. This is a contribution.

Notice what the strong version does not do. It does not overclaim. It does not say the finding disproves the theory. It says the effect held under conditions where attenuation was plausible, and proposes what that implies. Reviewers reward disciplined claim-making far more than confident-sounding language.

A theoretical contribution is rarely a new theory. It is almost always a small, careful adjustment to an existing one — clearly stated and clearly defended.

The cardinal mistakes

Three patterns kill discussion sections more often than any others. We see them in roughly two-thirds of drafts that come through our editorial review:

1. Stacked citations without engagement

Writing "This is consistent with Chen (2019), Park (2020), and Mehta (2022)" is not engagement — it is a list. A real comparison names what in those studies matches your finding, and where the relationship differs. If a discussion section has felt suspiciously easy to write, this is usually why.

2. Hedging until the contribution disappears

Caution is a virtue. Caution to the point of vagueness is not. Phrases like "this may possibly suggest that, in some contexts, there could be a tendency for…" communicate uncertainty about your own work. State the claim, then bound it.

3. No clear position on what changes for the field

By the end of the discussion, the reader should be able to answer: what does the field now know, or do, differently because of this paper? If the answer is unclear, the section has not finished its job. Linking back to the contribution you set up in the introduction is essential — see our guide on how to state your research contribution with examples.

One useful test: after drafting your discussion, read the first sentence of every paragraph in order. They should form a logical argument on their own — finding → interpretation → boundary → implication. If they read like a list of disconnected observations, the structure is the problem, not the prose.
Manuscript Health Checker

Check your discussion section's score in 60 seconds

Upload your draft. We score interpretation depth, theoretical engagement, and balance against your results section — and flag what reviewers will catch.

Check Discussion →

A reviewer-grade checklist

Before you submit, walk your discussion through this list. If three or more items fail, the section needs another pass before the manuscript leaves your desk.

Discussion section quality check

  • Each major finding has its own dedicated paragraph (not bundled).
  • Each paragraph runs the five moves — restate, interpret, compare, mechanise, bound.
  • No paragraph re-describes a table or restates a regression coefficient as its main point.
  • At least three prior studies are engaged with, not just cited in passing.
  • For each finding, a mechanism is proposed — even if tentatively.
  • Boundary conditions are acknowledged explicitly (population, context, design).
  • The contribution stated in the introduction is revisited and supported here.
  • The discussion's word count is roughly comparable to the results section.
  • Hedging language ("may," "could," "suggests") is used precisely, not as a stylistic default.
  • The first sentence of each paragraph, read in sequence, tells a coherent story.

The bottom line

A discussion section that adds theoretical value is not a matter of vocabulary — it is a matter of moves. Restate what you found, interpret it in substantive terms, compare it sharply to what was known, propose why it might work that way, and say where the claim ends. Do that for every major finding and you have a discussion that earns its place in the manuscript.

If you are unsure whether your draft clears the bar, that is exactly the kind of paragraph-by-paragraph check our editorial team runs on every manuscript through our manuscript editing service.

Free in 60 Seconds

Find out where your discussion is losing the reader

Upload your draft to the Manuscript Health Checker. You will see a scored breakdown of interpretation depth, theoretical engagement, comparison quality, and balance against your results section — with the exact paragraphs flagged.

Check Your Discussion — Free Not sure where to start? Book a free 30-minute consultation with a publication expert.