Many authors find the results and discussion the hardest sections to write, and for a good reason: these are the sections where the paper must stop describing what was done and start showing what was found and what it means. The methodology can be written from notes, and the introduction from the literature, but the results and discussion require judgement — deciding which findings matter most, how to present them so a reader grasps them quickly, and how far the interpretation can honestly go.
This guide covers both sections together, because they work as a pair. The results section presents the findings without interpretation; the discussion section interprets them, connects them to previous work, and states their limits. Keeping that division clean is the single most important habit in writing these sections. This guide follows our article on writing the methodology section and builds on the abstract and introduction, which promise the findings these sections must deliver.
The results section reports what the study found. It does not explain why the findings matter, compare them with other studies, or speculate about causes — that is the discussion's job. A results section that keeps to this rule is shorter, clearer, and easier for reviewers to check against the methodology. When a sentence in the results starts to explain or praise a finding, it has crossed into discussion and should be moved.
Organise the results in the same order as the research questions or hypotheses, and in the same order as the methods that produced them. If the methodology described the sample, then the intervention, then the outcome measures, the results should follow that same sequence. Reviewers check results against methods, and a matching order makes that check easy. Begin with the sample itself: how many participants or cases were included, the response or completion rate, and the key characteristics of the sample. This grounds everything that follows.
Present each main finding once, in the form that suits it best. Text is best for a small number of simple findings; tables are best for detailed numerical results a reader may want to look up; figures are best for patterns, trends, and comparisons that are easier to see than to read. Do not present the same finding in text, table, and figure — choose one primary form and use text only to point the reader to the main pattern. Every table and figure needs a clear title, labelled rows and columns, and a note of the sample size and any statistical detail the reader needs to understand it without hunting through the text.
Numbers in the results must be complete and precise. Report the actual values, not only whether a result was significant: give effect sizes, confidence intervals, or equivalent measures alongside p-values where your field expects them, because significance alone says nothing about the size or importance of an effect. Report negative and null results with the same completeness as positive ones. A study that reports only its significant findings misleads the reader about what was actually found, and reviewers treat selective reporting as a serious problem.
Be careful with language around numbers. A result is not "proved" by a single study, a difference is not "large" unless the size supports that word, and a trend is not a finding unless the analysis supports it. Use the past tense, because the study is complete, and keep the wording neutral: "the mean score was higher in group A than in group B" is a result; "group A performed impressively better" is interpretation dressed as a result. Small choices of wording like this are where many results sections lose credibility.
For qualitative studies, the results section works differently but follows the same principle: present what was found, organised by theme, with enough quoted evidence for the reader to judge the interpretation that will follow in the discussion. Choose quotations that illustrate the theme, not quotations chosen because they are dramatic, and show the range of views in the data, including views that complicate the main pattern. A qualitative results section that reports only one voice is as selective as a quantitative section that reports only significant p-values.
The discussion is where you earn the reader's trust as an interpreter. It should open with a direct answer to the research question, stated in plain language: what did the study find, in one paragraph? This opening matters because many readers read the discussion first, and a discussion that opens with background or a restatement of the methods loses them. After the answer, work through the findings in order of importance, not in the order they appeared in the results, and explain each one: what does it mean, why might it have happened, and how does it compare with previous work?
Comparison with the literature is the heart of the discussion. For each main finding, show whether it agrees with, extends, or contradicts earlier studies, and offer a reasoned explanation for agreement or disagreement — a difference in population, method, setting, or time period, for example. Do not simply list previous studies; explain what the comparison reveals. Where your findings contradict earlier work, say so plainly and consider the explanations fairly, including the possibility that your own study has a weakness. A discussion that only cites supporting studies reads as advocacy, not scholarship, and reviewers notice it.
The discussion must also explain the study's contribution: what is now known that was not known before, and who can use it? This is where the paper connects back to the research gap identified in the introduction. If the introduction promised that the study would fill a particular gap, the discussion should show, explicitly, whether and how it did. A paper whose discussion never returns to its own gap feels unfinished, however good the results are.
Every study has limitations, and reviewers expect to see them stated. A limitations paragraph that is vague — "the sample was small" with no explanation of why that matters — is almost as bad as no limitations at all. Name each real limitation, explain how it could affect the findings, and say what was done to reduce its impact. A small sample matters differently in a pilot study than in a study claiming a general effect; a single study site matters differently for a local service evaluation than for a claim about a whole population. Proportion is the key: state the limitation, its likely effect, and its boundary, then move on.
Do not apologise for limitations, and do not list limitations that are not real. Limitations are not confessions; they are part of the study's description, and stating them well is a strength. Equally, do not use the limitations paragraph to introduce new results, to blame participants, or to claim that a limitation "does not affect the findings" without a reason. If a limitation genuinely undermines a finding, say so and qualify the conclusion accordingly — a qualified conclusion that survives review is worth far more than a bold conclusion that does not.
Unexpected results deserve careful handling, not concealment. If a finding contradicts the hypothesis or the literature, report it fully and discuss the plausible explanations, including methodological ones. Some of the most useful papers are those that explain why an expected result did not appear. Hiding an unexpected result, or explaining it away with a speculation presented as fact, damages the paper far more than the result itself ever could.
Close the discussion with a conclusion that follows from the results actually reported — no broader than the evidence allows, and no narrower than the contribution deserves. The conclusion should restate the main finding, its implication for practice or future research, and a concrete suggestion for what should be studied next. Avoid ending with a generic call for "more research"; name the specific question the next study should answer, and why this study makes that question the right one. Our guide on responding to reviewer comments is useful when reviewers challenge your interpretation, and our guide on why papers get rejected shows how over-claiming in the discussion leads to rejection.
Discussion paragraphs are easier to write when they start from a single, plainly stated result. Take this invented example: in a small teaching study, students who practised with a simulator scored a mean of 78 out of 100 on a handling-skills test, while students taught by lecture alone scored a mean of 71. The results sentence writes itself, and it stays neutral: students in the simulation group scored higher than students in the lecture group (78 versus 71, n = 60 per group). No interpretation, no praise, no explanation yet. That restraint is the whole job of the results section, and it is what makes the discussion trustworthy.
The discussion paragraph then builds on that one result in four moves. First, restate it in interpretive language: simulation practice was associated with stronger handling performance. Second, compare it with earlier work: a previous study of a similar intervention reported a smaller difference, which suggests the effect may depend on how much practice time students actually received. Third, offer a plausible explanation tied to your own data, not to general theory alone: the simulation group had repeated, supervised attempts, while the lecture group had none, so the gap is consistent with practice rather than with extra teaching time. Fourth, state the limit plainly: the study ran at one institution with a short follow-up, so the finding supports a larger trial rather than a change of curriculum on its own.
Use this four-move pattern for each main finding, and keep one rule in force throughout: the discussion must not introduce a new result. If a number appears in the discussion that was never reported in the results, move it back to the results section or cut it. When every discussion paragraph follows the pattern of claim, comparison, explanation, and limit, the section reads as an argument rather than a list, and the conclusion follows naturally: one specific next study, such as testing the same intervention across several institutions with a longer follow-up, rather than a vague call for more research.
| Feature | Results section | Discussion section |
|---|---|---|
| Job | Present what was found | Explain what it means |
| Order | Same order as methods and research questions | Order of importance, main finding first |
| Evidence | Numbers, tables, figures, quotations | Comparison with previous studies |
| Language | Neutral, past tense, no praise or speculation | Interpretive but careful, claims tied to evidence |
| What to avoid | Interpretation, selective reporting, repetition | Over-claiming, ignoring contradictions, vague limits |
| Ending | Last finding, stated plainly | Conclusion and a specific next research question |
These authoritative external resources explain the standards and guidance mentioned in this article:
Should results and discussion be separate sections or combined?
Follow the target journal's structure. Many journals keep them separate; some short-format journals combine them. Separate sections make the division between finding and interpretation clearer, which reviewers prefer unless the journal asks otherwise.
Can I mention a finding in the discussion that is not in the results?
No. Every finding discussed must first appear in the results, with its numbers or evidence. A discussion that introduces a new result looks careless and forces reviewers to question what else is missing.
How do I report a result that was not statistically significant?
Report it fully, with its effect size and interval, and discuss it honestly. A null result is a result: it answers the question, and hiding it distorts the literature. Explain what it may mean without claiming more than the data support.
How long should the discussion be?
Long enough to answer the research question, compare the findings with previous work, state the limitations, and draw a conclusion — usually similar in length to the results section. A discussion that merely repeats the results is too long whatever its word count.
What if my findings contradict previous studies?
Say so plainly and discuss the plausible reasons: different population, method, setting, or time period, or a weakness in your own study. Contradiction handled fairly is a strength; contradiction hidden or dismissed is a weakness reviewers will find.
Should the conclusion repeat the abstract?
No. The conclusion should follow from the discussion: the main finding, its implication, and a specific next question. Repeating the abstract wastes the reader's last impression of the paper.