Research Paper
42 Research Paper Conclusion Examples by Methodology
See 42 real conclusion examples segmented by quantitative, qualitative, and mixed methods—with specific language patterns and structures you can adapt for your own research paper.
You’ve written the whole paper, the results are in place, and the conclusion still sounds like a warmed-over abstract. The fix is simple: your conclusion should say what your study now makes possible to know, decide, or investigate next.
Use the examples below by methodology. A quantitative conclusion earns trust through precision; a qualitative conclusion earns trust through interpretation; a mixed methods conclusion earns trust by showing how one data type changes the meaning of the other.
Contents:
Who This List Is For
How We Picked These 42 Examples
What Makes a Strong Research Conclusion
14 Quantitative Research Conclusion Examples
12 Qualitative Research Conclusion Examples
8 Mixed Methods Research Conclusion Examples
5 Action Research & Case Study Conclusion Examples
3 Chapter 5 (Dissertation) Conclusion Examples
How to Use This List in Your Own Writing
Who This List Is For
This list is for students and researchers who know what they found but can’t quite close the loop. That includes undergrads finishing a 12-page paper, PhD candidates drafting Chapter 5, and faculty trying to show students the difference between a summary and an actual conclusion.
It’s also for anyone switching methods. If you’ve spent a year writing regression-heavy conclusions, qualitative closure can feel strangely underpowered. No p-values. No confidence intervals. More interpretation.
The conclusion has a specific job. As the UNC Writing Center’s guide to conclusions puts it, the ending should help readers understand why the paper matters after they’ve followed the argument. That’s harder than repeating the thesis.
If you’re still shaping the paper around the conclusion, pair this with guides on research paper thesis examples, research paper introduction examples, and how to write the conclusion section of a research paper.
How We Picked These 42 Examples
These examples are modeled on the language patterns used in published academic writing: quantified findings, bounded claims, limitation statements, mechanism language, and future-work moves. They’re written as adaptable examples, not as passages to paste into a paper.
The sorting matters. A quantitative paper often closes by naming effect size, uncertainty, and generalizability. A qualitative paper usually closes by naming the interpretive contribution: a pattern, tension, identity process, or institutional mechanism.
Mixed methods conclusions are trickier. They have to integrate. The Nielsen Norman Group’s primer on mixed-methods research describes the core value as combining quantitative and qualitative evidence so the researcher can answer different parts of the same problem. A strong conclusion makes that integration visible.
We also screened for the mistakes advisors keep circling in red: vague “future research is needed” endings, limitation paragraphs that sound apologetic, and conclusions that introduce brand-new claims after the argument has already ended. Painful. Common.
What Makes a Strong Research Conclusion
A strong conclusion circles back to the research question with new authority. The reader should feel that the question has been changed by the evidence.
The Harvard College Writing Center’s guidance on conclusions is useful because it treats the ending as part of the argument, not a decorative final paragraph. The conclusion should widen the reader’s sense of consequence without wandering into claims the paper didn’t earn.
Weak research ending | Strong research ending |
|---|---|
Restates the abstract | Returns to the research question with a sharper answer |
Lists findings one by one | Shows how findings fit together |
Says “more research is needed” | Names the next design, population, or test |
Apologizes for limitations | Frames limits as boundaries of interpretation |
Uses the same voice for every method | Matches the logic of the methodology |
The methodology’s voice does a lot of work. Quantitative conclusions usually sound measured and bounded. Qualitative conclusions tend to sound interpretive and process-oriented. Action research conclusions should show change across cycles.
Reproducibility also belongs near the ending, especially in empirical work. A 2025 consensus paper on core reproducibility items in research argues for minimum expectations across planning, methods, data collection, analysis, and dissemination. Conclusions shouldn’t smuggle in certainty the methods can’t support.
For structure, keep five elements in view: question, synthesis, limits, implications, next step. Miss one and the ending usually feels thin.
14 Quantitative Research Conclusion Examples
Quantitative conclusions need numbers, but numbers alone don’t close the paper. The good ones interpret magnitude, explain uncertainty, and avoid pretending correlation has solved causality.
If you’re still working through the design behind the results, read different types of research methods and the methodology section of a research paper before polishing the final paragraph.
1. Correlational study conclusion
“Our findings indicate a statistically significant inverse relationship between social media use and academic performance (r = −0.42, p < 0.01). Because the design is correlational, future work should use longitudinal data to test whether reduced screen time changes academic outcomes.”
Works because it gives the statistic, then reins in causality.
2. Clinical intervention conclusion
“The intervention group showed a 23% reduction in hospital readmissions over 12 months compared with controls, exceeding the pre-specified threshold for clinical relevance. These results support testing the protocol beyond urban teaching hospitals, where staffing and follow-up conditions may differ.”
Works because it separates practical significance from generalizability.
3. Regression model conclusion
“Regression analysis showed that baseline cognitive function and medication adherence explained the largest share of variance in treatment response (R² = 0.31). The weak contribution of the remaining predictors suggests that the current model may overstate the number of mechanisms involved.”
Works because it treats weak predictors as useful information.
4. Dose-response conclusion
“The dose-response curve suggests a threshold effect: cognitive benefits emerged above 500 mg/day, with little additional gain after 1000 mg/day. This nonlinear pattern challenges the linear assumption used in prior dosing models and should be tested prospectively.”
Works because it turns a pattern into a testable claim.
5. Meta-analysis conclusion
“Across cohorts, the pooled effect remained consistent (OR = 1.87, 95% CI: 1.52–2.30), with low heterogeneity. Publication bias still cannot be excluded, so the estimate should be interpreted alongside funnel plot diagnostics and unpublished evidence where available.”
Works because it reports confidence and uncertainty in the same breath.
6. Mediation analysis conclusion
“Structural equation modeling supported a partial mediation pathway from socioeconomic status to academic achievement through educational investment. Because the direct effect remained significant, interventions at the investment stage may help, but they won’t fully address structural inequality.”
Works because it translates mediation into policy language without overstating it.
7. Survival analysis conclusion
“Survival analysis showed a median time-to-event of 36 months in the treatment group compared with 24 months in the control group. The benefit was concentrated among patients under 65, suggesting that age should be treated as a boundary condition in future trials.”
Works because it resists flattening subgroup results.
8. Interrupted time-series conclusion
“The time-series analysis detected a significant break in trend six months after policy implementation. The pattern is consistent with a policy effect, although unmeasured confounders mean replication in other jurisdictions is needed before broad adoption.”
Works because it uses cautious causal language.
9. Multilevel modeling conclusion
“Multilevel modeling showed that school-level factors explained 18% of the variance in student outcomes, while individual-level factors explained 52%. A policy response focused only on individual behavior would miss a meaningful source of institutional variation.”
Works because variance components become a decision point.
10. Meta-regression conclusion
“Meta-regression found larger effect sizes in studies using active comparators than in placebo-controlled studies. This difference suggests that efficacy estimates may shift depending on the benchmark used, so clinicians should read pooled effects with the comparator in mind.”
Works because it identifies a source of bias in the literature.
11. Logistic regression conclusion
“Logistic regression identified treatment adherence as the strongest independent predictor of treatment success, while baseline severity reduced the odds of response. With an AUC of 0.78, the model may support patient selection, though it is not accurate enough for stand-alone clinical decisions.”
Works because it names the model’s practical limit.
12. Underpowered trial conclusion
“The confidence interval for the primary endpoint crossed zero, leaving the hypothesized effect unresolved. A larger trial, powered for the observed variance rather than the planned estimate, is needed before claims about efficacy are warranted.”
Works because a null result doesn’t get buried.
13. Sensitivity analysis conclusion
“Sensitivity analysis showed that the estimated treatment effect changed only modestly across missing-data assumptions. This stability increases confidence in the finding, though the reason data were missing remains unresolved.”
Works because it reports assumption-testing clearly.
14. Incidence rate conclusion
“The intervention group had a lower event rate per person-year of follow-up, with the effect persisting among participants with higher baseline severity. These findings suggest the intervention may apply across a wider risk range than initially expected.”
Works because it connects subgroup evidence to generalizability.
12 Qualitative Research Conclusion Examples
Qualitative conclusions succeed when they name the pattern the study made visible. They should not imitate quantitative closure. No false precision, please.
A good qualitative ending often turns on a phrase: “informed negotiation,” “adaptive practice,” “identity reconstruction,” “rushing to solutions.” If you’re still choosing the right qualitative frame, start with types of qualitative research methods or qualitative research design.
1. Interview study conclusion
“Across 18 interviews, participants described a tension between wanting autonomy and needing expert guidance during uncertain treatment decisions. The study characterizes this process as informed negotiation, where patients used clinical input to make choices aligned with personal values.”
Works because it names the core process, not merely the topic.
2. Housing transition conclusion
“The analysis identified distinct pathways from homelessness to stable housing, shaped by family reconnection, institutional support, or peer networks. These pathways suggest that housing interventions should be matched to entry conditions rather than built around a single route to stability.”
Works because it develops a typology with practical use.
3. Aging narratives conclusion
“Participants described aging through shifting narratives of decline, wisdom, invisibility, and opportunity. Because individuals moved between these narratives depending on context, programs focused only on ‘positive aging’ may miss the ambivalence people actually live with.”
Works because it preserves contradiction.
4. Clinic observation conclusion
“Observations across community clinics showed that open-ended questions and reflective listening were associated with stronger patient engagement. Patients linked these interactions to feeling heard, suggesting that communication training may affect retention as much as formal follow-up protocols.”
Works because it connects observed behavior to participant meaning.
5. Autism diagnosis conclusion
“Parents of children with autism described diagnosis as both validating and threatening. This ambivalence should be understood as active meaning-making rather than resistance to care.”
Works because it reframes the obvious interpretation.
6. Safety culture conclusion
“Focus groups suggested that psychological safety shaped whether staff reported safety incidents or kept them hidden. Technical reporting systems may fail when organizational norms punish uncertainty or bad news.”
Works because it identifies the hidden condition behind compliance.
7. Medication adherence conclusion
“Narrative analysis showed that adherence was tied to identity, not only memory or knowledge. Patients were more likely to continue medication when treatment fit their sense of self.”
Works because it challenges the knowledge-deficit model.
8. Social work protocol conclusion
“Interviews with social workers showed frequent departures from agency protocols in response to client needs. The study frames these departures as adaptive practice, a form of professional judgment that complicates rigid fidelity models.”
Works because it treats deviation as data.
9. Policy implementation conclusion
“The case traced how a prior-authorization policy produced delays, substitute prescribing, and administrative burden across healthcare settings. Policies designed away from frontline conditions may produce outcomes misaligned with their intent.”
Works because it uses one case to clarify a broader mechanism.
10. School inclusion conclusion
“Ethnographic observation showed that students with disabilities were physically included but often socially isolated. Placement in mainstream classrooms, by itself, did not ensure access to instruction or peer belonging.”
Works because it distinguishes structural inclusion from lived inclusion.
11. Climate science emotion conclusion
“Interviews with climate scientists revealed anxiety, grief, and anger about the pace of policy response. Because these emotions rarely appear in formal scientific communication, the public record may underrepresent part of the work’s human cost.”
Works because it brings an invisible dimension into the argument.
12. Incident review conclusion
“The analysis found a recurring pattern after critical incidents: organizations acknowledged harm, then shifted too quickly toward solutions before causes were understood. Slower review processes may be necessary when the goal is learning rather than reputational repair.”
Works because it names the organizational reflex.
8 Mixed Methods Research Conclusion Examples
Mixed methods conclusions should feel integrated, not stapled together. The reader should see how the qualitative findings changed the interpretation of the numbers, or how the numbers tested the reach of the qualitative insight.
The SAGE overview of quantitative, qualitative, and mixed research is useful background here because it treats method choice as a logic problem. Different methods answer different kinds of questions. The conclusion has to respect that.
1. Intervention mechanism conclusion
“The quantitative data showed symptom improvement among 67% of participants after the intervention. Interviews suggested that participants attributed improvement less to the intervention content than to the increased attention and support surrounding it.”
Works because the qualitative finding changes the mechanism.
2. Teacher preparedness conclusion
“Survey data showed that most teachers felt unprepared to teach climate science. Interviews revealed that ‘unprepared’ meant different things: content gaps for some teachers, classroom strategy concerns for others, and political discomfort for another group.”
Works because it disaggregates a broad quantitative result.
3. Trial exit-interview conclusion
“The randomized trial showed that the intervention outperformed usual care on average. Exit interviews complicated that result by showing that a substantial subgroup experienced the program as too rigid.”
Works because average effectiveness doesn’t erase fit problems.
4. Organizational culture conclusion
“Survey measures of psychological safety improved after the leadership intervention. Interviews showed that the improvement was concentrated among managers, while frontline staff described little change.”
Works because qualitative data exposes uneven distribution.
5. Trauma recovery conclusion
“Latent class analysis identified distinct recovery trajectories after trauma. Interviews suggested that social support and meaning-making helped explain why people with similar trauma severity followed different paths.”
Works because the qualitative layer explains variation.
6. Patient satisfaction conclusion
“Patient satisfaction scores were high, with a mean rating of 8.2 out of 10. Interviews still revealed concerns about communication, suggesting that satisfaction ratings may miss parts of care quality that matter to patients.”
Works because it critiques the measure.
7. Responder analysis conclusion
“The intervention reduced symptoms among responders but had little effect for non-responders. Qualitative comparison showed that responders entered the program with stronger engagement and more support outside treatment.”
Works because it identifies context behind response.
8. Implementation fidelity conclusion
“Quantitative data showed wide variation in implementation fidelity across schools. Case studies suggested that leadership commitment explained more of the variation than resource levels alone.”
Works because it links measured variation to field conditions.
5 Action Research & Case Study Conclusion Examples
Action research and case study conclusions need to show what changed. The closure often has a narrative arc: starting condition, intervention, adjustment, observed result.
That can make the ending sound too anecdotal if the evidence is thin. Don’t let it. Strong case conclusions separate what happened in the case from what may transfer elsewhere.
1. Classroom action research conclusion
“Across three action research cycles, the peer feedback protocol moved from unstructured comments to guided review and peer coaching. By the final cycle, most feedback exchanges met the project’s quality criteria, and the protocol was adopted into regular classroom practice.”
Works because it shows iteration and sustainability.
2. Rural clinic case study conclusion
“This case study of a rural clinic’s EHR transition found that staff training and realistic timelines shaped adoption as much as technical configuration. The case suggests that similar clinics should treat implementation as organizational change, not software installation.”
Works because it extracts a transferable lesson from one site.
3. Manufacturing action research conclusion
“The lean process initiative reduced cycle time and defect rates over six months. Worker interviews also showed higher stress, indicating that efficiency gains should be evaluated alongside well-being before the revised process is scaled.”
Works because it names the unintended consequence.
4. Nonprofit scaling case conclusion
“The nonprofit scaled its pilot from one site to five by adapting curriculum, staffing, and partnerships to local conditions. The case suggests that fidelity to core principles may matter more than preserving every feature of the original model.”
Works because it reframes the fidelity problem.
5. Participatory environmental health conclusion
“The participatory project showed that residents held local environmental knowledge absent from official surveillance data. Combining community accounts with epidemiological evidence helped identify a health cluster that top-down monitoring had missed.”
Works because it validates community expertise as evidence.
3 Chapter 5 (Dissertation) Conclusion Examples
Chapter 5 has more room than a journal conclusion, but that room creates a trap. Students often re-summarize Chapters 1 through 4 instead of closing the dissertation’s argument.
The chapter should synthesize the study’s contribution, limitations, implications, and future research. If you’re writing the discussion before the final conclusion, use how to write discussion in a research paper to keep the two sections from blending together.
1. Mixed methods dissertation conclusion
“This dissertation investigated how early childhood trauma affects adult attachment patterns. Quantitative analysis identified an indirect pathway through childhood attachment disruption, while interviews showed that compensatory relationships could interrupt that pathway. The findings suggest that attachment-focused interventions may have long-term protective value and that resilience factors deserve equal attention to risk factors.”
Works because it integrates methods and avoids determinism.
2. Organizational case dissertation conclusion
“This dissertation examined organizational responses to healthcare reform across three systems. The case analysis showed that organizations with strong communication infrastructure adapted more effectively than organizations with siloed decision-making. The study extends organizational change theory by positioning communication capacity as preparation for policy change rather than a response after disruption begins.”
Works because it names the theoretical contribution.
3. Qualitative dissertation conclusion
“This dissertation investigated the lived experiences of adults with undiagnosed autism. Interviews showed that late diagnosis was both validating and disorienting, requiring participants to reinterpret earlier life experiences. The findings suggest that diagnostic practice should better account for adults and for gendered patterns of missed identification.”
Works because it connects lived experience to clinical implication.
How to Use This List in Your Own Writing
Start by identifying your methodology type. Sounds obvious; it isn’t. A lot of weak conclusions come from borrowing the voice of the wrong method.
If you’re writing a quantitative paper, your conclusion should probably include the main estimate, the practical meaning of that estimate, and the boundary of the claim. If you’re writing qualitative work, look for the strongest theme, process, contradiction, or reframing. Mixed methods? State the integration plainly.
Don’t copy the wording. Copy the logic.
A useful workflow is to put your draft conclusion beside two examples from the same method family. Then mark five items: research question, synthesis, limitation, implication, next step. If one is missing, the reader will feel it before they can name it.
This is where Otio’s unified research workspace is handy: upload your draft and a few model conclusions, then ask the chat to compare structure, tone, and missing elements across the examples. Keep the AI on a short leash, though. A 2026 arXiv paper on reference hallucinations in commercial LLMs and deep research agents found that 3–13% of citation URLs were hallucinated, with 5–18% non-resolving overall. Verify every source.
One last check: read the conclusion aloud. If it sounds like you’re reporting what the paper contained, revise it. If it sounds like the study has changed what the reader can responsibly say next, you’re close.
Try Otio for your next research paper draft if you want one place to compare sources, examples, notes, and revisions.
FAQ
Q: What's the difference between a conclusion and a discussion in a research paper?
A: A discussion interprets findings and connects them to existing literature; a conclusion gives the final synthesis, boundaries, and next step. Some papers combine them, but the closing paragraph still has to do conclusion work.
Q: How long should a research conclusion be?
A: For journal articles, 1–2 pages is common; for dissertations, Chapter 5 conclusions often run longer. The real test is whether you can restate the question, synthesize findings, name limits, and point forward without repeating the abstract.
Q: Can I introduce new citations in the conclusion?
A: Usually no. New sources belong in the literature review or discussion, because the conclusion should close the argument already built.
Q: How do I write a conclusion for a mixed methods study?
A: Show how the quantitative and qualitative findings interact. Don’t park them in separate paragraphs unless the design requires it.
Q: What if my research didn't find what I expected?
A: Treat null or unexpected findings as results, not failures. Explain what they challenge, what they rule out, and what design should test the revised explanation.
