How to Write a Research Paper Results Section
A traceable Results workflow for quantitative, qualitative, and mixed evidence without leaking interpretation into the findings.
Next step
Choose the next useful decision step first.
Use the guide or checklist that matches this page's intent before you ask for a manuscript-level diagnostic.
Quick answer: Write the Results section by organizing findings around the research questions or prespecified analyses, not the order in which data were collected. For each result, name the analysis, report the estimate with uncertainty, point to the relevant table or figure, retain meaningful null or mixed findings, and keep explanation for the Discussion unless the journal combines sections.
Evidence basis: We reviewed a peer-reviewed research-writing guide, the University of Guelph Results resource, and the UCSF Results guide on September 4, 2026. Those sources describe established reporting functions. The traceability grid below is Manusights guidance.
Result element | Reader check | Failure signal |
|---|---|---|
Question or hypothesis | Why is this analysis here? | Findings follow software output order |
Method link | Was the analysis prespecified and described? | A new variable or model appears without Methods support |
Estimate | What direction and magnitude were observed? | Only a p-value or adjective is reported |
Uncertainty | How precise is the estimate? | Certainty exceeds intervals, variation, or sample support |
Display | Does the table or figure add detail? | Text duplicates every number in the display |
Boundary | Which finding is primary, secondary, exploratory, null, or missing? | Exploratory patterns are presented as confirmatory |
Build a question-to-display outline
List each research question, the method that answers it, the primary estimate, and the best display. This outline should expose orphan analyses immediately. If a result has no question or Methods path, either label it as exploratory and describe the method or remove it.
Start with participant, sample, or corpus flow when that context controls interpretation. Then present primary findings, secondary findings, sensitivity checks, and exploratory results in a hierarchy the reader can follow.
Worked quantitative example
Weak version: “Temperature was significantly lower in high-canopy neighborhoods (p < 0.05), demonstrating that trees reduce urban heat.”
Evidence-matched Results sentence: “After adjustment for built density and elevation, blocks in the highest canopy quartile had afternoon temperatures 1.8°C lower than blocks in the lowest quartile (95% CI, 0.9°C to 2.7°C; Table 2).”
The revised sentence reports comparison, magnitude, uncertainty, model context, and display. It does not claim that planting trees caused the difference. That interpretation belongs in the Discussion.
Worked qualitative example
Weak version: “Participants felt the review process was difficult.”
Evidence-matched Results passage: “Participants described two recurring barriers: uncertainty about which reviewer requests required new analysis and difficulty tracking changes across manuscript versions. The first theme appeared across early- and late-career groups; the second was concentrated among multi-institution teams. Representative excerpts appear in Table 3.”
Qualitative Results still need traceability. Name how themes were developed, show variation and counterexamples, and link claims to quotations or other supporting records. Do not turn frequency language into prevalence unless the design supports it.
Decide what belongs in text, tables, and figures
Use text for the question, main pattern, estimate, uncertainty, and exceptions the reader must not miss. Use tables for exact values across several groups or models. Use figures for patterns, distributions, trajectories, or relationships that are easier to see than to read.
Do not repeat every cell in prose. A useful sentence tells the reader what to inspect in the display and why it answers the research question.
Preserve null, negative, and mixed evidence
Selective reporting makes a Results section look cleaner while weakening trust. Keep prespecified outcomes even when they are null. Report adverse, contradictory, or subgroup findings when they alter the answer. Label exploratory analyses so readers can distinguish discovery from confirmation.
Check Results traceability and claim language after tables and figures are locked.
Failure states and recovery
Symptom | Diagnosis | Recovery |
|---|---|---|
Results follow software output | Reader questions do not organize the section | Rebuild headings from objectives and hypotheses |
Only p-values appear | Magnitude and precision are hidden | Report estimates, intervals, denominators, and units |
Prose repeats tables | Text and display have the same job | Keep the pattern in text and detail in the display |
A new analysis appears | Methods and Results are disconnected | Add the method and label analysis status, or remove it |
Null findings disappear | Reporting is selective | Restore prespecified outcomes and explain missingness |
Interpretation dominates | Results and Discussion jobs are mixed | Move causes, comparisons, implications, and recommendations |
Final consistency audit
- Every primary question has a reported answer.
- Every result has a traceable Methods path.
- Estimates include units, denominators, and uncertainty where applicable.
- Primary, secondary, sensitivity, and exploratory analyses are labeled.
- Tables, figures, and text complement rather than duplicate one another.
- Null, negative, mixed, and missing results are handled transparently.
- Numbers agree across Results, abstract, tables, figures, and supplements.
- Interpretation is reserved for the Discussion unless the journal combines them.
Use the Methods section guide when an analysis cannot be traced backward, then use the Discussion guide to interpret the supported findings without expanding the claim.
Results are only traceable when the method is explicit. The Methods section workflow maps design, units, procedure, measurement, and analysis before results are reported.
Sources accessed September 4, 2026.
- How to Write and Publish a Research Paper for a Peer-Reviewed Journal, peer-reviewed guide archived by PubMed Central.
- Results, University of Guelph Writing in the Sciences.
- How to Write the Results Section, University of California, San Francisco.
Readiness check
Run the scan while the topic is in front of you.
See score, top issues, and journal-fit signals before you submit.
Submit if, and think twice if
Submit if every result has two-way traceability. Read forward from each question to a Methods subsection, analysis, result, and display. Then read backward from every table, figure, and reported model to the question it answers. Resolve orphan analyses, new outcomes, and unexplained denominators before copyediting.
Think twice if adjectives carry the finding. Words such as “marked,” “substantial,” “similar,” or “improved” should not substitute for an estimate and its uncertainty. Give the comparison, magnitude, units, denominator, interval or variation, and display reference that let a reader evaluate the description.
Hold if the hierarchy changed after seeing the data. If an exploratory subgroup or model became prominent because its result looked favorable, label that status and retain the prespecified outcomes. Explain the revised analysis path in the appropriate Methods or transparency statement rather than presenting hindsight as planning.
Run a cross-surface number audit. Check the abstract, Results prose, tables, figures, captions, supplements, and data availability materials for the same sample sizes, units, time points, model labels, signs, and rounding. The cited writing resources define accepted Results functions. This two-way traceability audit is Manusights editorial judgment. It cannot predict acceptance.
Frequently asked questions
Organize it by the research questions or prespecified analyses: sample or participant flow where needed, primary results, secondary results, sensitivity checks, and clearly labeled exploratory findings.
No. Report the estimate, direction, units, denominator, and uncertainty appropriate to the analysis. A p-value alone hides effect size and precision.
Prespecified null, negative, mixed, and missing outcomes should remain visible when they affect the answer. Omitting them creates selective reporting and weakens trust.
Use prose to name the research question, main pattern, estimate, uncertainty, and important exception. Let the table carry the complete values rather than repeating every cell.
Before you upload
Choose the next useful decision step first.
Move from this article into the next decision-support step. The scan works best once the journal and submission plan are clearer.
Use the scan once the manuscript and target journal are concrete enough to evaluate.
Private API processing. Your manuscript is not used to train models.
Put the guidance to work
Choose the practical next step for your research.
Start with the question in front of you, or use the guided learning path when you want a clearer view of the full publishing process.