How to Write a Conclusion for a Research Paper (With 3 Examples)
Dr. Rajni Garg
Executive Lead, Research & Content Strategy
8 min read
To write a strong conclusion for a research paper, answer your research question in plain words, say what your findings mean, and suggest what should happen next. Keep it short, and add nothing the reader has not already seen in the paper: no new data and no new sources.
This guide on how to write a conclusion for a research paper compares it with the discussion and the abstract, gives five steps and annotates three examples. The first is the closing paragraph of the writing groups study whose opening we annotated in our guide to the research paper introduction.
TL;DR
Answer the research question first, in a sentence or two.
Sum up the findings that support the answer, then say what they mean and for whom.
Name the main limit and one specific next step.
Leave out new results, new sources and claims your data cannot support.
What does a research paper conclusion do?
A conclusion answers your research question, says what the findings mean, and says what should come next. Where the introduction narrows from a broad topic to one question, the conclusion widens back out, from your answer to what it means for the field.
In a journal article it is usually brief. A thesis is different, since many schools expect a whole closing chapter, covered below. If you are still drafting the sections that come before it, our step-by-step research paper guide covers the whole process.
Conclusion vs discussion vs abstract
All three report what you found, so they are easy to blur. The difference is depth and purpose.
Section | What it does | How much detail | Cites other studies? |
|---|---|---|---|
Discussion | Interprets the results, compares them with earlier studies, explains surprises and sets out limitations in full | High: it works through each main result | Yes, often |
Conclusion | Answers the question and gives the main implication and next step | Low: only the findings that matter most | Rarely, and never new sources |
Abstract | A stand-alone summary of the whole paper, including methods and results, for readers who may read nothing else | Low, within the journal’s word limit | Usually not |
A simple test: the discussion asks “what do these results mean, and how do they fit with other work?” The conclusion asks “so what, and what now?” If your journal combines the two under one heading, save the final paragraph for the conclusion. The abstract often ends on a one-sentence version of your conclusion, so write the conclusion before the abstract. Our guide on how to write an abstract covers the rest.
How to write a research paper conclusion in 5 steps
Restate the question and the answer. In one or two sentences, remind the reader what you asked, then answer it directly. Rephrase it rather than copying your introduction.
Sum up the key findings. Pick the two or three results that support your answer. Leave out test statistics and most numbers, since the reader has already seen them.
Give the implications. Say what the findings mean for theory, practice or policy, and who should care. This is the “so what” step, and it is easy to skip.
Note the limits. In a sentence or two, name the limits that most affect how far your answer reaches, such as the sample, setting or design. The full account of limitations belongs in the discussion.
Suggest next steps. Propose a specific study or action that follows from your findings and limits. “More research is needed” is not a next step. “A trial across several universities” is.
If your findings and limitations are already written up, the AnswerThis research paper conclusion generator can help. Paste in your paper or key points, choose a length, and it writes one closing paragraph of about 40 to 70, 80 to 120 or 130 to 180 words. You get 3 free runs, and each result is a draft paragraph you can build on. It works for essays too, but for a research paper, give it your findings and limits rather than just your topic.
Research paper conclusion examples, annotated
All three are made-up teaching examples, so none of the studies, figures or results is real. The label after each sentence names the step it carries out.
1. Empirical study conclusion
Study: the doctoral writing groups study from our introduction guide, now at its end
This study asked whether joining a weekly writing group is linked to how much doctoral students write and how soon they submit a first manuscript, and the answer is a qualified yes. (Step 1: restate the question and the answer) Over one year, students who attended at least half of the sessions logged about two more writing hours a week than similar students who did not join, and were more likely to submit a manuscript. (Step 2: sum up the key findings) For departments, this suggests that writing groups do more than offer moral support and may be a low-cost way to help students start publishing. (Step 3: give the implications) Because students chose whether to join, more motivated students may have opted in, and all of them came from one university. (Step 4: note the limits) A study that offers group places by lottery, across several universities, would show whether the groups themselves make the difference. (Step 5: suggest next steps)
What to notice: the answer comes first and is hedged to match the design. The limit (students chose to join) leads straight to the next step (a lottery), a study someone could actually run.
2. Argumentative paper conclusion
Paper: Why promotion reviews should count datasets and software as research outputs
This paper has argued that universities should count well-documented datasets and research software as outputs in promotion reviews, alongside articles. (Step 1: restate the question and the answer) Three points support that claim: these outputs take skilled work to build, other researchers reuse them, and leaving them out steers early-career researchers away from work their fields depend on. (Step 2: sum up the key points) If committees accept this, they will need clear criteria for judging quality, such as documentation and evidence of reuse, not just a longer list of outputs. (Step 3: give the implications) The argument draws mainly on fields where sharing data and code is already common, and may apply less where it is not. (Step 4: note the limits) A sensible next step is for a few departments to pilot such criteria and report how committees used them. (Step 5: suggest next steps)
What to notice: in an argumentative paper, step 2 sums up your main reasons rather than findings, and the limit is about how far the argument reaches, not about data.
3. Literature review conclusion
Review: Night-time noise on hospital wards and patient sleep
This review asked how night-time noise on hospital wards affects patients’ sleep, and the short answer is that it clearly disturbs sleep, though by how much is still unclear. (Step 1: restate the question and the answer) Across the 38 studies reviewed, patients often named noise as a main cause of poor sleep, but few studies measured sleep directly. (Step 2: sum up the key findings) For hospitals, this makes quieter night-time routines, such as lower alarm volumes and quieter handovers, a practical place to start. (Step 3: give the implications) Most studies were small, took place on a single site and relied on patient surveys. (Step 4: note the limits) Future studies should pair noise readings with objective sleep measures, such as wrist-worn activity monitors, so that wards can be compared. (Step 5: suggest next steps)
What to notice: a review’s findings are patterns across studies, so step 2 describes the evidence as a whole. The limits are mostly limits of the literature, not of the review.
What not to put in a conclusion
The USC Libraries guide to conclusions sums up the rule as new insight, not new information. In practice, three things stay out.
New results. If a finding matters enough to mention, it belongs in the results and discussion first. A conclusion that reveals a new subgroup analysis gives the reader no way to check it.
New sources. The conclusion is about your study. If a source helps explain a finding, cite it in the discussion, where you have room to engage with it.
Overclaiming. A survey cannot show cause and effect, and one university cannot speak for all of them. Compare “writing groups make students publish sooner” with “joining a writing group was linked to earlier submission at one university.” The second is less exciting and much harder for a reviewer to reject.
Two smaller habits also weaken conclusions: apologizing at length for every limitation, and copying sentences from the introduction. Name the main limit once and move on.
Conclusions and recommendations in a thesis (Chapter 5)
In many schools, including many in the Philippines and the US, a five-chapter thesis ends with a chapter titled something like “Summary, Conclusions and Recommendations.” It does the same job as an article conclusion, at greater length and in a set order:
Summary of findings. The main results, often one short paragraph per research question, without repeating every table.
Conclusions. What the findings mean, answering each research question or hypothesis in turn.
Recommendations. Specific actions for practitioners, policymakers or future researchers.
Each recommendation should trace back to a finding. In the writing groups study, the finding on writing hours supports recommending that departments pilot writing groups. It does not support making them compulsory, because the study did not test that.
Programs vary. Some put the full discussion in this chapter, and others add implications and limitations as separate parts. For example, Western Kentucky University’s EdD rubric for Chapter 5 asks for implications, limitations, recommendations and final conclusions. Your school’s research manual sets the exact title and order.
FAQs
How long should a research paper conclusion be?
In a journal article, often one to three paragraphs. A thesis conclusion chapter can run to several pages, organized by research question.
Does every research paper need a separate conclusion section?
No. Some journals ask for a “Conclusions” heading, while others end with the discussion and expect its last paragraph to do the concluding. Either way, that final paragraph should answer your question.
What tense should I use in a conclusion?
Use the past tense for what you did and found (“the study found”), and the present tense for what the findings mean (“this suggests”).
Should I start my conclusion with “In conclusion”?
You can, but you rarely need to. The heading already tells readers where they are, and opening with the answer to your question is a stronger start.
Can I end a research paper with a question?
Only if it is a specific question your findings raise and the next study could answer. A vague rhetorical question sounds unfinished, and a concrete next step usually does the job better.
With your discussion written, paste your findings and limits into the research paper conclusion generator for a first closing paragraph, then edit it until every sentence matches what your paper shows.
.png)
