For students with a draft results section in an empirical psychology thesis or dissertation. The page covers how one test is reported in APA style, what comes before the hypothesis tests, the order of reporting, tables and figures, how planned and exploratory tests are kept apart, and the qualitative variant with themes and quotations. Department guidance is quoted where it sets a rule.
No one sees their own blind spots, with or without AI help. Esmond points them out in your draft from within your field, along with factual errors and inconsistencies.
The results section answers one question per hypothesis: what did the test show? It reports in the order the hypotheses were stated, with enough statistics for a reader to check each conclusion. The guidelines of the Cognitive Modelling and Decision Neuroscience Lab at the University of Hamburg set the boundary: “In the results section, all major findings of the paper are presented, but not yet interpreted in terms of the research question.”
Departments draw that boundary in slightly different places. The Psychological Methods group at Ruhr University Bochum asks students to describe each result substantively as well as statistically, and to say directly what it means for the hypothesis. Both fit one working rule: stating that a result is or isn't consistent with Hypothesis 2 is reporting; explaining why, or what it means for theory, is discussion. For the general line between the two chapters, see results vs. discussion; for what goes after it, see the psychology discussion section.
A hypothetical example of a commentYour text reports t(62) = 2.41 for Hypothesis 1, but Table 2 lists 33 and 32 participants per group, which would give 63 degrees of freedom. Which of the two is correct?
A first comment on one part of your own draft, right away and without an account.
Get a free previewA complete report of a group comparison gives the descriptive statistics, the test, and the size of the effect. For example: the intervention group (M = 2.9, SD = 0.8, n = 32) reported less stress than the waitlist group (M = 3.4, SD = 0.86, n = 32), t(62) = 2.41, p = .019, d = 0.60, 95% CI [0.10, 1.10]. Each element tells the reader something different.
A results section usually opens with what happened to the data before testing: how many cases were excluded and why (if the method section hasn't already said so), how missing values were handled, and how outliers were defined. A table of means, standard deviations, and, for correlational designs, correlations between all study variables follows.
Assumption checks come next, briefly. For a t-test or ANOVA that means the distribution of residuals and the equality of variances; for regression, also linearity and multicollinearity. Report what you checked and what you did when a check failed. If you switched to a Welch test because variances differed, the degrees of freedom are no longer whole numbers, for example t(58.3), and the text should say why.
Report the confirmatory tests first, one per hypothesis, in the order of the introduction and under the same variable names. If several tests address one hypothesis, state the correction for multiple comparisons (Bonferroni or Holm, for example) or the reason for none.
Tests you didn't plan go in a separate, labeled part. SWPS University's standards for master's theses in psychology allow additional relationships beyond the hypotheses, provided the thesis states clearly that those tests are exploratory and were run ad hoc. If the study was preregistered, every change to the planned test belongs here or in the method section; the University of Graz's differential psychology group requires deviations from the preregistration to be indicated in the master's thesis.
Non-significant results are reported with the same detail as significant ones. The Hamburg guidelines say that all results should be reported in full even when the test statistic is non-significant.
A table carries the numbers; the text points to the result that answers the hypothesis. Bochum's information sheet says: “If you report extensive result tables, you do not need to discuss every single value in the text.” SWPS University's standards add that the same numerical values should not appear both in a table and in the text.
In APA style, each table and figure has a number and a title, is mentioned in the text before it appears, and explains abbreviations and significance symbols in a note. A figure with error bars says whether they show standard errors or confidence intervals. The University of Warsaw lists the matching learning outcome for its master's thesis in psychology: a student “Communicates about data and conclusions from research understandably, using adequate indicators, tables, and graphs.”
In a qualitative psychology thesis (thematic analysis, interpretative phenomenological analysis), the results section presents the themes, each with a definition and quotations from participants. SWPS University's standards ask that qualitative results be supported by quotations from participants' statements, and that contradictory or deviant examples, the so-called negative cases, are included.
A reader checks that each quotation actually supports the theme it illustrates, that quotations come from more than one or two participants, and that words such as "most participants" match counts you could give. Whether the literature belongs in this section depends on your approach and your department; some qualitative theses combine results and discussion.
The University of Zurich's Chair of Developmental Psychology: Adulthood lists this among its criteria for the results of a master's thesis: “The presentation of results is clearly separated from the interpretation and discussion.” The Hamburg guidelines recommend reporting effect sizes and possibly confidence intervals alongside test statistics and p-values, and exact p-values unless they are below .001.
Bochum's group and Hamburg's lab also set rules on AI: Bochum permits AI tools only as assistants and requires each use to be documented in the thesis, and Hamburg requires any use of generative AI to be transparent and documented. Where rules like these apply, ask your supervisor first whether you may use Esmond on your results, and document the use as required.
| Check | What it catches |
|---|---|
| Every hypothesis has a test, in the order of the introduction | A hypothesis that disappears, or a test that answers a different question |
| Degrees of freedom match the sample sizes in the method section and tables | Unreported exclusions or a typo in a test |
| Each test has statistic, df, exact p, effect size, and CI | Results that can't be weighed or compared with other studies |
| Assumption checks are reported with the action taken | A test whose conditions weren't met, with no correction |
| Exploratory tests are labeled and kept apart | Unplanned findings presented as confirmed |
| The text states results without explaining them | Interpretation that belongs in the discussion |
| Numbers in text, tables, and abstract agree | A value changed in one place and not in the others |
These fragments are made up. None of them comes from a real thesis.
Esmond evaluates every assignment before you submit. It never writes it.
Esmond is an AI tool that doesn't write for you. If your course doesn't allow any AI tools, don't use Esmond for that work. If you're unsure, ask your instructor or advisor.
In a results section, Esmond can flag inconsistencies in your own text that a psychology reader would query: degrees of freedom that don't fit the reported group sizes, a value in the text that differs from the table, a hypothesis with no test, a test without an effect size, an unplanned comparison presented as confirmed, or interpretation where a result should simply be stated. The sample report shows how such comments are laid out, and how Esmond works describes the process. Other chapters are covered from thesis chapter feedback.
Esmond reads what you upload. If the file holds the whole draft, it can set this chapter against the others; if it holds one chapter, the review stays within that chapter and may raise a point that another chapter already answers.
Esmond reads the statistics as reported; it doesn't rerun your tests or open your data, and it doesn't check your tables against APA formatting rules or your department's template. Which points you act on is your decision.
Data preparation and exclusions if not covered earlier, descriptive statistics, assumption checks, then one test per hypothesis in the order of the introduction, followed by any exploratory tests under their own label. In a dissertation with several studies, each study has its own results section.
Give the group means and standard deviations, then the test with degrees of freedom, the exact p-value, and an effect size with its confidence interval, for example t(62) = 2.41, p = .019, d = 0.60, 95% CI [0.10, 1.10]. Values below .001 are reported as p < .001.
State whether each result is consistent with the hypothesis, and leave explanations and links to theory for the discussion. Hamburg's guidelines present findings without interpreting them in terms of the research question; Bochum's ask students to say directly what a result means for the hypothesis. Follow your department's version.
Yes. Report the test, the p-value, and the effect size with its interval, as for any other result. Hamburg's guidelines ask for all results in full even when they are non-significant. How much text they get can differ by department.
Values a reader needs and the text doesn't repeat: means and standard deviations per group or condition, correlations between all study variables, or regression coefficients with their standard errors and intervals. Number the table, give it a title, and explain abbreviations in a note.
Read on November 8, 2026.
Get your paper evaluated before you submit