The discussion section is often considered the most important part of a research paper because it most effectively demonstrates your ability as a researcher to think critically about an issue, to develop creative solutions to problems based on the findings, and to formulate a deeper, more profound understanding of the research problem. Yet it is consistently the section that PhD students and early-career researchers find hardest to write.
Many researchers stare at a blank page after completing their results section, unsure where to start, what to include, and how to distinguish what they are supposed to write in the discussion from what they already wrote in the results. The pressure to interpret your findings correctly — without overclaiming, without underclaiming, and without simply restating what you already reported — is real and significant.
This guide will show you exactly how to write a discussion section for a research paper or thesis — covering what the discussion is, what it must achieve, its essential components, a step-by-step writing process, before-and-after examples, discipline-specific considerations, and the most common mistakes to avoid.
What Is the Discussion Section of a Research Paper?
The discussion section is where you delve into the meaning, importance, and relevance of your results. It should focus on explaining and evaluating what you found, showing how it relates to your literature review and research questions, and making an argument in support of your overall conclusion.
The discussion is fundamentally an act of interpretation — not description. Your results section described what you found. Your discussion section explains what it means.
Think of the discussion as the intellectual centrepiece of your paper. In the introduction you told readers why the research was needed and what you set out to investigate. In the results you showed them what you found. In the discussion you complete the intellectual loop — you explain what those findings mean, why they matter, how they advance understanding in your field, and what should happen next as a result.
The discussion is NOT a second results section.
One of the most common mistakes researchers make is presenting a superficial interpretation of the findings that more or less re-states the results section of the paper. Obviously, you must refer to your results when discussing them, but focus on the interpretation of those results and their significance in relation to the research problem — not the data itself.
The difference between results and discussion writing is the difference between reporting and reasoning:
- Results: “Supervision quality was significantly positively associated with PhD completion likelihood (β = 0.43, p < .001).”
- Discussion: “The strong positive association between supervision quality and completion likelihood suggests that the supervisory relationship functions as a primary protective factor against doctoral attrition — a finding that challenges the dominant institutional narrative that completion is primarily determined by student characteristics rather than the quality of academic support received.”
The results tell the reader what happened. The discussion tells the reader what it means.
The Purpose of the Discussion Section
Before writing a single word of your discussion, be clear about what it must accomplish. The discussion section serves five distinct purposes:
1. Answer the research question. Your discussion must directly address whether and how your findings answer the research questions you posed in the introduction. Readers — and examiners — return to the introduction when reading the discussion to check alignment. If your research questions are not explicitly addressed, your discussion is incomplete.
2. Contextualise findings within existing literature. Your findings do not exist in isolation. They confirm, contradict, extend, or complicate what previous researchers have found. The discussion shows how your results fit — or do not fit — within the existing body of knowledge. This is where your engagement with the literature review pays off.
3. Explain unexpected or surprising results. If any of your findings were unexpected — if a hypothesis was not supported, if a result contradicted a well-established finding, or if an unusual pattern emerged in the data — the discussion is where you address this honestly and offer a credible explanation.
4. Acknowledge limitations. No study is perfect. A credible discussion acknowledges the limitations of your research — constraints in sample size, methodology, generalisability, or measurement — without being excessively self-critical. Addressing limitations builds credibility by showing you have thought carefully about the boundaries of what your findings can and cannot support.
5. Suggest implications and future directions. The discussion closes by stating what your findings mean for theory, practice, or policy — and what questions remain unanswered that future research should address.
The Five-Part Structure of a Strong Discussion Section
A strong discussion section follows a clear, logical structure regardless of discipline. Think of your discussion as an inverted pyramid — organise the discussion from the general to the specific, linking your findings to the literature, then to theory, then to practice where appropriate.
Part 1 — Opening statement: summarise key findings
Begin the discussion by briefly restating your key findings — without reproducing the statistical detail of the results section. This opening gives readers a clear reference point before you move into interpretation.
The opening should be one to three paragraphs. State what you found in plain language. Focus on the most significant findings — the ones that most directly answer your research questions. You are not summarising every finding; you are establishing the interpretive foundation for the discussion that follows.
Example opening: “This study found that supervision quality was the strongest predictor of PhD completion likelihood among the sample, accounting for 31% of variance in completion outcomes. Three specific supervisory practices — weekly structured meetings, written feedback within seven days of submission, and proactive career guidance — were uniquely associated with timely completion. These findings suggest that the quality of the supervisory relationship, rather than student characteristics alone, is the primary determinant of doctoral completion.”
This opening is specific about what was found, uses plain language rather than statistical notation, and immediately signals the interpretive direction of the discussion.
Part 2 — Interpretation: what do your findings mean?
The interpretation section is the intellectual heart of the discussion. For each major finding, you must ask and answer: what does this mean? Why did this result emerge? What does it tell us about the phenomenon under investigation?
Interpretation requires you to move beyond the data and engage in genuine intellectual reasoning. You are not simply noting that a relationship exists — you are explaining why it exists, what mechanisms underlie it, and what theoretical implications it carries.
Strong interpretation language:
- “These findings suggest that…”
- “This result indicates…”
- “One possible explanation for this pattern is…”
- “This finding is consistent with the theoretical proposition that…”
- “The data imply that…”
Notice the use of hedging language — “suggests”, “indicates”, “implies” — rather than absolute language like “proves” or “demonstrates”. Research findings support or challenge conclusions; they rarely prove them definitively. Appropriate hedging is not weakness — it is intellectual honesty.
Part 3 — Contextualisation: how do findings relate to existing literature?
Once you have interpreted each major finding, contextualise it by showing how it relates to the existing research you reviewed in your literature review. Does your finding confirm a well-established result? Does it challenge or contradict it? Does it extend existing knowledge into a new context, population, or time period?
This is the section where you show that you understand the conversation your field is having — and that your research contributes to that conversation meaningfully.
There are four relationships your findings can have with existing literature, each requiring a different discussion approach:
Confirming existing findings: Your results replicate or support what previous studies found. Discuss why this consistency matters — does it validate the finding across a new context? Does it strengthen the evidence base?
“The finding that supervision quality predicts completion outcomes is consistent with Johnson and Smith’s (2023) longitudinal study of doctoral attrition in UK universities, suggesting that this relationship holds across different national higher education contexts.”
Contradicting existing findings: Your results differ from or contradict what previous studies found. This is actually the most intellectually interesting scenario — explain why the discrepancy might exist. Is it methodological? Contextual? Does it suggest the previous finding was overstated?
“Contrary to Brown et al.’s (2022) finding that student resilience was the primary predictor of completion, the current study identified supervision quality as the stronger predictor. This discrepancy may reflect differences in institutional context — Brown et al. studied students at research-intensive universities with well-resourced doctoral support structures, while the current study examined institutions with less formalised doctoral programmes.”
Extending existing findings: Your results apply a previously established finding to a new context, population, or variable — broadening the scope of what is known.
“While previous research has established the supervision-completion relationship in Western research universities, the current study extends this finding to the Indian university context for the first time, suggesting the relationship is cross-culturally robust.”
Filling a gap: Your results address a question that existing research had not investigated — they are not confirming or contradicting but adding genuinely new knowledge.
“No previous study had examined the specific supervisory behaviours associated with timely completion. The current study’s identification of three specific practices — structured weekly meetings, rapid written feedback, and career guidance — represents a new contribution to understanding the mechanisms through which supervision affects doctoral outcomes.”
Part 4 — Limitations and alternative interpretations
In my experience working with PhD students and early career researchers, they are often excessively critical of their work in the limitations section. Being honest about limitations is important — but the goal is accuracy, not self-flagellation. State each limitation clearly, explain why it exists and why it does not invalidate your findings, and suggest how future research could address it.
What to include in the limitations section:
Sample limitations — size, representativeness, and access constraints. If your sample was limited by institutional access or response rates, acknowledge this and discuss its effect on generalisability.
Methodological limitations — constraints in your research design, measurement instruments, or analysis approach. If self-report data is vulnerable to social desirability bias, acknowledge it. If cross-sectional design prevents causal claims, state this.
Scope limitations — the deliberate boundaries of the study (which you defined as delimitations in the methodology) that affect what conclusions can be drawn.
What NOT to include in limitations:
- Limitations that genuinely undermine the validity of your findings — if a methodological flaw is serious enough to invalidate the results, that is a research integrity issue, not a limitation to briefly acknowledge
- Limitations shared by virtually all research in your field — mentioning that a qualitative study is not generalisable is not informative if qualitative generalisability was never claimed
- An exhaustive catalogue of every possible weakness — focus on the two or three most significant limitations relevant to your specific findings
Addressing alternative explanations:
For your most significant findings, briefly acknowledge and address the most plausible alternative explanation. This demonstrates intellectual rigour — you have considered whether your interpretation is the only or most credible one.
“While the current study interprets the supervision-completion relationship as causal, an alternative interpretation is that high-quality supervisors may preferentially select or retain higher-ability students, creating a confound between supervision quality and student characteristics. Future research using longitudinal designs with random assignment to supervisory conditions would be needed to establish causation.”
Part 5 — Implications and future research
The final component of the discussion answers the “so what?” question. Your findings mean something — for theory, for practice, for policy, or for future research. This is where you articulate that meaning explicitly.
Theoretical implications address how your findings contribute to, extend, or challenge existing theoretical frameworks. What does the field now understand differently as a result of your research?
Practical implications address how your findings should inform practice — what should supervisors, doctoral coordinators, institutions, or policymakers do differently in light of what you found?
Future research directions identify the questions your study raises but does not answer. What would the next logical study investigate? What methodological approaches would address the limitations you have acknowledged?
Keep this section focused — two or three implications are more persuasive than a long list. Specific implications (“doctoral programmes should implement mandatory weekly supervision meetings with a standardised feedback protocol”) are more useful than vague ones (“institutions should improve supervision quality”).
Step-by-Step: How to Write Your Discussion Section
Step 1 — Return to your research questions before writing anything
Open your introduction and read your research questions again before you begin the discussion. Think of your objectives as the North Star guiding your discussion — they should influence what you emphasise, what you spend time on, and what you mention only briefly. For each point you plan to discuss, ask yourself: “How does this relate back to my research objectives?” If the connection is weak, consider whether it belongs in your discussion or might be better suited to a brief mention in your limitations or future research sections.
Step 2 — List your key findings before writing prose
Before writing a single paragraph, make a bullet list of your three to five most significant findings — the ones that most directly answer your research questions. For each finding, write one sentence interpreting what it means. This list becomes the structural skeleton of your discussion.
Finding 1: Supervision quality = strongest predictor (β = 0.43)
→ Meaning: supervisory relationship is primary protective factor against attrition
Finding 2: Three specific supervisory behaviours predict completion
→ Meaning: general “good supervision” can be operationalised into specific actions
Finding 3: Student demographics not significant predictors
→ Meaning: challenges student-deficit explanations of attrition
Unexpected finding: Part-time students showed higher completion rates than full-time
→ Meaning: needs explanation — possibly greater intrinsic motivation or clearer goals
This planning exercise prevents the most common discussion-writing trap: writing descriptively rather than analytically because you are not sure what point you are making.
Step 3 — Structure your discussion using temporary headings
Transform each key point you have identified into a temporary heading. Use italics and your word processor’s heading styles to create a clear structure. This approach helps you visualise the flow of your discussion and ensures you do not miss important points.
Once the full discussion is written, these temporary headings can be removed or converted into proper subheadings depending on your target journal or thesis requirements.
Step 4 — Write interpretation before contextualisation
For each finding, write your interpretation first — what you think the finding means — before going to your reference list to find supporting citations. This preserves your original analytical voice and prevents the common pattern of writing a discussion that is almost entirely “as Smith found… as Jones argued… consistent with Brown…” without any of your own interpretive reasoning.
Your interpretation comes first. Literature comes second — as evidence that supports, challenges, or contextualises your interpretation.
Step 5 — Address unexpected or negative results specifically
If any findings were unexpected or if any hypotheses were not supported, these must be addressed specifically and honestly in the discussion. Unexpected results are not failures — they are often the most intellectually interesting part of a study. A credible explanation for why an expected result did not emerge is more valuable than a list of findings that all confirmed predictions.
Step 6 — Write the limitations section honestly but proportionately
State your two to three most significant limitations, explain why each exists, note what it means for the conclusions you can draw, and suggest how future research could address it. Each limitation should be addressed in three to four sentences — enough to show you have thought about it seriously, not so much that it dominates the discussion and undermines confidence in the study’s value.
Step 7 — End with implications — not just limitations
Many researchers end their discussion with the limitations section, leaving readers with a focus on what the study could not do. This is a weak ending. Always close the discussion with implications and future directions — ending on what the research contributes, what it means for the field, and what it opens up for future inquiry. This is a much stronger and more memorable closing.
Before and After Examples
The following examples show weak and strong versions of a discussion paragraph for the same finding.
Finding: Students who received weekly supervision meetings had significantly higher completion rates than those who met monthly (85% vs 61%, p < .001).
Weak discussion paragraph: “The results showed that students who had weekly supervision had higher completion rates than those with monthly supervision. This is an important finding. Previous studies have also found that supervision is important for PhD completion. These results suggest that supervision matters and that universities should think about how to improve supervision practices for their doctoral students. More research is needed in this area.”
Why it is weak: It re-states the result rather than interpreting it. The citation is vague (“previous studies have found”). The implication is generic (“universities should think about how to improve”). The closing sentence is a placeholder, not a genuine direction.
Strong discussion paragraph: “The significantly higher completion rates among students receiving weekly supervision (85% vs 61%) suggests that regularity of contact — not merely the quality of individual interactions — is a critical structural feature of effective doctoral supervision. This finding extends Johnson et al.’s (2023) finding that supervision quality predicts completion by identifying frequency of contact as one of its key operational components. An explanatory mechanism may lie in accountability and momentum: weekly meetings create consistent short-term deadlines that prevent the motivational drift and task avoidance commonly associated with the extended, self-directed nature of doctoral study (Brown & Williams, 2024). These findings have direct implications for doctoral programme design — specifically, the need for formalised minimum supervision frequency requirements rather than leaving contact frequency to individual supervisor-student negotiation. This is particularly relevant in institutional contexts where informal norms of infrequent supervision persist unchallenged.”
Why it is strong: It interprets the finding rather than restating it. It identifies a specific mechanism (accountability and momentum). It engages critically with existing literature. It states a specific, actionable implication.
The Discussion Section in Different Research Contexts
Journal article discussion
Typically 600 to 1,500 words depending on the journal. Usually covers all five components of the structure above but in a more compressed format. Check your target journal’s recent publications for typical length and structure — conventions vary significantly between journals and disciplines.
Some journals combine the discussion and conclusion into a single final section. Others — particularly in medicine and health sciences — keep discussion and conclusions separate. Follow your target journal’s format exactly.
PhD thesis discussion chapter
In a PhD thesis, the discussion chapter is typically 8,000 to 15,000 words — much longer than a journal article discussion because it must address the full complexity of a multi-year doctoral study. It may be organised around the research questions, around the theoretical framework, or thematically.
The PhD thesis discussion must demonstrate doctoral-level thinking — going beyond description and surface interpretation to engage critically with theory, methodology, and the broader implications of the research. Examiners specifically look for evidence of independent scholarly judgment in the discussion chapter.
Qualitative research discussion
Qualitative discussions follow the same five-part structure but with important differences in content and language. Findings are expressed as themes or categories rather than statistics. Contextualisation involves showing how themes relate to theoretical frameworks and prior qualitative work in the field. Limitations address issues of trustworthiness, credibility, and transferability rather than statistical validity and generalisability.
Be careful not to use inappropriate language in qualitative discussions — avoid phrases like “the data proved” or “the results demonstrated” which imply statistical certainty your methodology cannot support. Use language consistent with your epistemological position: “the themes suggest”, “participants described”, “the analysis revealed patterns consistent with”.
Language and Style for the Discussion Section
Use hedging language appropriately
Academic discussions require careful hedging — language that signals the degree of certainty with which claims are made. This is especially important in the discussion where you are making interpretive claims that go beyond what the data directly shows.
Strong hedging phrases for discussion writing:
- “These findings suggest that…”
- “This pattern may reflect…”
- “One possible explanation is…”
- “The data are consistent with the interpretation that…”
- “This finding tentatively supports…”
- “Further research would be needed to confirm…”
Avoid both under-hedging (“this proves that…”) and over-hedging (“it might perhaps be possible that this could potentially suggest…”). The goal is appropriate epistemic modesty — not paralysis.
Maintain present tense for interpretation
Use present tense when interpreting and discussing findings: “These results suggest that…” not “These results suggested that…” Results reported in the results section use past tense. Interpretations and implications in the discussion use present tense because they describe relationships and meanings that exist in the present, not just at the time of data collection.
Use first person where appropriate
Many disciplines now permit and encourage first-person writing in the discussion section. “We interpret this finding as evidence that…” or “I argue that this pattern reflects…” are clear, confident, and intellectually honest. Check your target journal’s style guidelines — if first person is permitted, use it in the discussion where it is most natural.
Connect discussion paragraphs explicitly
Each discussion paragraph should connect to the one before and the one after. Use transitional phrases to signal shifts in focus:
- Moving to a new finding: “Turning to the second research question…”
- Moving to limitations: “Despite these contributions, several limitations must be acknowledged…”
- Moving to implications: “The foregoing analysis has several implications for theory and practice…”
- Moving to future research: “These findings open several directions for future investigation…”
Common Mistakes in Discussion Sections
Restating the results. The most universal mistake. The discussion interprets — it does not repeat. If you find yourself writing “the results showed that X…”, ask yourself: what does X mean? Why did X occur? What does X imply? Those are discussion questions. “The results showed X” is a results statement.
Ignoring results that did not support hypotheses. Researchers often unconsciously downplay or omit findings that were unexpected or that contradicted their hypotheses. This is a form of selective reporting that undermines research integrity. All significant findings must be addressed in the discussion — particularly unexpected ones, which are often the most theoretically interesting.
Overclaiming beyond what the data supports. Do not use exaggerated claims about the significance of your research. For example, do not state that a correlation equals causation. As another example, do not generalise findings beyond the sample selected for your study. A cross-sectional survey cannot establish causation. A study of 50 students at one university cannot generalise to all PhD students globally. Match your claims precisely to what your methodology and sample support.
Writing an exhaustive limitations section. Listing every possible limitation of your study is not rigorous — it is excessive. Focus on the two or three limitations most relevant to your findings. A limitation that does not affect your conclusions does not belong in the discussion.
Failing to connect findings to the research questions. Every significant finding discussed must be connected back to the research questions posed in the introduction. Readers should be able to read your discussion and understand clearly whether and how each research question was answered.
Generic implications. Implications like “this research will contribute to the growing body of literature on doctoral education” are meaningless. State specifically what your research contributes — what new knowledge it adds, what theory it extends, what practice it should change, what policy it informs.
Starting the discussion with a methods summary. Beginning the discussion with a paragraph summarising what the study did — “the purpose of this study was to…” followed by a full description of the methodology — is an unnecessary repetition of the introduction and methods sections. The discussion starts with findings, not methods.
Introducing new data. Do not mention new data that has not been presented in the earlier chapters. Do not just list your findings. Synthesise them in a way that conveys a general takeaway message for your readers. The discussion interprets data already presented — it does not introduce new results or cite new empirical studies not already referenced in the literature review.
How Long Should the Discussion Section Be?
Length varies significantly by discipline, paper type, and publication venue. The following guidelines apply broadly:
Journal article: 600 to 1,500 words. Some short communications have discussions of 300 to 400 words. Some theoretical papers have longer discussions of 2,000 words. Check your target journal.
PhD thesis discussion chapter: 8,000 to 15,000 words. Must address the full complexity of the research comprehensively. Considerably longer than a journal discussion because it engages with theory, methodology, and implications in greater depth.
Research report or dissertation (Master’s level): 1,000 to 3,000 words depending on total word count.
Conference paper: 300 to 600 words — highly compressed.
A useful check: your discussion should be approximately 25 to 35% of the total paper length. If your discussion is shorter than your results section, it is probably not sufficiently interpretive. If it is significantly longer, it may be including material that belongs elsewhere.
Frequently Asked Questions
What is the difference between the results section and the discussion section? The results section presents your findings objectively — what you found, expressed in data, statistics, themes, or observations. The discussion section interprets your findings — what they mean, why they emerged, how they relate to existing knowledge, and what implications they carry. Results describe; discussion explains. The results section uses past tense and avoids interpretation. The discussion section uses present tense for interpretation and actively engages with theory and literature.
Should I include citations in the discussion section? Yes — the discussion should cite relevant prior studies when contextualising your findings. When confirming an existing finding, cite the studies you are confirming. When contradicting a prior finding, cite the study you are contradicting. When proposing a theoretical explanation, cite the theoretical work that supports it. However, unlike the literature review, the discussion should not be dominated by citations — your own analytical voice and interpretation should be the primary thread, with citations providing supporting evidence.
How do I discuss unexpected or negative results? Unexpected results should be addressed honestly and with genuine intellectual engagement. First, acknowledge clearly that the result was unexpected or that the hypothesis was not supported. Second, offer one or two credible explanations for why this occurred — methodological constraints, contextual factors, or theoretical reasons why the expected relationship may not hold in your specific setting. Third, discuss what the unexpected finding means for the field — it may be more theoretically significant than the expected findings. Avoid minimising or glossing over unexpected results.
Can the discussion section be combined with the conclusion? In some disciplines and journals, the discussion and conclusion are combined into a single section. In others — particularly in medicine, health sciences, and many social science journals — they are kept separate. A combined section typically flows from interpretation and contextualisation into implications and a final summative conclusion. If combining, ensure that the conclusion element is clearly identifiable — usually a final paragraph or two that synthesises the overall contribution and makes a clear closing statement.
How do I start the discussion section? Begin with a brief summary of your most significant finding — not with a restatement of the research purpose or a summary of the methodology. The most effective opening immediately establishes the interpretive direction of the discussion. For example: “The central finding of this study — that supervision quality accounts for 31% of variance in doctoral completion outcomes — challenges institutional frameworks that attribute non-completion primarily to student characteristics.” This opening immediately signals what the discussion will argue and why it matters.
What tense should I use in the discussion section? Use past tense when referring to what your study did or found: “The analysis revealed…”, “The survey showed…” Use present tense when interpreting findings and their implications: “These results suggest…”, “This finding indicates…”, “The evidence implies…” Use present tense when referring to the conclusions of cited studies: “Smith (2023) argues that…”, “Brown et al. (2024) demonstrate…” This tense convention is standard across most disciplines.
Conclusion
The discussion section is where your research comes alive. It is the section that transforms data into knowledge — where statistical relationships and qualitative themes become contributions to the field’s understanding of the world.
Writing a strong discussion requires three things: clarity about what your findings actually mean, courage to interpret boldly while hedging appropriately, and the discipline to connect every interpretive claim back to your research questions, your literature, and the boundaries of what your methodology can support.
The five-part structure — opening summary, interpretation, contextualisation, limitations, and implications — provides a reliable framework that works across disciplines and research designs. Master this structure, and you will never again face a blank page at the start of your discussion section.
Published by EaseWrite — writing made easy for PhD scholars and researchers worldwide.