13 Results Chapter
The results section is a matter-of-fact description of your findings. It is written in past tense and focuses on answering your research questions and testing your hypotheses. Extensive interpretation belongs in the Discussion. Here you only report what you found.
That said, good results sections are well-structured and nonetheless tell a coherent “story” that guides the reader through the analyses step by step. It should usually align with the order of the hypotheses or research questions.
Reporting a finding (“H1 was supported”) is fine. Explaining why it occurred, what it means theoretically, or how it relates to prior literature is not — save that for the Discussion.
Descriptions of your findings in the result section → past tense (“The results showed that…”)
13.1 Structure for quantitative research
Report sections in this order:
13.1.1 1. Descriptive statistics
Report means (M), standard deviations (SD), and correlations among measured variables. Present these in a table or figure. In the text, describe the main and noteworthy correlations. Note whether control variables correlate with (in)dependent variables. If so, you may want to include them in follow-up analyses.
Missing data: Briefly state how missing data were handled (e.g., listwise deletion, multiple imputation) and how many cases were excluded or imputed. Reviewers will expect this.
Scale reliability: For each multi-item scale, report internal consistency using Cronbach’s α or McDonald’s ω (preferred in APA 7). This can appear in the descriptive statistics table or in a brief sentence in text.
Example:
“All scales showed acceptable reliability: social media use (α = .84), privacy concern (ω = .79), and self-disclosure intentions (α = .91).”
13.1.2 2. Manipulation check (for experiments)
Verify that the experimental manipulation worked as intended. Also check randomization: are demographic variables (gender, age, education) balanced across conditions? Report these checks even if no significant differences were found.
Example:
“As a manipulation check, participants in the high-exposure condition reported significantly more perceived peer disclosure than those in the low-exposure condition (M = 4.12, SD = 0.73 vs. M = 2.31, SD = 0.68), t(198) = 17.34, p < .001, d = 2.43, confirming that the manipulation was successful. Randomization checks showed no significant differences between conditions in age, t(198) = 0.84, p = .402, gender distribution, χ²(1) = 0.62, p = .431, or education level, χ²(2) = 1.14, p = .565.”
13.1.3 3. Hypothesis testing (confirmatory analyses)
For each hypothesis:
- Restate the hypothesis briefly
- Name the test used and which variables are included (e.g., “We conducted a linear regression with X as predictor and Y as outcome”)
- Report results with all relevant statistics
- State a brief conclusion (e.g., “H1 was supported” or “H1 was not supported”)
Example (all four steps):
H1 predicted that higher exposure to peer disclosures would increase participants’ own disclosure intentions. We tested this using a simple linear regression with peer disclosure exposure as the predictor and disclosure intentions as the outcome. The analysis revealed a significant positive effect, b = 0.34, SE = 0.08, 95% CI [0.18, 0.50], t(298) = 4.25, p < .001, β = .24. H1 was supported.
Follow the same order as in the Theory chapter. Also briefly report non-significant findings — do not omit them.
13.1.4 3a. Reporting regression models
When reporting a multiple regression, first report the overall model, then individual predictors.
Overall model: Report R², adjusted R², F(df1, df2), p, and — if comparing nested models — ΔR² and its significance.
Individual predictors: For each predictor, report b, SE, 95% CI, t(df), p, and β.
Example:
“A multiple regression was conducted with social media use and age as predictors of well-being. The model was significant, F(2, 297) = 14.32, p < .001, R² = .09, adjusted R² = .08, indicating that the predictors explained 9% of the variance in well-being. Social media use was a significant negative predictor, b = −0.31, SE = 0.07, 95% CI [−0.45, −0.17], t(297) = −4.43, p < .001, β = −.28. Age was not a significant predictor, b = 0.04, SE = 0.03, 95% CI [−0.02, 0.10], t(297) = 1.33, p = .184, β = .07.”
13.1.5 3b. Mediation and moderation (if applicable)
Mediation (indirect effects): Report the indirect effect with a bootstrap confidence interval. If the CI excludes zero, mediation is supported.
Example:
“The indirect effect of social media use on well-being through social comparison was b = −0.12, SE = 0.04, 95% CI [−0.21, −0.04], indicating significant mediation.”
Moderation (interaction effects): Report the interaction term and, if significant, follow up with a simple slopes analysis or plot to illustrate the pattern.
Example:
“The interaction between social media use and gender was significant, b = 0.18, SE = 0.06, t(296) = 3.01, p = .003, β = .15. Simple slopes analysis showed that the effect of social media use was significant for women (b = 0.31, p < .001) but not for men (b = 0.08, p = .21).”
13.1.6 3c. SEM and CFA (if applicable)
If you use confirmatory factor analysis (CFA) to validate your measurement model, report it before hypothesis testing. If you use structural equation modeling (SEM), report model fit before presenting path coefficients.
Fit indices to report: χ²(df), p, CFI, RMSEA (with 90% CI), SRMR. Common thresholds: CFI ≥ .95, RMSEA ≤ .06, SRMR ≤ .08 — but treat these as guidelines, not hard cutoffs.
Example (CFA):
“A CFA confirmed the three-factor structure of the scale. Model fit was acceptable: χ²(84) = 112.43, p = .024, CFI = .97, RMSEA = .04, 90% CI [.01, .06], SRMR = .05. All factor loadings were significant and exceeded .50.”
Example (SEM path):
“The structural model showed a significant direct path from social media use to privacy concern (β = .31, SE = .06, p < .001) and from privacy concern to protective behavior (β = .24, SE = .07, p = .001). Model fit was good: CFI = .96, RMSEA = .05, SRMR = .06.”
13.1.7 4. Secondary (exploratory analyses, usually optional)
If you want to explore the data further (e.g., covariate analyses with control variables), include these in a clearly labeled separate section.
13.2 APA statistical reporting
Follow APA 7th edition formatting:
- Statistical symbols are italicized: M, SD, N, n, F, t, r, p, b, β, χ²
- Round all numbers to two decimal places
- No leading zero on p values, correlations, or standardized effect sizes that have an upper bound of 1: p = .03, r = .45 (not 0.03, 0.45)
- Report exact p values: p = .003 — except use p < .001 for very small values
- Put spaces after = and <
- Report effect sizes (Cohen’s d, η², f², etc.)
- Report 95% confidence intervals where applicable: b = 0.34, 95% CI [0.18, 0.50]
Examples:
The regression analysis revealed a significant positive effect of social media exposure on vaccine hesitancy, b = 0.34, SE = 0.08, 95% CI [0.18, 0.50], t(298) = 4.25, p < .001, β = .24.
An independent-samples t-test showed no significant difference between conditions, t(148) = 1.12, p = .264, d = 0.18, 95% CI [−0.15, 0.51].
The ANOVA yielded a significant main effect of logo color, F(1, 196) = 8.34, p = .004, η² = .04, 95% CI [.01, .10].
Non-significant findings are still findings and should also be reported fully!
13.3 Tables and figures
- Use tables and figures only when they add information that cannot be conveyed efficiently in text
- Format all tables in APA style (not copy-pasted from SPSS output)
- Every table and figure needs:
- A title (above for tables, below for figures)
- Clear axis labels or column headers
- A note if abbreviations are used
- Variable labels in tables must match those used in the method section
13.4 Results in qualitative research
In qualitative results, the chapter demonstrates your analysis and presents empirical findings. The exact structure depends on your analytical approach, but the general logic is the same: organize around your sub-questions or themes, present your analytical interpretation, and support it with data.
| Approach | Typical structure |
|---|---|
| Thematic analysis (Braun & Clarke) | Organized by themes derived from the data |
| Grounded theory | Organized by categories from selective coding |
| Discourse/content analysis | Organized by discursive patterns or code categories |
For each sub-question or theme:
- Introduce the theme or sub-question
- Present the analytical finding (your interpretation, not just a summary)
- Support findings with representative quotes from the data
- Briefly conclude what this finding means for the sub-question
Quotes serve as illustrations — not as substitutes for analysis. Contextualize each quote before and after: explain what it illustrates and why it is representative. Format quotes in italics, indented. Use a respondent ID (e.g., R04, P12) rather than identifying information.
Example:
Participants frequently described feeling pressure to present an idealized version of themselves online. One respondent captured this tension clearly:
“I know it’s not real, but I still spend like 20 minutes picking the right photo. You want people to think you’re doing well, you know?” (R07)
This illustrates the gap between participants’ awareness of impression management norms and their continued compliance with them — a pattern present across 14 of the 18 interviews.
How many quotes? Aim for 2–4 quotes per sub-question. Each quote should be the clearest illustration of its point — prefer fewer, well-contextualized quotes over many loosely connected ones.
On saturation and frequency: You may note how widespread a theme was (e.g., “this pattern appeared across 12 of 18 interviews”), but avoid reducing qualitative findings to counts. The goal is analytical depth, not frequency tallying.
Avoid discussing theoretical implications extensively in the results section — save that for the Discussion.
13.5 Results checklist
13.6 References
- Masur, P. K., DiFranzo, D. J., & Bazarova, N. N. (2021). Behavioral Contagion on Social Media: Effects of Social Norms, Design Interventions, and Critical Media Literacy on Self-Disclosure. PLOS One, 16(7). e0254670. https://doi.org/10.1371/journal.pone.0254670