The Discussion section is where most papers lose their reviewers. Not because the writing deteriorates. Not because the science weakens. But because the author, having presented clean results, suddenly faces an open field with no guardrails. The Results told you what to report. The Methods told you what to describe. The Discussion section structure tells you nothing. It just asks: now what? (See also: Why Most Discussions Fail)
The default response is to discuss everything. Every finding gets a paragraph. Every comparison to previous literature gets a mention. Every limitation gets acknowledged. The result is a Discussion that reads like a second Results section with citations attached–comprehensive, dutiful, and flat.
After years of writing Discussions that were too long, too scattered, and too easily criticized, I developed a filtering system. It is not elegant. But it is reliable. And it has changed how I decide what earns space in the most consequential section of a paper.
The Problem With Discussing Everything
When you discuss every finding with equal weight, you send an implicit message to the reviewer: I do not know which of these findings matters most.
This is a problem because reviewers are looking for a hierarchy of importance. They want to know what you consider your primary contribution, what is secondary, and what is context. When everything is treated equally, the reviewer is forced to construct that hierarchy themselves–and they will not always construct it in your favor.
A Discussion that covers eight findings in eight paragraphs, each with the same depth and the same structure (restate finding, compare to literature, offer tentative interpretation), is not thorough. It is undifferentiated. And undifferentiated discussions invite the most damaging reviewer comment: “The Discussion lacks depth.”
The irony is that depth comes not from covering more, but from covering less with more precision. (Related: The Real Reason Your Papers Feel Fragmented)
The Filtering Principle
The principle I use is simple: a finding earns Discussion space only if it meets at least one of three criteria. If it meets none, it does not get discussed–regardless of how much effort went into producing it.
The three criteria are:
- Primary: The finding directly answers the paper’s main research question.
- Surprising: The finding was unexpected given existing literature or the study’s hypothesis.
- Contradictory: The finding conflicts with established understanding or widely cited prior work.
Everything else–confirmatory results, expected secondary outcomes, findings that align with what everyone already assumes–gets, at most, a single sentence of acknowledgment. It does not get its own paragraph. It does not get a literature comparison. It does not get an interpretation.
This is not about hiding results. The Results section reports everything. The Discussion section interprets selectively.
The Discussion Filtering Matrix
Below is the matrix I use when planning a Discussion. For each finding from the Results, I score it against the three criteria, then assign it a discussion priority.
Step 1: List All Findings
Before writing a single word of Discussion, list every finding from your Results section. Be specific. Not “there were differences between groups” but “Group A showed 23% higher incidence of complication X compared to Group B (p = 0.03).”
Step 2: Score Each Finding
For each finding, answer three questions with Yes or No:
| Finding | Primary? | Surprising? | Contradictory? | Priority |
|---|---|---|---|---|
| Finding 1: [describe] | Yes/No | Yes/No | Yes/No | — |
| Finding 2: [describe] | Yes/No | Yes/No | Yes/No | — |
| Finding 3: [describe] | Yes/No | Yes/No | Yes/No | — |
| Finding 4: [describe] | Yes/No | Yes/No | Yes/No | — |
| Finding 5: [describe] | Yes/No | Yes/No | Yes/No | — |
Step 3: Assign Priority
Use the scoring to assign priority:
- Full Discussion (dedicated subsection, 150-300 words): Finding is Primary, OR meets two or more criteria.
- Brief Discussion (1-2 sentences within another subsection): Finding meets exactly one criterion (Surprising or Contradictory only, not Primary alone).
- Acknowledge Only (single sentence, or omit from Discussion): Finding meets zero criteria. Confirmatory. Expected. Already established.
Step 4: Order by Importance
Arrange your Discussion findings in this sequence:
- Primary finding that answers the research question (always first).
- Surprising findings that challenge assumptions.
- Contradictory findings that conflict with prior work.
- Brief acknowledgments of confirmatory results (grouped, not individual paragraphs).
Completed Example
Here is what the matrix looks like when filled out for a hypothetical surgical outcomes study:
| Finding | Primary? | Surprising? | Contradictory? | Priority |
|---|---|---|---|---|
| Complication rate 23% higher in Group A | Yes | No | No | Full Discussion |
| No difference in mortality between groups | Yes | Yes | Yes | Full Discussion |
| Length of stay correlated with BMI | No | No | No | Acknowledge Only |
| Readmission rate lower than national average | No | Yes | No | Brief Discussion |
| Age distribution matched between groups | No | No | No | Acknowledge Only |
| Wound infection rate contradicts Smith et al. 2021 | No | No | Yes | Brief Discussion |
In this example, only two findings receive full discussion. Two get brief mentions. Two are acknowledged in a sentence or excluded from the Discussion entirely. The paper has six findings but a focused Discussion built around two.
How Each Priority Level Looks in Practice
Full Discussion
A finding receiving full discussion gets its own subsection or a substantial paragraph. The structure within that paragraph follows a specific logic:
- Restate the finding concisely (one sentence, not a repetition of the Results).
- Interpret what it means–not what it shows, but what it implies.
- Compare to the most relevant prior work (two to three citations, not a literature review).
- Address why any discrepancy with prior work might exist.
- State the practical or theoretical implication.
This is where depth lives. One finding, fully explored, is worth more than five findings superficially mentioned.
Brief Discussion
A finding receiving brief discussion gets one to two sentences, typically embedded within a Full Discussion paragraph where it is contextually relevant. It does not get its own paragraph. It does not get a literature comparison. It gets mentioned and connected to a primary finding.
Example: “The lower-than-expected readmission rate, while not the focus of this study, may reflect the standardized discharge protocol used across both groups.”
Acknowledge Only
A finding at this level may not appear in the Discussion at all. If it does, it appears in a single sentence, usually grouped with other confirmatory findings: “Baseline characteristics, including age distribution and comorbidity profiles, were balanced between groups, consistent with the randomization protocol.”
No interpretation. No literature comparison. No paragraph.
The Decisions That Improve the Discussion
Using this matrix forces several decisions that improve Discussion quality without requiring better writing:
Decision 1: What is your primary finding?
If you cannot identify which finding is primary, the paper has a structural problem that precedes the Discussion. Return to the research question. The primary finding is the one that directly answers it.
Decision 2: What do you leave out?
This is the harder decision. Leaving out a finding you spent weeks analyzing feels wasteful. But a Discussion is not a comprehensive account of everything you found. It is a curated interpretation of what matters most.
The findings you exclude from Discussion are still in your paper. They live in the Results. They are available to any reader. You are not hiding them. You are choosing not to spend the reader’s attention on interpretations that add volume without adding meaning.
Decision 3: How much literature is enough?
For a Full Discussion finding, two to four well-chosen citations are usually sufficient. You are not writing a review article. You are contextualizing a specific finding within specific prior work.
For a Brief Discussion finding, one citation or none. For an Acknowledge Only finding, zero.
Researchers who over-cite in the Discussion are often compensating for weak interpretation. If you do not know what your finding means, adding more references will not clarify it.
When to Break the Rules
This matrix is a starting framework, not a rigid protocol. There are legitimate reasons to discuss a finding that scores zero on all three criteria:
- A reviewer specifically asked about it in a prior round. (See: Why Reviewer Comments Sometimes Miss the Point)
- The finding has methodological implications that affect interpretation of the primary results.
- The absence of an expected finding is itself informative.
But these exceptions should be conscious decisions, not defaults. The matrix exists to make the default selective rather than comprehensive.
What This Changes
Before I used this system, my Discussions ran 1500 to 2000 words and received comments about lacking focus. After adopting it, they typically run 800 to 1200 words and receive comments about clarity and depth.
The word count went down. The perceived quality went up. This is not a coincidence.
A focused Discussion signals to the reviewer that the author has done the intellectual work of distinguishing what matters from what merely exists. That signal, more than any individual insight, is what separates a Discussion that satisfies from one that frustrates. (See also: Why Good Studies Still Get Rejected)
The Discussion is not where you report what you found. That already happened. The Discussion is where you demonstrate that you understand what your findings mean. And understanding, by definition, requires selection.
Not everything you found deserves to be discussed. Knowing what to leave out is the skill that makes a Discussion work.
