Results vs discussion: the line that gets crossed
The single most common committee feedback on results chapters is: "This is discussion, not results. Save interpretation for Chapter 5."
The line is simple in theory, hard in practice. A results sentence reports a finding and a statistic. A discussion sentence interprets why that finding matters, what it means in context, or how it relates to prior research.
Here are two sentences about the same hypothetical experiment (a study of student study habits and final grades):
Students who studied in groups reported an average final grade of 82.4 (SD = 8.1) compared to 76.3 (SD = 9.7) for solo learners; this difference was significant, t(58) = 2.31, p = .024, d = 0.61.
Group study may enhance performance because peer explanation reinforces conceptual understanding. Our finding suggests that collaborative learning environments could be a cost-effective intervention to boost student achievement.
The first sentence: states a finding (group learners scored higher), reports the numbers and descriptive statistics (means and standard deviations), and includes the statistical test and its outcome (p-value and effect size). It does not say why. The second sentence: interprets why the difference exists, speculates on mechanism, and draws implications. It belongs in discussion, not results.
Your committee will accept results that seem obvious—"high group grades support collaboration"—as long as you show the number, not tell the meaning. The moment you add "this suggests," "this implies," or "this may reflect," you have crossed into discussion.
Organizing results
Organize your findings in a way that mirrors your research design and makes patterns visible. There is no single correct order; choose one and be consistent.
| Organizing principle | Best for | Example order |
|---|---|---|
| By research question or hypothesis | Quantitative studies with distinct, numbered RQs | RQ1 results, then RQ2, then RQ3. Subheadings for each. |
| By methods chapter order | Mixed-methods studies or multi-phase projects | First present pilot results, then main study results, then follow-up analysis. |
| By theme or construct | Qualitative or interview-based studies | Present all findings related to "barriers to entry," then all related to "support systems," etc. |
| By population or subgroup | Comparative studies (control vs. treatment, cohort A vs. cohort B) | Demographic profile of each group, then findings for Group 1, then Group 2. |
Within each section, use a logical cascade: demographics or sample characteristics first, then overall findings, then subgroup or condition-specific breakdowns. This gives your reader a frame before you layer on complexity.
Reporting quantitative results
A well-reported statistical finding contains six elements. Every inferential statistic you report should have all of them.
- The test used: t-test, ANOVA, Pearson correlation, chi-square, Mann-Whitney U, etc.
- Descriptive statistics: means and standard deviations (or medians and ranges for non-normal data).
- Degrees of freedom: the sample size and structure information the test requires (e.g., df = 58 for a t-test; F(2, 57) for ANOVA).
- The test statistic: the value your analysis produced (t = 2.31, F = 4.15, r = 0.48).
- The p-value: .024, .001, or > .05 (exact or threshold, depending on your field and journal).
- Effect size: Cohen's d for t-tests, eta-squared for ANOVA, r for correlations. Essential for interpreting practical significance in discussion.
Here are two fully worked examples using realistic hypothetical data:
Online learners (M = 73.8, SD = 11.2) scored significantly lower on the midterm exam than in-person learners (M = 81.6, SD = 9.4), t(142) = −3.48, p = .001, d = −0.71. This represents a medium-to-large effect size.
Time spent on homework was positively correlated with course grade (r = 0.62, 95% CI [0.45, 0.75], p < .001). This strong relationship held across all four course sections.
Notice that both examples follow the same order: descriptive data first (means and SDs, or the correlation coefficient), then the test statistic and p-value, then effect size or confidence interval. Many fields now expect confidence intervals instead of or in addition to p-values; check your discipline's style guide.
For non-significant results, report them the same way. "There was no significant difference in motivation between male and female participants, t(88) = 1.12, p = .266" is a complete, reportable finding. Do not omit the test statistic or p-value because the result was null.
Reporting qualitative results
Qualitative results present themes, patterns, and representative quotes. Your job is to show how you arrived at each theme and let the data speak.
The structure is: theme label → brief definition → one or more participant quotes → minimal gloss (context only, not interpretation). Anonymize participants with codes (P01, P02) or pseudonyms (Jordan, Alex). Do not name participants or provide identifying details unless your ethics approval explicitly allows it and your institution's handbook requires it.
Here is a worked example from a hypothetical interview study on graduate student motivation:
Advisor support as a pivot point. Five of eight participants named their advisor's regular feedback as crucial to staying on track. One participant described the turning point this way:
"I was ready to quit after my second committee meeting. But my advisor sat down and said, 'This is fixable. Let's break it into three smaller pieces.' That one conversation changed how I saw the entire project." (P03)
Another highlighted the inverse: "I maybe saw my advisor three times my whole second year. By the time I asked for help, I felt too far behind to catch up." (P07)
Participants distinguished between feedback that clarified the path forward (described as helpful) and critical-only feedback without concrete next steps (described as demoralizing).
The callout above shows all three elements: the theme ("Advisor support as a pivot point"), two quotes that illustrate it, and a brief summary of what the quotes illustrate. The gloss does not interpret the psychological mechanism—it simply frames what the data shows.
Use this same pattern for every major theme. You may have 4–8 themes in a results chapter for an interview study, depending on data volume and research questions. Each theme deserves at least one quote and a 2–3 sentence gloss.
Tables and figures
Tables and figures must follow three rules, consistently applied:
- Every table and figure has a number and a caption. Table 1, Table 2, Figure 1, Figure 2, etc. Captions go above tables, below figures. Each caption is a complete sentence or phrase that stands alone—a reader should understand the data's purpose without re-reading the text.
- Every table and figure is referenced in the text before it appears. Write "As shown in Table 1, group A outperformed group B" and then place Table 1. Do not embed a table or figure without a reference sentence immediately preceding it.
- Do not duplicate the same data in both a table and a figure. If you show means and standard deviations in a table, do not repeat them in a bar graph. Choose one. A table may present raw numbers; a figure may show a trend or comparison the table does not emphasize—but not both.
Captions should be concise but complete. Good: "Table 2. Mean exam scores and standard deviations by course modality (online vs. in-person)." Weak: "Exam scores." (Captions are not titles; they are part of your result, not decoration.)
Number all tables and figures sequentially across your entire results chapter. Do not restart numbering per section. This keeps your cross-references unambiguous.
Common mistakes in results chapters
- Interpreting instead of reporting. "These results indicate that group study improves performance" is interpretation. "Students in group study conditions scored 6 points higher on average" is reporting. Save the first for discussion.
- Including unreferenced tables or figures. Every table and figure must be named and referenced in the text before it appears. A floating table with no introduction baffles readers and breaks the flow. Write the reference sentence first.
- Inconsistent decimal places. If you report one mean to two decimal places, report all means to two decimal places. If one p-value is .024 and another is .05, show both to three decimal places (or write .050). Consistency signals rigor and makes scanning easier.
- Cherry-picking only statistically significant results. Omitting null or non-significant findings because they "didn't work" is a form of cherry-picking. Report what you found, significant or not. Your reader needs the full picture.
- Omitting negative results or unexpected findings. If you hypothesized that gender would moderate the effect of study mode on grades and it did not, say so. Null results are real results and belong in the results chapter.
Frequently asked questions
Can results and discussion be combined into one chapter?
Some institutions and advisors allow a combined Results & Discussion chapter. Check your graduate handbook and ask your committee chair. If allowed, use clear subheadings to distinguish results (findings with statistics) from discussion (interpretation and implications). Many dissertations still separate them, and that division enforces the thinking you need: one chapter asks "what did I find?" the next asks "what does it mean?" If your handbook does not forbid combining them, err toward separation unless your advisor requests otherwise.
Should I write results in past or present tense?
Use past tense for your study and its findings: "Students scored higher," "We observed a correlation." Use present tense for established facts or generalities: "Higher homework completion is associated with better outcomes" (present tense because it is a general principle, not specific to your study). Most results chapters use past tense throughout; be consistent within your chapter and check your field's style guide (APA, Chicago, your discipline's handbook).
How many decimal places should I report?
Report p-values and effect sizes to two or three decimal places (.024, .001, or .050). Report means and standard deviations to one decimal place more than your raw data (if you measure grade point average to two decimals, report means to three: 3.452). Report correlations to two decimal places (.62, .48). Check your field's style guide; APA Publication Manual gives detailed guidance. The goal is precision without false confidence—reporting a mean to five decimal places overstates your precision.
Do all research questions need their own subsection?
Use subsections for clarity and organization, especially if you have three or more research questions or hypotheses. Each subsection should have its own heading and group all results pertaining to that question before moving to the next. If you have only one or two questions, a single flow may work. Ask your advisor whether your committee prefers subsections; when in doubt, organize by RQ. This mimics your methods chapter and makes your results easier to follow.
What if my data are messy or incomplete?
Report what happened. "Five of 120 participants did not complete the follow-up survey; analyses were conducted on the remaining 115." "Responses to the open-ended question were analyzed by two independent coders; inter-rater reliability was κ = 0.78, indicating substantial agreement." Transparency about limitations strengthens your results, not weakens them. Your committee expects real-world data; missing the occasional response or coding disagreement is normal. Document it, handle it consistently, and move on.
Sources
- APA Style — Official guidance on reporting statistical results, formatting, and academic writing standards.
- Purdue OWL: APA Style — University resource for APA formatting and results section guidelines.
- Purdue OWL: Results Section — Detailed guidance on writing results sections of research papers and theses.
- Duke University Libraries: Thesis & Dissertation Resources — Academic guidance on thesis structure and chapter organization.
- University of North Carolina Writing Center: Writing in Psychology — Discipline-specific guidance on reporting quantitative results and APA style.