To write the statistics in your methods section, state which statistical test you used, why you chose it, the variables it was applied to, the alpha level (usually .05), and the software you ran it in — all in past tense, before you report any results. The methods section tells the reader what you did and how you'll analyse it; the actual numbers (your p values and effect sizes) belong in the results section. Most students lose marks here because they either blur the two sections together or forget to justify their test choice.
Key Takeaways
- The statistical analysis subsection of your methods describes your plan — the tests, variables, and software — not your findings.
- Every test you mention must be justified by your research question and your data type (e.g. "an independent-samples t-test was used to compare the two groups").
- State your significance threshold explicitly: "Statistical significance was set at p < .05."
- Name the software and version you used (e.g. SPSS 29, R 4.3, or StatRyx) so your analysis is reproducible.
- Write the whole subsection in the past tense and in APA 7 style, with test statistics and p italicised.
What goes in the statistics part of a methods section?
The statistical analysis subsection of your methods section should contain five things: the tests you ran, the justification for each, the variables involved, your alpha level, and the software used. A clean version reads like a recipe — anyone with your data should be able to reproduce your analysis exactly.
Here is the standard order:
- Descriptive statistics — how you summarised your data (means, standard deviations, frequencies).
- The inferential test(s) — the specific test chosen to answer each research question or hypothesis.
- The justification — why that test fits your design and data type.
- Assumption checks — how you tested for normality, homogeneity of variance, etc.
- Alpha and software — your significance threshold and the program (and version) you used.
Crucially, no results go here. "A t-test was used" belongs in methods; "t(48) = 2.31, p = .025" belongs in results.
How do I justify my statistical test choice?
You justify a statistical test by linking it to your research design, your independent and dependent variables, and their measurement levels. Examiners want to see that you didn't pick a test at random — you matched it to the question and the data.
A justification sentence usually follows this pattern: "A [test] was conducted to [compare/examine the relationship between] [variable(s)], because [variable type/design reason]."
Examples:
- "An independent-samples t-test was conducted to compare mean anxiety scores between the treatment and control groups, as the independent variable was dichotomous and the dependent variable was continuous."
- "A Pearson correlation was used to examine the association between sleep duration and exam performance, as both variables were continuous and approximately normally distributed."
- "A one-way ANOVA was conducted to compare reaction times across the three dosage groups."
If you're unsure which test matches your variables, our guide on how to choose the right statistical test walks through it by data type — or StatRyx can pick the test for you from the structure of your data.
How do I report assumption checks in the methods section?
State in the methods section which assumptions you checked and how, so the reader knows your test was appropriate before you report the outcome. You describe the procedure here; the actual assumption results can go in your results section.
Example wording:
"Prior to analysis, the normality of residuals was assessed using the Shapiro–Wilk test, and homogeneity of variance was examined using Levene's test. Where assumptions were violated, a non-parametric alternative (the Mann–Whitney U test) was used."
This single sentence signals three things examiners look for: you knew the assumptions existed, you tested them, and you had a fallback plan. If you're weighing a parametric test against a non-parametric one, see our explainer on Mann–Whitney vs the t-test.
A worked example: writing a full statistics subsection
Imagine a thesis comparing memory-recall scores between 50 participants who slept 8 hours versus 50 who slept 5 hours, plus a correlation between sleep and recall. Here is how the statistical analysis subsection would read:
Statistical Analysis
All analyses were conducted in StatRyx (or SPSS version 29). Descriptive statistics (means and standard deviations) were calculated for recall scores in each sleep condition. An independent-samples t-test was used to compare mean recall scores between the 8-hour and 5-hour sleep groups, as the independent variable was dichotomous and the dependent variable was continuous. The assumption of normality was assessed with the Shapiro–Wilk test, and homogeneity of variance with Levene's test. A Pearson correlation was then conducted to examine the relationship between total sleep duration and recall score. Statistical significance was set at p < .05, and effect sizes (Cohen's d and r) were reported to indicate the magnitude of effects.
Notice what's absent: no t value, no p value, no correlation coefficient. Those numbers — e.g. t(98) = 3.12, p = .002, d = 0.62 — appear only in the results. This separation is the single most common fix supervisors request.
Methods vs Results: what goes where?
| Element | Methods section | Results section |
|---|---|---|
| Name of the test | ✅ "A one-way ANOVA was conducted…" | ❌ |
| Justification for the test | ✅ "…because the DV was continuous." | ❌ |
| Alpha level (.05) | ✅ "Significance was set at p < .05." | ❌ |
| Software and version | ✅ "Analyses were run in SPSS 29." | ❌ |
| Which assumptions you checked | ✅ "Normality was assessed via Shapiro–Wilk." | Assumption outcomes can appear here |
| Test statistics and p values | ❌ | ✅ F(2, 97) = 4.31, p = .016 |
| Effect sizes | Mention you'll report them | ✅ η² = .08 |
| Tables and figures of results | ❌ | ✅ |
What software should I report — and does it matter?
Yes, you must name your statistical software and version, because reproducibility is a core requirement of a credible methods section. SPSS is the academic incumbent — a single-user licence runs to roughly £80–£100 per year through many universities, and it's cited in tens of thousands of theses — but it's not your only option.
| Tool | Cost | Best for | Picks the test for you? | APA-ready output |
|---|---|---|---|---|
| StatRyx | Free | Non-statisticians who want the right test chosen + the write-up done | ✅ Yes | ✅ APA 7 |
| SPSS | ~£80–£100/yr | Established labs, point-and-click | ❌ No | Partial |
| R | Free | Full control, custom analyses | ❌ No (code) | Via packages |
| JASP | Free | Bayesian + frequentist, desktop | ❌ No | ✅ Good |
| jamovi | Free | Clean desktop interface | ❌ No | ✅ Good |
| Stata | Paid | Econometrics, large datasets | ❌ No | Partial |
StatRyx is an AI-powered statistical analysis tool that replaces manual SPSS workflows with automated, APA 7-formatted reporting. The practical advantage for your write-up: it selects the appropriate test from your data, runs it, checks assumptions, and generates APA-formatted output you can paste straight into your methods and results sections — which removes exactly the two steps students stumble on.
**Stop calculating this by hand — run it