Video: How to Write the Discussion Section: Examples & Expert Tips

Want to write the perfect discussion section in your research paper? In this video, Dr. Daniel Wasser provides a succinct 6-step guide for you!  

Contents

• What Is a Discussion Section?
• What to Include in the Discussion: A Quick Overview
• Step 1: Summarize Your Key Findings
• Step 2: Interpret Your Results
• Step 3: Compare With Existing Literature
• Step 4: Discuss Implications
• Step 5: Acknowledge Limitations
• Step 6: Suggest Future Directions and Provide a Take-Home Message
• Discussion Section vs. Conclusion: Key Differences
• Quick-Reference Checklist for an Effective Discussion Section
• Frequently Asked Questions

The discussion section is the heart of your research paper: it’s where you stop reporting and start thinking. Yet it’s also the section researchers struggle with most. Unlike the results section, which presents findings objectively, the discussion asks you to interpret, contextualize, and argue for the significance of your work.

This guide (extending the above video) walks you through six clear steps, with examples from multiple disciplines, common mistakes to avoid, and annotated samples so you can see exactly what good looks like.

What Is a Discussion Section?

The discussion section explains what your results mean, not just what they are. It connects your findings to the broader literature, addresses limitations honestly, and tells readers why your work matters.

Results Section vs Discussion: Key Differences

What to Include in the Discussion: A Quick Overview

Before diving into each step, here’s what a strong discussion section covers:

• ✅ A brief summary of your key findings
• ✅ Your interpretation of what those findings mean
• ✅ Comparison with existing literature
• ✅ Implications for theory, practice, or policy
• ✅ Honest acknowledgment of limitations
• ✅ Suggestions for future research
• ✅ A clear take-home message

What NOT to Include

• ❌ New data or results not mentioned earlier
• ❌ Simple repetition of the results section
• ❌ Overconfident claims unsupported by your data
• ❌ Excessive self-criticism that undermines your study
• ❌ Vague statements like “more research is needed” without specifics

Step 1: Summarize Your Key Findings

Open your discussion by briefly restating the research problem and your central findings. This anchors the reader and creates a bridge from the results section. Keep this to one concise paragraph: you are not repeating your results, you are distilling them into a clear statement that directly answers your research question.

How to Do It

• Restate the research objective in one sentence
• Summarize the most important finding (not all findings)
• Directly answer your research question

Sentence Starters

“This study investigated… and found that…” “The central aim of this research was to… The results indicate that…” “Consistent with our hypothesis, the data demonstrate that…” “Contrary to expectations, the analysis revealed that…”

Annotated Examples by Field

Medicine / Clinical Research

“This study examined the effect of a 12-week resistance training program on HbA1c levels in adults with Type 2 diabetes. The results indicate a statistically significant reduction of 0.8% in HbA1c compared to the control group.”

🔍 Annotation: The opening immediately restates the intervention and outcome measure. It answers the research question directly: no preamble, no hedging.

Environmental Science

“This study assessed microplastic concentration in three coastal estuaries over two tidal cycles. Findings demonstrate that microplastic load was significantly higher during ebb tide, suggesting terrestrial runoff as the dominant input pathway.”

🔍 Annotation: The summary includes the mechanism (runoff), not just the finding: this signals interpretation is coming and draws the reader forward.

Social Sciences / Education

“This qualitative study explored how first-generation university students experience academic identity. Across all 24 interviews, participants described a persistent sense of ‘not belonging,’ regardless of academic performance.”

🔍 Annotation: For qualitative research, the “finding” is a theme, not a number. The summary captures the core theme concisely.

Step 2: Interpret Your Results

This is the most intellectually demanding part of the discussion. Interpretation means explaining why your results look the way they do: not just restating what they show.

What Interpretation Looks Like

Common Mistakes in Interpretation

• Confusing correlation with causation: if you have observational data, avoid causal language (“X causes Y”); use “X is associated with Y” instead
• Over-interpreting non-significant results: a p-value above 0.05 is not proof that no effect exists
• Ignoring unexpected findings: reviewers notice when inconvenient results are glossed over; address them directly

Annotated Examples by Field

Psychology

“The significant positive correlation between sleep duration and working memory capacity (r = 0.61, p < .001) is consistent with consolidation theory, which posits that slow-wave sleep facilitates the transfer of short-term memories to long-term storage. Notably, this relationship remained significant even after controlling for age and anxiety levels, suggesting it is not an artifact of demographic confounding.”

🔍 Annotation: The interpretation cites a named theory, references the statistical result without re-reporting it, and proactively rules out a confound: exactly what reviewers want to see.

Engineering / Materials Science

“The reduction in tensile strength observed at elevated curing temperatures was unexpected. One plausible explanation is accelerated polymer chain scission above 80°C, a phenomenon documented in analogous epoxy systems by Vasilovitch et al. (2019). This interpretation is supported by the SEM micrographs showing increased surface microfractures in high-temperature samples.”

🔍 Annotation: The author flags the surprise, offers a mechanistic explanation, cites prior work, and uses their own additional data (SEM images) to corroborate. This is robust interpretation.

Step 3: Compare With Existing Literature

Your findings don’t exist in isolation. This step situates your work within the broader conversation in your field: showing what you confirm, what you challenge, and what you add.

Three Moves to Make Here

• Confirm: Show where your results align with prior work and explain why this convergence matters
• Contrast: Where your results diverge from prior studies, explore why: different populations, methods, contexts?
• Contribute: What does your study add that didn’t exist before?

Sentence Starters

“These findings are consistent with those of [Author, Year], who observed that…” “In contrast to [Author, Year], the present study found…, which may be attributable to differences in sample composition.” “While previous research has focused on X, this study extends that work by examining Y.” “To our knowledge, this is the first study to demonstrate…”

Annotated Example

Public Health

“The 23% reduction in emergency department visits following the community health worker intervention aligns with findings from Kumar et al. (2021), who reported a 19% reduction in a comparable urban setting. However, our effect size was larger, possibly reflecting our program’s additional focus on medication adherence coaching: a component absent from the Kumar et al. protocol. Unlike Brennan & Osei (2020), who found no effect in rural cohorts, our rural subgroup showed comparable gains to the urban sample, suggesting the intervention may be more scalable than previously assumed.”

🔍 Annotation: This paragraph makes all three moves: confirms (Kumar), explains the difference, and challenges a prior conclusion (Brennan & Osei). It positions the study’s contribution clearly without overstating it.

Step 4: Discuss Implications

Implications answer the “so what?” question: why should anyone outside your immediate research team care about these results? There are two types:

How to Frame Implications Without Overstating

• Use hedged language: “These findings suggest…”, “This may indicate…”, “One implication is…”
• Be specific about who the implications apply to
• Distinguish between what your data supports directly vs. what it implies more loosely

Annotated Examples by Field

Education Policy

“These findings have direct implications for how schools allocate reading support resources. The strong association between early phonological awareness and reading fluency at age eight suggests that phonological screening in kindergarten (rather than waiting for reading failure to emerge) could allow for earlier, lower-cost intervention. School districts with limited resources may prioritize phonological training in pre-K programs as a cost-effective preventive strategy.”

🔍 Annotation: The implication is specific (kindergarten screening), actionable (districts can act on it), and realistic (cost language anchors it in real-world constraints).

Economics / Behavioral Science

“From a policy perspective, the finding that default enrollment significantly increased retirement savings participation: from 34% to 89%: supports the use of opt-out rather than opt-in program designs in public pension schemes. This aligns with nudge theory and suggests that administrative design choices may be more powerful levers than financial incentives in promoting savings behavior.”

🔍 Annotation: Connects to a named theory (nudge theory), provides a concrete policy lever, and makes the magnitude of the effect vivid with real numbers.

Step 5: Acknowledge Limitations

Limitations are not a confession of failure: they are a demonstration of scientific integrity and self-awareness. Reviewers and readers trust authors more when limitations are addressed clearly and without defensiveness.

What to Address

• Sample size or composition: Was your sample small, non-representative, or restricted to a specific demographic?
• Study design: Observational designs cannot establish causation; cross-sectional designs cannot assess change over time
• Data collection constraints: Self-reported data, short follow-up periods, measurement tools with known reliability issues
• Generalizability: Can your findings be applied to populations or contexts beyond your sample?
• Confounding variables: Were there variables you could not control for?

How to Frame Limitations Constructively

Annotated Example

Epidemiology

“Several limitations should be considered when interpreting these results. First, data were collected via self-report, which may introduce recall bias, particularly for dietary intake variables. Second, the study’s cross-sectional design prevents causal inference: the observed association between ultra-processed food consumption and depressive symptoms does not establish that diet causes depression. Third, the sample was drawn from a single urban health region and may not reflect patterns in rural or lower-income populations. Nonetheless, the large sample size (n = 4,217) and the consistency of findings across demographic subgroups strengthen confidence in the observed associations.”

🔍 Annotation: Three limitations are named precisely (not vaguely), but the final sentence pivots to strengths: maintaining credibility without undermining the study. This is the gold standard.

Step 6: Suggest Future Directions and Provide a Take-Home Message

Your discussion should not end with limitations. Close by pointing forward: what research should come next, and what is the single most important thing you want readers to carry away?

Future Directions: Be Specific

Vague suggestions like “future research should explore this further” are almost useless. Instead, specify:

• What should be studied (the variable, population, or mechanism)
• How it should be studied (the design or method)
• Why it matters (the gap this would fill)

❌ “More research is needed in this area.”

✅ “Future longitudinal studies should track HbA1c over 24 months and include participants from lower-income groups, who were underrepresented in the current cohort, to determine whether effects persist and generalize across socioeconomic strata.”

The Take-Home Message

This is your last sentence or short closing paragraph. Make it count. It should be:

• Clear: no jargon
• Specific: tied to your actual findings
• Memorable: the one sentence a conference audience would write down

Annotated Examples by Field

Neuroscience

“These results open promising avenues for non-pharmacological interventions in early-stage Alzheimer’s disease. Future work should examine whether the cognitive gains observed here persist beyond six months and whether they translate to functional independence in daily tasks: a question with direct implications for caregiver burden and healthcare costs. Ultimately, this study demonstrates that structured cognitive engagement can meaningfully slow hippocampal volume loss, reinforcing the brain’s plasticity as a therapeutic target even in the context of neurodegeneration.”

🔍 Annotation: Future direction is specific (six-month follow-up, functional outcomes) and tied to real-world stakes (caregiver burden). The take-home message names the mechanism (hippocampal volume loss) and the therapeutic principle (neuroplasticity).

Ecology

“Longitudinal monitoring across additional river systems, particularly those with varying agricultural intensity, would clarify whether the microplastic dynamics documented here are regionally specific or represent a broader pattern. Conservation practitioners should consider tidal cycle timing when designing sampling protocols, as ebb-tide sampling may significantly overestimate baseline pollution levels. In sum, tidal hydrology is not merely a sampling consideration: it is a substantive driver of microplastic distribution that ecosystem models must incorporate.”

🔍 Annotation: Practical implication for practitioners (sampling timing), a specific future research direction (agricultural intensity gradient), and a punchy take-home that reframes the finding as a modelling principle.

Discussion Section vs. Conclusion: Key Differences

A common source of confusion is the difference between the discussion and conclusion section; here’s how to keep them distinct:

Quick-Reference Checklist for an Effective Discussion Section

Use this before submitting your manuscript:

• Does my opening paragraph directly answer my research question?
• Have I interpreted why my results look the way they do: not just what they show?
• Have I compared my findings with at least 2–3 prior studies?
• Have I articulated both theoretical and practical implications?
• Have I named specific limitations without being self-defeating?
• Have I offered concrete (not vague) future research directions?
• Does my closing paragraph contain a clear, jargon-free take-home message?
• Have I avoided introducing any new data?
• Have I avoided causal language where I only have correlational data?

Frequently Asked Questions

1. How long should a discussion section be?

There is no universal rule, but as a general guide, the discussion section is typically 15–25% of your total manuscript length. In a 5,000-word paper, that is roughly 750–1,250 words. For a PhD dissertation chapter, it may run several thousand words. Journals often specify word limits: always check author guidelines. Err on the side of depth over brevity; a thin discussion is one of the most common reasons for peer review rejection.

2. Can I use the first person (“I” or “we”) in the discussion section?

Yes, and in many disciplines it is now preferred. Phrases like “We interpret this as…” or “We argue that…” are clearer and more direct than passive constructions such as “It can be argued that…” Check your target journal’s style guide: some fields (particularly in the humanities and social sciences) strongly favor first person, while certain clinical journals still use passive voice by convention.

3. What is the difference between a discussion section and a results section?

The results section reports your findings objectively: what the data show, supported by statistics, tables, and figures. The discussion section interprets those findings: what they mean, why they matter, and how they relate to existing knowledge. A useful rule: if you are describing a number or a pattern, it belongs in results. If you are explaining what that number means, it belongs in the discussion. In some qualitative research traditions, the two sections are integrated, but in quantitative research they are usually kept strictly separate.

4. Should I address every single result in the discussion?

No. Focus on your most significant, surprising, or theoretically important findings. Trying to discuss every result leads to a bloated, unfocused discussion that buries your key contributions. Results that are largely confirmatory and unsurprising can be acknowledged briefly (“As expected, X was associated with Y, consistent with prior work”). Reserve your analytical depth for findings that are novel, unexpected, or especially consequential.

5. How do I write a discussion section when my results were not statistically significant?

Null or non-significant results are legitimate and valuable scientific contributions. In your discussion, consider: (a) whether the null result is a true null or a consequence of insufficient power (small sample size); (b) what the confidence intervals suggest about the plausible range of effect sizes; (c) whether the result challenges an existing assumption in the field; and (d) what conditions might produce a different result. Frame the non-significant finding as informative, not as a failure: many journals now actively seek well-powered null results as a corrective to publication bias.

6. How do I avoid overlap between the Discussion and the Introduction?

The Introduction and Discussion are mirror images of each other but they face opposite directions. The Introduction moves from broad context to your specific research question; the Discussion moves from your specific findings back out to broader significance. Overlap happens when authors either re-explain background theory in the Discussion or preview conclusions in the Introduction.

Here’s how to keep them distinct:

The key distinction on literature: You may cite the same papers in both sections but for different purposes. In the Introduction, you cite Smith (2021) to show what was known before your study. In the Discussion, you cite Smith (2021) to show how your results confirm, contradict, or refine what Smith found. The reference is the same; the analytical move is entirely different.

A practical test: Read each sentence in your Discussion and ask “Does this sentence require knowledge of my results to make sense?” If yes, it belongs in the Discussion. If it could have been written before you collected any data, it likely belongs in the Introduction (or nowhere at all).

Common overlap traps to avoid:

• ❌ Re-explaining a theory or concept in the Discussion that was already defined in the Introduction. Simply apply it, don’t re-define it
• ❌ Restating the research gap in the Discussion as if the reader has forgotten it. A brief callback is fine (“This study addressed the gap identified by…”), but don’t re-argue for why the gap mattered
• ❌ Saving a key piece of background literature for the Discussion because it “fits better there”. If it’s foundational context, it belongs in the Introduction; if it’s a direct comparator to your findings, it belongs in the Discussion.