The discussion chapter is where a dissertation stops reporting and starts arguing: what the findings mean, how they sit beside published work, and how far they can be trusted. It is the discussion chapter guide of How to write a dissertation, from EdCitation, and it follows the literature review and methodology guides in the same series.
The advice comes from six universities' own library and writing pages, read on 26 September 2026: Southampton, Sheffield Hallam, Westminster and Manchester in the UK, the University of Southern California (USC) in the US, and UNSW Sydney in Australia. They agree on the purpose and part company on the details, and we say where. None sets a length for the chapter; your handbook does, if anyone does.
A discussion is built from comparisons with other people's studies, so its references carry weight. EdCitation finds those studies with Find sources, builds each entry with Cite a source and checks the finished list with Verify references. Nothing in EdCitation writes a word of the chapter itself.
What is the discussion chapter of a dissertation for?
To say what your findings mean for the research question and for what was already known. University of Southern California Libraries (2026) defines the discussion as interpreting the findings against prior knowledge of the problem, carrying the reader beyond where the literature review left them.
Southampton puts the test as one question: what do we know now that we did not know before? Its guide expects the discussion to hold the dissertation's most original insight (University of Southampton Library, 2026). Sheffield Hallam University Library (n.d.) adds a second job: judging how reliable the findings are.
How is the discussion different from the results and the conclusion?
The results report, the discussion interprets, and the conclusion sums up. The universities draw the lines slightly differently.
| Chapter | Its job | Where the guides differ |
|---|---|---|
| Results or findings | What the data show, without interpretation | UNSW allows a brief comment on a key result; Manchester says do not interpret yet |
| Discussion | What the results mean, set against the literature | Southampton pictures the results as the tip of an iceberg and the discussion as the part under water |
| Conclusion | A short summary of the answer and what it adds | UNSW: no new information; Manchester puts limitations and future research here |
Results and discussion may also be one chapter. University of Westminster Library (2025b) suggests separate chapters for scientific and quantitative dissertations, and calls a combined analysis and discussion common in qualitative work, where themes are presented and interpreted together. Manchester leaves the choice to you but asks you to explain it (University of Manchester, 2020). In a qualitative thesis the first half is often called the findings chapter.
What goes in a discussion chapter, and in what order?
A common order runs from the question to the future: restate, answer, compare, explain, apply, qualify. Sheffield Hallam and USC give lists close to this one; treat it as a default that your department's guidance overrides.
| Part | What it does | Where the guides differ |
|---|---|---|
| 1. The question restated | The problem, the method in a sentence, the main finding | USC: one direct paragraph |
| 2. Key findings | Each research question or hypothesis answered in turn | Westminster: say plainly whether the hypothesis was supported |
| 3. Interpretation against the literature | Where findings agree or disagree with earlier studies, and why | Manchester needs little literature here; USC expects the most citations after the review |
| 4. Unexpected findings | Stated, then a reasoned explanation | USC allows opening the chapter with one |
| 5. Alternative explanations | Readings other than the one you prefer | Sheffield Hallam and USC both ask for them |
| 6. Implications | What the findings mean for theory, practice or policy | UNSW adds practical applications |
| 7. Limitations | What limits how far the findings reach | Southampton, USC: here or in the conclusion; Manchester: the conclusion |
| 8. Recommendations | Further research tied to the limitations | Southampton: few, and not repeated in the conclusion |
Where the order changes
USC keeps the discussion in the order the results were reported, except that a striking or unexpected finding may come first; Sheffield Hallam puts the most important findings first. Pick one and say so in the opening paragraph.
Southampton shapes each theme like a small essay: a topic sentence answering part of the question, a body that brings in the literature, and an evaluative close that answers "So what?"
Limitations after the research, not before
A methodology chapter names the limits you planned for; the discussion names the ones the data revealed. Southampton draws that line, giving the example of generalisability that only becomes clear once the data are analysed, and warns against overstressing weaknesses. Sheffield Hallam asks for limitations without apology; Manchester calls them evidence that you have thought critically, not weaknesses. A recommendation for further research belongs next to the limitation it answers, and USC warns against recommending work you could have done within the study itself.
How do I relate the findings back to the literature review?
Return to the studies your literature review set up, and compare your findings with them directly. Southampton and Sheffield Hallam both describe the discussion as revisiting the key sources from the review, to show where your findings fit, where they conflict and what is new (University of Southampton Library, 2026; Sheffield Hallam University Library, n.d.).
The guides disagree on how much literature belongs. Manchester's chapter-by-chapter tips say you need little, because most of the chapter is your own interpretation of your results (University of Manchester, 2020). USC says the discussion usually holds more references than any section except the literature review, mostly recent ones used to interpret a finding (University of Southern California Libraries, 2026). Both are right about different failures: Manchester guards against a second literature review, USC against findings discussed in a vacuum.
A practical rule that satisfies both: cite a study in the discussion only when it changes how a finding should be read. A source that agrees, one that disagrees, and the reason for the difference usually cover a finding. Twelve sources making the same point do not.
Adding a source you did not review
Sometimes a finding, usually an unexpected one, needs literature the review never mentioned. USC allows studies to be saved for the discussion when they matter more as comparisons than as background. If a new source becomes central, add a sentence about it to the literature review too, so the reader meets it before it carries weight.
A source that disagrees with you stays in. USC says to explain why your findings differ, and Southampton warns that ignoring contradictory data undermines your credibility. The research gap your review identified is the yardstick: say how far your findings close it.
How cautious should the claims in a discussion be?
As cautious as the evidence, and no more. Hedging means wording a claim to show how certain it is. Manchester's Academic Phrasebank notes that discussion sections are especially rich in hedging, above all where writers explain their findings (University of Manchester, n.d.), and Southampton asks for it wherever a result could be read another way.
USC adds the other half: read no more into the findings than the evidence supports, and label speculation as speculation. Over-hedging fails the other way, making a clear result sound unsure.
Words that match the evidence
This table is our own guide, drawn from the phrasings the Phrasebank lists; no university prescribes it.
| What the evidence is | Wording that fits | Wording to avoid |
|---|---|---|
| An experiment in which the method was assigned | "The results indicate that…" | "This proves…" |
| A correlation or a survey | "X was associated with Y", "may reflect" | "X causes Y" |
| A small or single-site sample | "In this sample…", "may not extend to…" | "Students prefer…" |
| A result that is not statistically significant | "No reliable difference was found" | "There is no difference" |
| Qualitative themes | "Participants described…", "suggests" | "Nurses feel…" |
Report the size of an effect, not only whether it passed a test: an effect size with its confidence interval shows how large the difference might plausibly be. For qualitative work, write about transferability rather than statistical generalisation.
What does a discussion paragraph look like?
Below is a paragraph of our own, written for this guide. The student's finding is invented for the example; the two published studies and every figure attributed to them are real, read in the papers themselves.
Imagine a dissertation that surveyed second-year students on how they took lecture notes and compared that with their marks on a test's conceptual questions.
This study found that students who reported taking lecture notes by hand scored somewhat higher on conceptual questions than those who typed them. The finding is consistent with Mueller and Oppenheimer (2014), whose three experiments found that students taking notes on laptops did worse on conceptual questions than those writing by hand. It sits less easily with a preregistered direct replication of their first study, in which 142 students showed no reliable difference on conceptual questions (Hedges' g = −0.13, 95% CI [−0.45, 0.20]), and whose mini meta-analyses of similar studies found the longhand advantage on conceptual questions small and not statistically significant (g = 0.14, 95% CI [−0.07, 0.35]; Urry et al., 2021). Because the present study relied on self-reported note-taking and did not assign students to a method, the difference may reflect who chooses to write by hand rather than the pen itself. The finding therefore suggests, rather than shows, a longhand advantage, and a study that assigned the method would be needed to test it.
What the paragraph does
It states the finding in one sentence, then sets it beside an agreeing study and a disagreeing one, with the disagreeing study's own figures. It names a rival explanation, confounding by who chooses which method, and states the limitation that allows it. It ends with a claim worded to fit, and a recommendation tied to that limitation.
It also shows why the comparison must be checked at source. Urry et al. (2021) report that the original Study 1 had 65 participants and a conceptual-question effect of g = 0.34 with an interval running from −0.16 to 0.83, and that they used updated data files posted with Mueller and Oppenheimer's 2018 correction. Comparing a result with the 2014 headline alone would set it beside a stronger claim than the later evidence supports.
How do I write the discussion chapter, step by step?
In this order, starting before a sentence is drafted.
- Read your handbook or module guide for what the chapter must contain and where limitations go. Check your paper turns those instructions into a free checklist and quotes the line behind every rule.
- List your research questions or hypotheses and write a one-sentence answer to each.
- For each answer, pick the studies from your literature review that it agrees or disagrees with, and reread the passages you rely on.
- Where a finding needs a study the review lacks, search for the claim in Find sources, and add the study to the review if it becomes central.
- Build each new entry from its DOI in Cite a source, which also screens it for retraction.
- Draft each finding as topic sentence, comparison and evaluation; add unexpected findings and alternative explanations.
- Write the implications, the limitations the data revealed, and a few recommendations tied to them.
- Reread every claim against its evidence and adjust the wording.
- Run the reference list through Verify references, then the whole draft through References from a file, before the chapter goes to your supervisor.
How EdCitation helps with a discussion chapter
EdCitation is, in our view, the best tool for a discussion chapter's references, because each entry it returns comes from the publisher's record, never composed, and no chatbot can give that guarantee. The worked paragraph above rests on which study came first and what was later corrected; a misdated or invented reference would put that comparison on the wrong footing.
We ran Find sources on 26 September 2026 for "laptop versus longhand note taking", sorted by citations. It returned 443 works, led by Morehead, Dunlosky and Rawson's 2019 replication in Educational Psychology Review, cited 107 times, and Luo, Kiewra, Flanigan and Peteranetz's 2018 study in Instructional Science, cited 98 times. The original's record in the search index carries only its main title, without the words laptop and longhand, so it was not near the top; searching "the pen is mightier than the keyboard" put it first, cited 1,039 times, with its 2018 correction listed below it. In our run a short phrase worked better than a whole claim sentence, which returned a noisier list.
Cite a source built the original from its DOI, in a rerun on 27 September 2026:
Mueller, P. A., & Oppenheimer, D. M. (2014). The pen is mightier than the keyboard: Advantages of longhand over laptop note taking. Psychological Science, 25(6), 1159–1168. https://doi.org/10.1177/0956797614524581
The record gives the title in title case, and Cite a source set it in sentence case, as APA 7 wants. Beside the entry, the tool listed the correction published in 2018, with its link.
We then gave Verify references three entries: that one, Morehead and colleagues' replication with 2017 as its year, and a reference we made up. Verified, for the first: "The DOI resolves to this record and the title matches." Doubtful, for the second: "A record with this title exists, but it was published in 2019, not 2017." Not found, for the made-up one: "No publisher's or registry's record matches this reference." Where an index does not answer, the tool says could not check, never not found.
Verify references is free with no account. References from a file takes the whole draft and matches every in-text citation to the list, which catches a source added late to the discussion and never listed; a free account gets three checks a month, and Pro, at $8 a month, removes the cap. Theoretics QA, part of Max at $24 a month (see pricing), weighs the draft against your own handbook's instructions, quoting your text and rewriting nothing. None of these tools writes any of your discussion, and none can promise a mark.
Quick questions
What is the difference between the results and the discussion chapter?
The results chapter reports what the data show, without interpretation; the discussion explains what those results mean and compares them with published studies. Some departments combine the two, which Westminster describes as common in qualitative dissertations.
Should limitations go in the discussion or the conclusion?
Either, depending on your department. Southampton and USC accept both places, and Manchester's tips put them in the conclusion; wherever they go, they should be the limitations your data revealed, not a copy of the methodology chapter's.
Can I cite new sources in the discussion chapter?
Yes, when a finding needs one, often an unexpected result. USC allows studies to be saved for the discussion; if a new source becomes central, mention it in the literature review too.
What tense is a discussion chapter written in?
Mostly the present for what the findings mean and for established knowledge, and the past for what specific earlier studies found. USC's guide gives that split.
Can EdCitation check the references in my discussion chapter?
Yes. EdCitation's Verify references checks each reference against the publisher's record, free, and References from a file matches every in-text citation to the list.
References
- Mueller, P. A., & Oppenheimer, D. M. (2014). The pen is mightier than the keyboard: Advantages of longhand over laptop note taking. Psychological Science, 25(6), 1159–1168. https://doi.org/10.1177/0956797614524581
- Sheffield Hallam University Library. (n.d.). Discussion. Dissertations and research projects. https://libguides.shu.ac.uk/researchprojects/discussion
- University of Manchester. (n.d.). Being cautious. Academic Phrasebank. https://www.phrasebank.manchester.ac.uk/using-cautious-language/
- University of Manchester. (2020). Tips for writing your dissertation chapter by chapter [PDF]. https://documents.manchester.ac.uk/display.aspx?DocID=50610
- University of New South Wales. (n.d.). How to write an honours thesis. UNSW Current Students. https://www.unsw.edu.au/student/managing-your-studies/academic-skills-support/toolkit/honours-postgraduate
- University of Southampton Library. (2026, June 18). Results and discussion. Writing the Dissertation: Guides for Success. https://library.soton.ac.uk/writing_the_dissertation/results_discussion
- University of Southern California Libraries. (2026, September 8). 8. The discussion. Organizing Your Social Sciences Research Paper. https://libguides.usc.edu/writingguide/discussion
- University of Westminster Library. (2025a, October 1). Discussion. Dissertations 5: Findings, Analysis and Discussion. https://libguides.westminster.ac.uk/c.php?g=696975&p=5004236
- University of Westminster Library. (2025b, October 1). Dissertations 5: Findings, analysis and discussion. https://libguides.westminster.ac.uk/Findings-Analysis-Discussion
- Urry, H. L., Crittle, C. S., Floerke, V. A., Leonard, M. Z., Perry, C. S., III, Akdilek, N., Albert, E. R., Block, A. J., Bollinger, C. A., Bowers, E. M., Brody, R. S., Burk, K. C., Burnstein, A., Chan, A. K., Chan, P. C., Chang, L. J., Chen, E., Chiarawongse, C. P., Chin, G., … Zarrow, J. E. (2021). Don't ditch the laptop just yet: A direct replication of Mueller and Oppenheimer's (2014) Study 1 plus mini meta-analyses across similar studies. Psychological Science, 32(3), 326–339. https://doi.org/10.1177/0956797620965541