The results section is where you present what you found - and only what you found. It is the most fact-bound part of a research paper: numbers, themes, observations, all reported as cleanly as possible, with interpretation saved for the discussion.
Most weak results sections fail in one of two ways. They either bury the findings in interpretation ("This shows that..." instead of "Group A scored higher than Group B"), or they dump every number from the analysis without selecting what matters. This guide shows you how to avoid both, with templates and side-by-side examples for quantitative and qualitative work.
Need help writing your results section?
ThesisAI helps you turn your data into a clear, well-organised results section, with citations checked against your sources.
Start WritingWhat is a Results Section?
The results section reports the findings of your study in a clear, organised, and neutral way. It answers the question: What did the data show?
It does not answer:
- What does this finding mean? (discussion)
- How did you collect the data? (methodology)
- How does this fit with prior research? (discussion)
If you find yourself starting a sentence with "This suggests..." or "This indicates...", it probably belongs in the discussion.
Results vs Discussion: One Quick Test
If a sentence reports something that was true in your data, it's a result. If it reports something about what your data means, it's discussion. Keep this divide sharp.
Standard Structure of a Results Section
The structure depends on the type of research, but the underlying logic is the same: organise findings around your research questions, and report the most important findings first.
Quantitative Results: Standard Structure
- Sample description - basic descriptive statistics for the final sample.
- Descriptive results - means, frequencies, distributions for key variables.
- Inferential results - hypothesis tests, model output, effect sizes.
- Subgroup or secondary analyses - any planned secondary tests.
- Tables and figures - referenced from the text, never standalone.
Qualitative Results: Standard Structure
- Brief overview of participants and data corpus - sample summary, total interview hours, etc.
- Themes or categories - typically 3-6 main themes, presented in the order they answer the research question.
- Within each theme - definition, supporting interview quotes, and frequency or distribution where appropriate.
- Relationships between themes - if relevant.
Results Templates (Copy and Adapt)
Template 1: Quantitative Results Paragraph
Fill-in-the-blanks template
The final sample comprised [N] [participants/units] (M_age = [X], SD = [Y]; [%] [demographic]). [Variable 1] was [normally / non-normally] distributed (Shapiro-Wilk p = [value]). On average, the [intervention/group A] showed [statistic] ([CI]), compared with [statistic] ([CI]) in the [control/group B]. A [test] revealed a significant difference, [statistic, e.g. t(118) = 3.24, p = .002, d = 0.59]. The effect direction supports the prediction that [hypothesis]. Full descriptive statistics are reported in Table 1; group differences are visualised in Figure 1.
Template 2: Qualitative Theme Paragraph
Fill-in-the-blanks template
Theme [N]: [Theme name]. [Theme appeared in N of N interviews]. This theme captured [definition - the kind of meaning the theme represents]. Participants described [observation 1] and [observation 2], often in the context of [setting]. As one [participant role] put it: "[Verbatim quote.]" (Participant [pseudonym]). A second participant echoed this, while extending it to [aspect]: "[Quote.]" (Participant [pseudonym]). Within this theme, two patterns were visible: [sub-pattern A] and [sub-pattern B], which are explored further in Theme [N+1].
Full Results Examples (Strong vs Weak)
Example 1: Quantitative - Mindfulness and Exam Anxiety
Weak version:
The mindfulness group did better. Their anxiety went down a lot, which is great, and clearly shows that mindfulness works. The control group did not improve as much, which makes sense given they did not receive any training. More research should be done in this area.
Why it's weak: Vague claims ("did better", "a lot"), no statistics, no test specified, interpretive language ("great", "clearly shows", "makes sense"), and a closing recommendation that belongs in the discussion.
Strong version:
The final sample comprised 118 participants (M_age = 20.4, SD = 1.7; 64% female), with two participants withdrawn for non-attendance. Pre-intervention STAI-S scores did not differ between groups, t(116) = 0.41, p = .684. Post-intervention, the mindfulness group reported significantly lower state anxiety (M = 32.4, SD = 8.1) than the wait-list control (M = 41.7, SD = 9.6), t(116) = 5.65, p < .001, d = 1.04, 95% CI [6.0, 12.6]. The reduction within the mindfulness group from baseline to follow-up was also significant, t(58) = 7.12, p < .001, while the control group showed no significant change, t(58) = 0.92, p = .362. Descriptive statistics for both groups are presented in Table 1; group means with 95% confidence intervals are visualised in Figure 1.
Why it works: Sample size given, named test, exact statistics, effect size with confidence interval, baseline equivalence checked, both between- and within-group comparisons reported, and clear pointers to the table and figure.
Example 2: Qualitative - Doctoral Supervision Experiences
Weak version:
The interviews showed that PhD students often felt anxious about their supervisors. Many participants felt unsupported. They said their supervisors were too busy, which is a common problem in academia. This is concerning and suggests universities need to do more.
Why it's weak: No themes named, no quotes, no participant attribution, vague intensifiers ("often", "many"), interpretive claim ("concerning", "suggests universities need to do more").
Strong version:
Three themes were generated through reflexive thematic analysis: (1) Negotiating absence, (2) Performing readiness, and (3) The hidden curriculum of feedback. Each is presented below with representative extracts.
Theme 1: Negotiating absence (10 of 12 interviews). Participants described an ongoing effort to manage their supervisors' limited availability. The absence was rarely framed as personal neglect; instead it was treated as a structural feature of academic life. As Maya (Year 2, Sociology) put it: "You learn pretty fast that 'I'll get back to you' means three weeks, and you start drafting twice as much so something is always in flight." This pattern was echoed by Daniel (Year 2, Politics): "It's a kind of solo running. I check in when I have output, not when I have questions, because the questions don't survive the wait." Two sub-patterns were visible within this theme: buffering (drafting redundant work to keep the relationship moving) and question rationing (saving up questions to fit a single meeting). Theme 2 builds directly on this rationing dynamic.
Why it works: Themes named and numbered, frequency reported, theme defined, attributed verbatim quotes, sub-patterns identified, and a connecting sentence to the next theme. No interpretation about what universities should do - that belongs in the discussion.
Common Results Mistakes (And How to Fix Them)
Mistake 1: Interpreting Findings Inside the Results
Problem: "The intervention reduced anxiety, which is exactly what previous research has shown and confirms our hypothesis."
Fix: Report the finding plainly. "The intervention group reported significantly lower anxiety than the control group, t(116) = 5.65, p < .001, d = 1.04." Save the link to prior research and the hypothesis verdict for the discussion.
Mistake 2: Reporting Every Number
Problem: A page-long paragraph dumping every descriptive statistic, demographic, and post-hoc test the analysis produced.
Fix: Put detailed statistics in a table and report only the key findings in the text. The narrative should highlight the headline numbers, not repeat the table.
Mistake 3: No Effect Sizes
Problem: "The difference was significant (p < .05)." Without an effect size, the reader cannot tell if it was meaningful.
Fix: Always pair p-values with effect sizes (Cohen's d, eta-squared, odds ratio) and confidence intervals where appropriate.
Mistake 4: Tables and Figures Without Narrative
Problem: Three tables are pasted with no text between them. The reader has to infer the story from the data.
Fix: For each table or figure, write a short sentence in the body referring to it and pointing out what to look at: "Mean scores by group are shown in Table 1; the largest difference appears in the post-test condition."
Mistake 5: No Quotes (Qualitative)
Problem: Themes are described abstractly with no direct quotation. Readers cannot judge whether the theme is grounded in the data.
Fix: Every theme should have at least two illustrative quotes, attributed to pseudonymised participants, and selected to show the pattern rather than the most dramatic case.
Mistake 6: Hand-picked Quotes (Qualitative)
Problem: Only the most quotable participants appear, distorting how widespread the theme actually was.
Fix: Report theme frequency or distribution ("appeared in 10 of 12 interviews"), and choose quotes that represent both typical and edge expressions of the theme.
How to Write the Results: A 5-Step Process
Step 1: Re-read Your Research Questions
Open your introduction and write the research questions at the top of a fresh page. Every paragraph in your results section should help answer one of them.
Step 2: Decide What Goes in Tables/Figures vs the Body
Anything detailed and tabular goes in a table. Anything visual and comparative often works better as a figure. The body text reports the headline finding and points the reader where to look for detail.
Step 3: Draft Findings First, Sentences Second
For quantitative work, list the key statistics; for qualitative work, list the themes. Then turn each into a clean reporting sentence using the templates above.
Step 4: Cut Interpretation
Read your draft and remove any sentence that says what the finding means. Move those sentences into a "discussion" file - they're not wasted, just relocated.
Step 5: Cross-check Numbers
Every statistic in the body should match the table, and every effect size should match the analysis output. Inconsistent numbers are the most common reason results sections lose credibility.
Discipline-Specific Tips
Sciences and Engineering
Report uncertainty: confidence intervals, error bars, or measurement uncertainty for each headline number. Follow your field's reporting standard (CONSORT for trials, ARRIVE for animal research, etc.).
Psychology and Social Sciences
APA-style reporting expects test statistic, degrees of freedom, exact p-value, and effect size. Confidence intervals are increasingly standard.
Qualitative Research
Themes should be named, defined, and supported with quotes. Report distribution where appropriate. Avoid "many" and "most" without a number behind them.
Mixed-Methods Research
Present strands separately, then add a short integration paragraph at the end showing how the quantitative and qualitative findings converge or diverge.
Humanities
"Results" may be merged with analysis. Present evidence (quotes, archive entries, close readings) before drawing argumentative conclusions, even if the two sections aren't formally separated.
FAQs About the Results Section
How long should the results section be?
Usually 15-25% of total word count. Empirical theses with rich datasets can justify more; humanities papers often less.
What tense should I use?
Past tense for what was found ("Group A scored..."). Present tense only for stable references to tables and figures ("Table 1 shows...").
Can I use first person ("we" / "I")?
Yes, increasingly. Some style guides still prefer passive voice. Be consistent with the rest of the paper.
How many tables and figures should I include?
Only as many as you need. A table or figure that you do not refer to in the body is a candidate for the appendix. Two or three well-chosen tables usually beat six mediocre ones.
Do I need to include non-significant results?
Yes. Reporting only significant findings introduces bias. Mention non-significant tests briefly with the relevant statistic.
Should the results section have its own conclusion?
No. End on the last finding. The discussion takes over from there.
Strong results sections feel almost boring to write - that's the point. Plain reporting now leaves room for thoughtful interpretation later.