The discussion chapter is where your dissertation stops reporting what happened and starts explaining what it means. Your results chapter presented the findings as plainly as possible; the discussion takes those same findings, answers each research question in turn, sets your evidence against the published literature, and shows what your study adds. Markers look here for independent judgement more than anywhere else, so a discussion that merely repeats the results in different words will lose marks even if the underlying research is sound. Written well, this chapter is often the longest analytical section after the literature review — typically around 20 to 25 per cent of the total word count, so roughly 2,000 to 2,500 words in a 10,000-word dissertation — and it is the place where a competent project becomes a convincing one.
Interpret, do not restate: the core discipline
The single most common weakness in discussion chapters is summary dressed up as analysis. Writing “Questionnaire respondents rated flexible working at 4.2 out of 5” belongs in the results. The discussion version asks why that score came out that way, what it suggests about the people surveyed, and whether it changes anything you believed at the outset. A useful test is to read each paragraph and ask whether it contains a verb of interpretation — suggests, indicates, implies, challenges, supports, complicates. If a paragraph could be moved into your results chapter without looking out of place, it does not yet belong in the discussion.
Work through your research questions one by one and give each a clear verdict. If your first question asked whether remote working affected self-reported productivity among call-centre staff, open that section with a direct answer: productivity rose for experienced staff and fell for new starters, for example, then build the reasoning underneath. Markers should never have to hunt for your answer. State it, support it with the strongest pieces of evidence, acknowledge the evidence that points the other way, and only then move on.
Use numbers sparingly here. Select the two or three figures that carry each argument rather than repeating every percentage from the results.
Comparing your findings with the literature
A strong discussion puts your findings into conversation with the studies you reviewed earlier, and it does so in three distinct moves: confirming, contradicting and extending. Most students manage the first move well. If your finding that peer support predicted retention matches earlier work, say so plainly, name the studies, and note where your context was similar. Agreement is valuable — it strengthens confidence in both sets of results — but agreement alone will not earn high marks, because it asks little of you as a researcher.
Contradiction is where the analytical work happens. Suppose previous research in large hospitals found that handover checklists reduced errors, while your small clinic study found no reduction. Do not hide that clash or apologise for it. Work through the plausible explanations in order: differences in setting and scale, differences in how the checklist was introduced, differences in measurement, and the possibility that the earlier finding does not travel to your context. Each explanation you test against your own data raises the quality of the chapter. Markers reward a student who can explain a disagreement far more than one whose results conveniently match everything already published.
Extension is the third move. Your study may confirm a known pattern but in a population nobody had examined — part-time students, night-shift workers, family-run firms — or add a mechanism the literature had only guessed at. Make that addition explicit: “These findings extend earlier work by showing the effect holds among zero-hours staff, a group absent from the samples in the studies reviewed.”
Handling unexpected results honestly
Almost every dissertation produces at least one result the researcher did not predict, and how you treat it says a great deal about your maturity as a researcher. The honest approach has three steps. First, report the surprise without softening it: if you expected training to improve confidence and it did not, write that clearly. Second, check the boring explanations before reaching for interesting ones — a small sample, a leading question, a coding error, a change in circumstances halfway through data collection. Third, offer your best reasoned interpretation and label speculation as speculation.
What you must not do is quietly drop an inconvenient finding, reframe a failed hypothesis as though it had been your intention all along, or invent a confident explanation unsupported by anything in your data. Examiners spot all three quickly, and each damages trust in the rest of the chapter. An unexpected result, discussed candidly, frequently becomes the most memorable part of a dissertation. A null result is still a result: discovering that an intervention made no measurable difference in your setting is useful knowledge for the next researcher and for any practitioner tempted to spend money on it.
Implications, limitations and recommendations that hold together
Implications answer the question “if my findings are right, what follows?” Split them into two kinds. Theoretical implications concern the ideas in your field: perhaps your evidence supports one model of motivation over another, or suggests a concept needs refining for smaller organisations. Practical implications concern what people should do differently: a ward manager, a school leader or a marketing team reading your work should finish the section knowing which of their current habits your evidence supports and which it questions. Keep each implication tethered to a specific finding. Grand claims about transforming an entire sector, built on interviews with nine people, will undermine your credibility rather than raise it.
Limitations need the same balance, handled without self-sabotage. Every study has limits — sample size, a single site, self-reported data, a cross-sectional design that captures one moment. Name the two to four limitations that genuinely affect how far your conclusions can travel, explain the direction of the risk (does this limit likely inflate or deflate the effect you found?), and state what you did to reduce it where you could. Avoid two traps: the apology list, where eight weaknesses in a row make the whole project sound worthless, and the token limitation, such as “time was limited”, which applies to every student project ever written and tells the marker nothing. One well-analysed limitation is worth more than five listed without thought.
Recommendations for future research should grow directly out of those limitations and surprises. If your single-site design limits generalisability, propose the multi-site comparison. If an unexpected subgroup difference appeared, propose the study that would test it properly. Vague suggestions such as “more research is needed with a bigger sample” earn nothing; a specific design, population and question show you understand your own project deeply enough to improve on it.
Stating your contribution and avoiding the classic mistakes
Near the end of the chapter, pull the threads together into a short, explicit statement of contribution. Undergraduate and masters contributions are usually modest, and that is entirely normal: a new context, a combined method, a tested intervention in an understudied setting, or evidence that challenges a comfortable assumption all count. Write two or three sentences you could defend in a viva without blushing. For example: “This study contributes evidence from independent retailers, a setting missing from research dominated by large chains, and shows that the loyalty effects reported elsewhere weaken sharply where customers know staff personally.” Modest, specific and defensible beats sweeping every time.
The mistakes markers mention most often are predictable. The first is the results repeat, covered above. The second is introducing brand-new literature in the discussion that never appeared in the review, which looks like padding; a small number of new sources to explain a surprise is acceptable, a parallel review is not. The third is overclaiming from small samples, especially using words such as “proves”. The fourth is a discussion organised by questionnaire item or interview theme rather than by research question, which forces the marker to assemble your argument for you. Finally, many students write the discussion in a rush at the end, so if you need structured support with sequencing and depth, our dissertation chapter help page explains how individual chapters can be strengthened without rewriting your whole project.
Frequently asked questions
How long should a discussion chapter be?
As a guide, allow around 20 to 25 per cent of your total word count — roughly 2,000 to 2,500 words in a 10,000-word dissertation, or 3,000 to 3,750 words in a 15,000-word project. Always check your own course handbook, because some programmes set chapter-by-chapter limits and empirical projects in the sciences often run shorter discussions than social science projects.
Can I include new references in the discussion?
Yes, in moderation. Most of your comparisons should draw on the literature review, but bringing in one or two new sources to explain an unexpected result is normal and often sensible. If your discussion relies heavily on studies the reader has never met, the balance is wrong and the review probably needs revisiting instead.
Should the discussion and conclusion be separate chapters?
That depends on your university’s template. Many programmes combine them into one “Discussion and conclusion” chapter; others expect a short separate conclusion that summarises the contribution without new analysis. Follow your handbook, and if the chapters are combined, keep the concluding section brief rather than repeating the discussion a second time.
What if my results contradict my hypothesis completely?
Report it plainly, explore the methodological and contextual explanations, and treat the contradiction itself as a finding. A well-discussed negative result can score highly because it demonstrates honest, critical research. What loses marks is disguising the outcome or pretending the hypothesis was something else.
In what order should I write the discussion?
Follow your research questions in the order you introduced them. Answer each one, compare it with the literature, then move on. Finish with implications, limitations, recommendations and your contribution statement, so the chapter narrows from detailed interpretation to the wider meaning of the study.
The discussion is where markers decide whether you simply carried out a project or genuinely understood it. Give it the drafting time it deserves, answer your questions head-on, and be candid about what your evidence can and cannot support. If you want broader guidance across every stage of the project, from proposal to final chapter, our dissertation help page sets out the support available at each step.