If your Results section is ten clean paragraphs of numbers and your Discussion is two paragraphs of "this is important," the paper doesn't land. The Discussion section of a research paper is where interpretation lives. It's the move from "here are the numbers" to "here's what they mean, how they fit with prior work, and why it matters." Done well, it's the section reviewers remember. Done poorly, it's where papers get rejected — either for overclaiming or for failing to commit to an interpretation at all.
PaperDraft is a writing assistant, not a paper generator — the draft is your starting point, not your submission. You are responsible for editing, verifying sources, and following your school's academic integrity policy.
This guide walks through the four-move structure of a strong discussion section, what goes where, common errors, and examples. For the section that comes before it, see writing the results section. For paper structure overall, see our pillar on research paper structure.
The Four-Move Structure of a Discussion
A research paper discussion section, across nearly every empirical field, follows a four-move pattern:
- Restate the key finding. One or two sentences. What did you find?
- Compare to prior work. How does it fit, extend, or contradict what was known?
- Acknowledge limitations. What constrains the interpretation?
- Point to implications and future research. What should practitioners or researchers do next?
Not every paper does all four in that order, but nearly every strong Discussion includes all four moves. Skipping any one weakens the section.
Move 1: Restate the Key Finding
Open the Discussion by telling the reader — plainly — what the headline result was. Don't re-report the numbers; summarize their meaning.
Example: "This study found that participants in the mindfulness condition reported significantly lower state anxiety than those in the waitlist control, with a medium effect size."
One or two sentences is enough. The reader just read the Results section; they don't need a second full tour.
Move 2: Compare to Prior Work
Place your finding in the literature. Three common relationships:
- Consistent with prior work. "This is consistent with Smith (2022), who found similar reductions in anxiety following a brief mindfulness intervention."
- Extends prior work. "Prior research has focused on clinical populations; this study extends the finding to a non-clinical undergraduate sample."
- Contradicts prior work. "In contrast to Lee (2023), who reported no effect, this study found a significant reduction. Possible reasons include the longer intervention duration and the younger sample."
Name the studies, don't just gesture at "prior research." Reviewers check.
Move 3: Acknowledge Limitations
Every study has limitations. Stating them is not weakness — it's credibility. Cover three common types:
- Sample limitations. Single site, narrow demographic, self-selection.
- Method limitations. Self-report only, cross-sectional, short follow-up.
- Generalizability. Findings may not apply to other populations or settings.
Example: "Several limitations warrant consideration. First, the sample was drawn from a single large public university, limiting generalizability to other institutional types. Second, anxiety was measured by self-report, which may be subject to social desirability bias. Third, the four-week follow-up does not capture long-term effects."
A limitations paragraph takes 4-6 sentences. Longer than that, and the paper starts sounding apologetic.
Move 4: Implications and Future Research
Close with where this points. Two directions:
- Practical or theoretical implications. "These findings suggest that brief mindfulness interventions may be a cost-effective addition to college counseling services, particularly in under-resourced settings."
- Future research directions. "Future research should examine longer-term effects, whether dose-response relationships exist, and whether findings replicate in more diverse samples."
Keep future directions grounded. "More research is needed" alone is filler. Name what research, on what question, in what population.
What Not to Put in the Discussion
A Discussion is not a dumping ground. Leave out:
- New results. If a number wasn't in Results, it doesn't belong in Discussion.
- Uncited claims about the field. Every comparison needs a reference.
- Repetition of the literature review. The reader read it; synthesize, don't re-summarize.
- Unsupported speculation. "This may suggest a paradigm shift" — unless you have evidence, cut it.
- Conclusions that contradict the data. If the effect was small, don't claim a large impact. If the p-value was marginal, don't treat it as definitive.
Example: Discussion Paragraph by Paragraph
A condensed example of the four moves in flow, from a psychology intervention study:
Paragraph 1 (Finding): "This study examined whether a four-week mindfulness intervention reduced state anxiety in undergraduates. Participants in the mindfulness condition showed significantly lower post-test anxiety than those in the waitlist control, with a medium effect size."
Paragraph 2 (Prior work): "These results align with Smith (2022) and Garcia (2021), both of whom reported anxiety reductions following brief mindfulness interventions in clinical samples. The current study extends this finding to a non-clinical undergraduate population, suggesting mindfulness may have preventive as well as therapeutic value."
Paragraph 3 (Limitations): "Several limitations warrant consideration. The sample was drawn from a single university, and anxiety was assessed by self-report at a single post-test. The study did not include a longer-term follow-up, leaving open whether the effect persists beyond four weeks."
Paragraph 4 (Implications): "These findings suggest that brief mindfulness interventions may be feasible additions to college mental health resources. Future research should examine longer-term effects, optimal dose, and whether findings replicate in more diverse student populations."
Four paragraphs, roughly 250 words, each doing one clear job.
Results solid but the blank Discussion section is still staring at you? PaperDraft gives you a structured first draft — IMRaD skeleton, Discussion scaffolding across the four moves in academic register — so you can spend your time interpreting your data instead of formatting. It's a drafting assistant, not a submission. Try PaperDraft — free
Common Mistakes in Discussion Sections
A few errors sink Discussion sections reliably.
Overclaiming. A small effect on 80 undergraduates does not "revolutionize" a field. Calibrate the claim to the evidence.
Underclaiming. The opposite problem. Real findings get buried in hedges. If the effect is significant and meaningful, say so — then add the appropriate caveats.
Restating Results verbatim. If your Discussion opens with "Table 2 shows a mean of 14.2," you're still in Results. Summarize the meaning, not the number.
No limitations section. Reviewers read this paragraph first. Missing it signals inexperience or overclaim.
Generic "future research is needed." Specify the research. Which question? Which population? Which method?
No connection to prior work. A Discussion without citations is an opinion piece, not a research paper.
Sneaking in conclusions not supported by the data. If you tested anxiety but write about "overall well-being," the Discussion has drifted from what you measured.
How a Drafting Assistant Fits
A drafting tool can scaffold the Discussion's four-move structure — paragraph stubs, transition language between moves, and consistent academic register. What it can't do is interpret your specific findings honestly, know which prior studies your result actually fits or contradicts, or judge whether your claim size matches your evidence. PaperDraft handles the scaffolding. You handle the interpretation, the comparison to literature, and the limitations that only you know.
FAQ
How long should the Discussion section be?
Usually 15-25% of the paper's total length. For a 4,000-word research paper, expect 600-1,000 words.
Can the Discussion and Conclusion be the same section?
Some journals combine them; some separate them. If combined, the Conclusion is a final paragraph summarizing the take-home message. If separate, the Conclusion is a shorter standalone section after the Discussion.
Should I include unexpected findings in the Discussion?
Yes, but frame them honestly. Don't call an exploratory finding a primary result. Distinguish pre-registered hypotheses from post hoc observations.
How many citations belong in a Discussion?
Typically 10-25 for an undergraduate paper, more for a thesis. Every comparison to prior work needs a citation.
What if my hypothesis was not supported?
Report the null result honestly, then discuss why. Possible reasons: underpowered sample, measurement issues, the prediction was wrong. Null findings are part of the evidence base. See null hypothesis examples for how to frame non-significant results.
Once the Discussion lands, the paper has a thesis-level payoff. For the next section — deciding what goes in the Appendix and what doesn't — see research paper appendix.