You just finished collecting your data, running your analyses, and writing up your results. Now comes the section that makes most students want to cry: the discussion.
Here’s the truth — the discussion section isn’t a summary of your results. It’s where you answer the single most important question a reader cares about: “So what?”
When you write a discussion section well, you transform raw data into a coherent argument. You show your reader why your findings matter, how they fit into the scholarly conversation, and what they mean beyond your specific study. That’s the difference between a paper that sits on a professor’s desk and one that gets cited, graded highly, and remembered.
- A strong discussion section has five core components: restate findings, interpret them, compare with existing literature, acknowledge limitations, and discuss implications.
- Use the inverted-hourglass model: start narrow (specific finding) and widen gradually (implications, theory, future research).
- After every paragraph, ask “so what?” — if you can’t answer in one sentence, rewrite it.
- Fill-in-the-blank templates and the roadmap statement technique turn vague interpretation into structured, defensible writing.
- Common mistakes include restating results instead of interpreting, overclaiming from correlational data, and ignoring contradictory findings.
Here’s a comprehensive guide that walks you through everything you need to write a discussion section that actually works — including templates, annotated examples, and the kind of practical tips that make the difference between a 70% and a 90% paper.
Where the Discussion Section Fits: The IMRaD/CARS Framework
Before you can write a discussion section, you need to understand where it lives. Most empirical research papers follow the IMRaD structure — Introduction, Methods, Results, and Discussion. This framework was developed by scientists and has spread across disciplines because it works. It provides a logical flow: introduce the problem, describe how you solved it, report what you found, and then explain what it means.
The CARS model (Create a Research Space) takes this further. It helps you understand the argument structure of a paper: establishing a territory (introduction and literature review), indicating a gap (your research question), and occupying the gap (your results and discussion). The discussion section is where you occupy the gap — you step into the scholarly conversation your literature review set up and show how your findings change things.
Here’s what most students don’t realize: the discussion section is the only section where you get to be interpretive. Everything before it is reporting or describing. Everything after it (conclusion, if your paper has one) is wrapping up. The discussion is your space to think out loud about what your findings actually mean.
When you write the discussion, you’re not just filling a section. You’re building the argument your paper was always heading toward. And that’s what makes it one of the most important parts of your research paper.
The 5-Part Discussion Section Structure
Every strong discussion section follows five core components. These aren’t rigid rules — they’re a framework that works across disciplines. You can reorder them based on what matters most in your study, but they should all be present.
Component 1: Restate the Major Findings
Your opening paragraph should state your most important results in plain, conceptual language. Not numbers. Not statistics. Words.
Think of it as an executive briefing for someone who will never read your results section. If they can understand your main finding without flipping back to check the numbers, you’ve done it right.
Template 1 — Restating Findings:
“This study set out to determine whether [research question]. Our findings demonstrate that [major finding in plain language].”
Template 2 — ThesisAI’s Mechanism Template:
“This suggests that [mechanism] and indicates that [broader implication].”
Here’s what that looks like in practice with a real quantitative example:
Strong example:
“Students who received structured peer feedback during collaborative writing assignments showed significantly higher critical thinking scores than those in the control group. These findings support our initial hypothesis that structured collaborative instruction enhances analytical reasoning skills.”
Weak example:
“As shown in Table 3, the peer feedback group scored M = 14.3, SD = 2.1, compared to the control group M = 12.1, SD = 2.4, t(38) = 2.87, p = .006.”
The weak example just repeats results. The strong example interprets them. Notice that the strong example doesn’t mention the t-statistic or p-value — that belongs in the results section.
Component 2: Interpret Your Results
This is the heart of the discussion. You explain what your findings mean and why they occurred. This is where you demonstrate scholarly thinking — moving from observation to insight.
When you interpret results, consider these angles:
- What theoretical framework do your results support?
- What mechanisms might explain your findings?
- Are there alternative explanations for the same data?
- What patterns emerged that you hadn’t anticipated?
Component 3: Compare with Existing Literature
A strong discussion doesn’t stand alone. It enters an ongoing scholarly conversation. Compare your findings with previous research — do they confirm, contradict, extend, or refine prior work?
Johnson PhD’s Roadmap Statement Technique
End your summary paragraph with a roadmap statement that orients your interpretation: “This study extends the literature in [N] ways” — and then structure your interpretations around that statement. This gives you a ready-made scaffold for the next three paragraphs.
If you just found something that contradicts previous research (which happens more often than you expect), frame it as an opportunity. Discrepancies aren’t failures — they’re potential contributions. Explain possible reasons: different sample populations, methodological differences, contextual factors, or measurement instrument variation.
Component 4: Acknowledge Limitations
Every study has limitations. Acknowledging them doesn’t weaken your paper — it strengthens credibility by demonstrating methodological awareness and intellectual honesty.
Common limitations include:
- Sample size (small or convenience samples)
- Sampling method (non-random or self-selected participants)
- Geographic scope (findings applicable only to specific populations)
- Measurement tools (reliability or validity concerns)
- Temporal factors (cross-sectional data limiting causal inference)
The key is balanced acknowledgment. Be honest but not apologetic. State the limitation, explain its potential impact, and suggest how it affects interpretation.
Component 5: Discuss Implications and Future Research
End with the broader impact of your study. Answer the reader’s question: “What should happen next?”
Discuss both theoretical and practical implications. How do your findings advance academic understanding? How might practitioners apply these results? What specific research questions should follow?
Future research recommendations should be concrete. Avoid “More research is needed.” Instead, propose specific studies — perhaps with different populations, methodologies, or variables.
The Inverted-Hourglass Pyramid Model
The University of Guelph’s writing center offers the best visual metaphor I’ve found for how a discussion should flow. Imagine an inverted hourglass — narrow at the top, widening as you go down.
┌─────────────────────┐
│ Summary of your │ — narrow opening: specific finding
│ main result │
├─────────────────────┤
│ Interpretation │ — widening: what it means, mechanisms
│ of findings │
├─────────────────────┤
│ Comparison with │ — widening: how it fits in literature
│ existing research │
├─────────────────────┤
│ Implications │ — wider: theoretical, practical
│ & future research │
└─────────────────────┘
You start narrow (restating your specific finding) and widen gradually (what it means, how it compares, what it implies). This structure prevents the common error of spiraling into abstraction too early. Your opening should be tight. Your ending should be expansive.
The “So What” Self-Check (thesisAI)
After every paragraph, write one sentence completing: “This matters because…” If you can’t answer it, that paragraph belongs in the results section, not the discussion. This single habit prevents the most common student mistake — describing instead of interpreting.
How to Write Each Component: Annotated Examples
Quantitative Discussion — Strong vs. Weak Example
Strong quantitative example (annotated):
“Regression analysis indicated that program length was significantly associated with research identity stability (β = 0.42, p < 0.01). [Restates finding in plain language] This result supports the theoretical prediction that extended immersion in academic communities strengthens researcher self-concept. [Interpretation — connects to theory] However, the interaction term between program length and funding type was not significant (β = 0.08, p = 0.23), suggesting that financial support does not moderate the identity development trajectory. [Addresses unexpected/null result] This contrasts with scholarship on academic socialization (Martinez, 2020), which found that funding status significantly shapes identity outcomes. [Compares with literature] The discrepancy may reflect sample differences: our participants were all graduate students in structured degree programs, whereas Martinez’s study included non-degree research assistants. [Explains discrepancy]”
Why this works: It moves smoothly from finding → interpretation → literature comparison → explanation. Every paragraph earns its place. The author asks “so what?” after each result and provides a clear answer.
Weak quantitative example (annotated):
“Table 2 shows the regression results. The program length variable had a beta of 0.42 with a p-value of less than 0.01. The funding interaction had a beta of 0.08. The R-squared was 0.34. These results are consistent with previous studies.”
Why this fails: Every sentence is a result restatement. There’s no interpretation. The phrase “consistent with previous studies” is a placeholder — it doesn’t name any specific studies or explain why they’re consistent. This reads like a results section, not a discussion.
Qualitative Discussion — Strong vs. Weak Example
Strong qualitative example (annotated):
“The data revealed three distinct patterns in how graduate students navigate interdisciplinary training. [Topic sentence] First, students described a period of epistemological tension during their first year, where exposure to conflicting methodological paradigms produced identity uncertainty. [Finding] This finding extends the identity development literature (Cross & Rogers, 2019) by demonstrating that disciplinary tension is not a barrier to growth, but a catalytic mechanism. [Interpretation + literature] Second, the transition from tension to integration occurred most frequently during structured peer mentorship — suggesting that social context mediates epistemological adaptation. [Finding + mechanism] Third, students who engaged in interdisciplinary fieldwork reported higher identity confidence than those who completed only coursework, indicating that embodied research experience strengthens theoretical identity. [Finding + implication] Taken together, these findings suggest that interdisciplinary training operates as an identity transformation process rather than a simple skill-building exercise. [Broad implication]”
Weak qualitative example:
“Theme 1 was tension. Students felt confused about their discipline. Theme 2 was mentorship. Theme 3 was fieldwork. These themes are important.”
Why this fails: It lists themes without interpretation. No literature comparison. No mechanism. No implication. This is a themes section, not a discussion.
Field-Specific Conventions
Your discussion looks different depending on your discipline. Here’s a practical comparison:
Sciences (Natural Sciences, Medicine, Engineering)
- Organize around hypotheses tested
- Reference specific statistical tests and effect sizes
- Use signal phrases: “Our findings indicate that…” (tentative), “Our results demonstrate that…” (stronger)
- Compare numerical findings with meta-analyses or prior studies
- Emphasize methodological rigor and replicability
- Follow NIH/PMC’s seven-element framework closely
Signal phrase guidance (University of Guelph):
- Strong evidence: “Our results demonstrate that…” or “We clearly found that…”
- Moderate evidence: “Our findings suggest that…” or “These results indicate that…”
- Tentative: “These data imply…” or “We propose that…”
Verb choice: Use “demonstrate” for strong causal claims (only with experimental designs), “suggest” or “indicate” for correlational data, “may explain” or “could account for” for mechanisms.
Social Sciences (Psychology, Education, Sociology)
- Organize around research questions or themes
- Use extended quotations or case excerpts as evidence
- Emphasize context and meaning over generalization
- Acknowledge researcher positionality and reflexive bias
- May include narrative summaries rather than hypothesis-driven framing
- Strong connection to theoretical frameworks
Humanities (History, Literature, Philosophy)
- Argument-driven rather than hypothesis-driven
- Deep engagement with primary sources and theoretical texts
- Less emphasis on empirical comparison; more on interpretation and argument
- Theoretical engagement with existing scholarship
- Often combines results and discussion into a single continuous section
Comparison Table: Discipline Approaches
| Aspect |
Sciences |
Social Sciences |
Humanities |
| Organization |
Hypotheses tested |
Research questions/themes |
Argument-driven |
| Evidence |
Statistical tests, effect sizes |
Extended quotes, themes |
Primary sources, theory |
| Language |
Signal phrases, tentative/strong |
Contextual, reflexive |
Interpretive, argumentative |
| Structure |
Separate results & discussion |
Sometimes combined |
Often combined |
| Comparison |
Meta-analyses, prior studies |
Theory engagement |
Scholarship dialogue |
Common Mistakes and How to Fix Them
Mistake 1: Restating Results Instead of Interpreting Them
The problem: You summarize raw data without moving to analysis. The examiner has already read the results section. They want “so what?” — not “what happened.”
The fix: After every finding, write one sentence completing “This matters because…” If you can’t answer it, the paragraph belongs in the results section.
Mistake 2: Introducing New Results in the Discussion
The problem: You present new findings that weren’t reported in the results section.
The fix: Any new finding discussed must be tied to data already presented. Go back to the results section and add it there first.
Mistake 3: Overclaiming from Correlational Data
The problem: Using “causes” or “leads to” when your design was cross-sectional or correlational. This is the most common overclaiming error.
The fix: Use “associated with,” “related to,” or “predicts” instead of “causes.” Reserve causal language for experimental or longitudinal designs.
Mistake 4: Ignoring Contradictory Findings
The problem: You suppress unexpected results out of embarrassment.
The fix: Unexpected results are valuable. Frame the discrepancy honestly: “Contrary to our hypothesis, …” Offer plausible explanations. Compare with prior literature. Avoid overinterpreting, but don’t hide.
Mistake 5: Using Generic Language for Limitations
The problem: “This study had some limitations.” That’s not a limitation — that’s a placeholder.
The fix: Be specific. “Our cross-sectional design prevents causal inference.” “Our convenience sample of undergraduate students limits generalizability to working professionals.” Name the limitation and its impact.
Mistake 6: Inflating Claims Beyond Your Data
The problem: “These results prove that…” or “This completely changes our understanding of…”
The fix: Use cautious, qualified language. “Suggests,” “indicates,” “may explain,” “appears to.” Absolute claims undermine credibility unless your study design truly warrants them.
Talandis (2025) Self-Assessment Rubric
JALT Publications (JALT 49.3, 2025) offers a practical six-point checklist you can use to grade your own discussion before submitting:
- Structural awareness: Does the discussion follow a clear logical progression from specific to general?
- Interpretive depth: Are findings interpreted beyond description, with theoretical connections?
- Literature integration: Are findings compared with at least 2-3 specific prior studies?
- Limitations acknowledgment: Are limitations named specifically without excessive self-criticism?
- Implications clarity: Are theoretical and practical implications stated clearly?
- Language precision: Is tense, voice, and hedging language consistent and appropriate?
If you score 4 or above on each point, your discussion is publication-ready. If you score below 3 on any point, revise that component before submitting.
How to Handle Unexpected or Contradictory Results
You expected X. You got Y. What do you write?
First, don’t panic. Unexpected results are common and valuable.
Second, don’t hide them. Suppressing unexpected findings creates dishonesty and wastes analytical opportunities.
Here’s how to handle them:
Frame the discrepancy honestly: “Contrary to our hypothesis, participants in the intervention group showed no significant improvement in [measure].”
Offer plausible explanations: Was your measurement instrument adequate? Could external factors have influenced results? Was the sample different?
Compare with prior literature: Check whether other researchers found similar unexpected results. If no prior work found the same thing, you may have something genuinely novel.
Avoid overinterpretation: Do not claim unexpected findings are conclusive. Use cautious language. Overinterpreting unexpected results is one of the most common mistakes novice writers make.
Word Count and Length Guidance
While length varies by discipline and journal, a common rule of thumb for undergraduate and graduate papers is 1,000 to 1,500 words (roughly 10-15% of total paper length). For a thesis or dissertation, discussions can run 3,000 to 5,000 words.
The key is substance over length. A concise, focused discussion outperforms a long, repetitive one every time.
Length guidelines by paper type:
- Undergraduate research paper: 800-1,200 words
- Graduate-level research paper: 1,000-1,500 words
- Journal article discussion: 1,000-2,000 words
- Master’s thesis discussion: 2,000-3,500 words
- PhD dissertation discussion: 3,000-5,000+ words
Focus on what matters. If you can say something in two sentences, don’t write it in eight.
Your Pre-Submission Discussion Checklist
Before you submit, run through this combined checklist (integrating Talandis 2025 rubric with practical checks):
- [ ] Did you state the major findings in plain language (no statistics) in the opening paragraph?
- [ ] Did you interpret the findings without merely repeating the results section?
- [ ] Did you compare your results with at least 2-3 prior studies?
- [ ] Did you address unexpected findings honestly?
- [ ] Did you acknowledge limitations without excessive self-criticism?
- [ ] Did you propose specific future research directions?
- [ ] Is your discussion focused on answering the research question?
- [ ] Did you avoid introducing new results?
- [ ] Is your tense consistent within paragraphs?
- [ ] Did you avoid inflated or overgeneralized claims?
- [ ] Is the discussion concise (typically 10-15% of total paper length)?
- [ ] Did you use cautious, qualified language throughout?
- [ ] Did you connect findings to theoretical framework or literature review?
Putting It All Together: Your Discussion Section Framework
Here’s a paragraph-by-paragraph framework you can adapt:
Paragraph 1: Summary of major findings, direct answer to research question (2-3 sentences)
Paragraph 2: Interpretation of first key finding in context of literature (2-3 sentences)
Paragraph 3: Interpretation of second key finding; comparison with prior studies (2-3 sentences)
Paragraph 4: Unexpected results or contradictory findings; possible explanations (2-3 sentences)
Paragraph 5: Theoretical and practical implications (2-3 sentences)
Paragraph 6: Limitations (1-2 sentences)
Paragraph 7: Future research recommendations; closing statement (1-2 sentences)
This framework scales. Add more paragraphs for multiple key findings. Merge or reorder based on what matters most in your study.
What We Recommend
Start with the templates. Don’t try to write a discussion section in one pass. Open your fill-in-the-blank templates (see below) and work through each component sequentially. The “so what” self-check after every paragraph is the single most effective habit for avoiding the results-description confusion.
When to go deep vs. broad: Use the inverted-hourglass model to decide. If your study has one high-impact finding, go deep on that one. If you have multiple equally important findings, broaden across them. Don’t spiral into abstraction before you’ve earned the right to do so — stay grounded in your data until the implications paragraph.
What to avoid: Don’t treat the discussion as a results summary. Don’t use absolute claims when your design doesn’t support them. Don’t ignore contradictory findings. Don’t write walls of text without subheadings or structure.
Related Templates (Quick Reference)
Template 1 — Opening Statement: “This study set out to determine whether [research question]. Our findings demonstrate that [major finding].”
Template 2 — Interpretation Bridge: “This suggests that [mechanism] and indicates that [broader implication].”
Template 3 — Literature Comparison: “These findings are consistent with [Author] (Year), who also observed [similar pattern]. However, our results diverge in that [specific difference], which may reflect [reason].”
Template 4 — Limitations: “This study was limited to [scope]. While findings are robust within this context, they cannot be generalized to [different population]. A [different methodology] would provide deeper insight.”
Template 5 — Roadmap Statement: “This study extends the literature in [N] ways: [Way 1], [Way 2], and [Way 3].”
Related Guides
Final Thoughts
The discussion section is where you demonstrate that you understand not just what you found, but why it matters. It’s where the scholarship begins — where your data enters the scholarly conversation.
Don’t treat it as a perfunctory obligation. Treat it as your best opportunity to show what you actually learned, to connect your work to the broader academic world, and to explain why anyone should care.
When you write the discussion, remember: interpret, don’t repeat. Compare, don’t isolate. Be honest, not apologetic. And above all, answer the question that every reader carries into your paper — “So what?”
Need help writing your research paper? Get professional academic writing assistance from expert writers who understand the discussion section inside and out. Order your custom paper today.
Stuck interpreting your results? Our academic writing team can help you turn raw data into a compelling discussion. Visit QualityCustomEssays.com for tailored support.
