For students with a draft method section in an empirical psychology thesis or dissertation who want to know what a reader will test in it. This page covers the parts that are specific to psychology: how the sample size was justified, whether the measures are reliable in your own sample, whether the procedure can be repeated, and how deviations from a plan or preregistration are reported.
From within your field, Esmond points out where your findings claim more than your method can show and where your design choices contradict each other.
In empirical psychology, the method section is the record a reader uses to judge every number that follows. If it is vague about who took part, how a construct was measured, or which test belongs to which hypothesis, the results can't be checked. The University of Warsaw's Faculty of Psychology puts the standard plainly in its rules for the master's thesis: “The empirical part should describe the selection of indicators and present the research procedure in such a way that it can be replicated by an independent team.”
The usual APA subsections are Participants, Measures or Materials, Procedure, and often Design and Data Analysis. The general questions of any method chapter, such as method versus methodology, are on the page about the thesis methodology chapter. For where this section sits among the other parts, see the overview of the psychology thesis. This page stays with what a psychology reader looks for inside the section.
A hypothetical example of a commentYour Participants section reports 112 students, yet your first independent-samples t-test in the results has 104 degrees of freedom, which fits 106 participants. Where does the method section account for the other six?
A first comment on one part of your own draft, right away and without an account.
Get a free previewA reader follows the sample from the first contact to the final analysis: how many people were invited or signed up, how many started, how many were excluded and for which reason, and how many remain in each test. The degrees of freedom in your results are a quick test of that chain. An independent-samples t-test with 120 participants has 118 degrees of freedom; a correlation with 60 participants has 58. If the numbers differ, the method section owes the explanation.
Exclusion rules need a date as well as a reason. A rule set before the data were seen (for example, a failed attention check, or a response time under a set limit) is part of the design. A rule chosen after looking at the results changes the test. SWPS University in Poland, in its standards for master's theses in psychology, calls it good practice to formulate exclusion criteria before the study or the data analysis begins, and to report how many people took part and how many were included in the analysis.
Describe the sample on the characteristics that matter for the question: age, gender, and the population it was drawn from (psychology undergraduates, a clinical sample, an online panel). That description is what the discussion will later need when it says how far the findings extend.
Statistical power is the probability that a test detects an effect of a given size, if an effect of that size exists, at the chosen alpha level. A power analysis turns that into a sample size, and it needs four inputs: the test, the alpha level, the target power, and the effect size you expect or care about. For a two-sided independent-samples t-test with alpha = .05 and power = .80, an expected effect of d = 0.50 requires 64 participants per group, 128 in total. Halve the expected effect and the required sample roughly quadruples.
The weakest input is usually the effect size. Taking it from one published study tends to overestimate it, and copying that study's sample size is no safer. The University of Guelph's Department of Psychology warns: “Note that given that most psychology studies typically have statistical power of less than .50, looking at the sample size of a previous study to set your sample size is generally discouraged.” Alternatives are a smallest effect size of interest, or an estimate taken from a meta-analysis and adjusted downward.
If the sample was fixed by circumstance (a school with 90 pupils, a clinic with 40 patients), say so, and report what effect size that sample could detect with reasonable power. That sentence gives your later discussion of a non-significant result something to stand on.
For each instrument, a reader expects the construct it measures, the number of items, the response format, how the total was computed, a reference for the original version, and evidence that it works in your data. The guidelines of the University of Hamburg's Cognitive Modelling and Decision Neuroscience Lab state one part of this directly: information on the reliability of the measures usually belongs in the method section.
Reliability from the original validation study describes that study's sample. Your sample may differ in age, language, or culture, so report an internal-consistency coefficient (Cronbach's alpha or McDonald's omega) computed on your own data. For a translated or shortened scale, say who translated it, how, and whether the factor structure was checked. If the introduction defines a construct one way and the scale measures something narrower, the method section is where a reader will notice.
For an experiment, the equivalent is the manipulation: what exactly differed between conditions, and whether a manipulation check showed that it worked.
The procedure describes what a participant experienced, in order: consent, instructions, the tasks or questionnaires in their sequence, any deception and the debriefing that followed, and how long it took. Standard instruments can be cited; anything you built yourself needs enough detail to rebuild it.
Ethics belongs here in concrete form: which committee gave approval, the reference number, and that approval came before data collection. The guidelines of the Cognitive Modelling and Decision Neuroscience Lab at the University of Hamburg ask for the ethics approval to be mentioned in the Participants part, together with a justification of the sample size, ideally a formal power analysis.
The analysis plan links each hypothesis to one test, names the software and version, and says how missing data and assumption violations will be handled. If the study was preregistered, the method section points to the preregistration and reports every change. The differential psychology group at the University of Graz is explicit about this: “If you have deviated from the preregistration when carrying out your examination or evaluation, this must be indicated in the Master's thesis.”
| Part | What a reader asks | Where drafts often leave a gap |
|---|---|---|
| Participants | Do recruitment, exclusions, and the analyzed N add up, and do they match the degrees of freedom? | Exclusions reported only as a final number, with no reason or rule |
| Sample size | Which test, alpha, power, and effect size went into the calculation, and where does the effect size come from? | "A power analysis was conducted" with none of the inputs |
| Measures | Is each construct in the hypotheses measured, and how reliable was each measure in this sample? | Reliability quoted only from the original validation study |
| Manipulation | What differed between conditions, and did a check show it worked? | A manipulation described in one sentence, with no check |
| Procedure | Could another team run the same session from this text? | Order of tasks, timing, or debriefing left out |
| Ethics | Which body gave approval, with which reference, and when? | "Ethical approval was obtained" with no detail |
| Analysis plan | Is there one planned test per hypothesis, and are changes to the plan reported? | Tests in the results that the plan never mentions |
At the University of Waterloo, the honours thesis handbook in psychology lists the required parts of the method section for an empirical project: “methods (participants, design, task or test to be given, testing procedures, measures)”. The Psychological Methods group at Ruhr University Bochum asks for the sample description to include a “justification of sample size or power analysis”, next to the design, the measurement instruments, and the statistical methods.
SWPS University's standards for psychology master's theses go further for experimental and correlational studies: the sample size should be adequate for the expected effect size, and the decision process that led to the number of participants should be described in the thesis under Method / Participants. They name tools such as G*Power as good practice for setting the sample size before the study.
Two of the guides cited on this page also set rules on AI. Bochum's group allows AI tools only as assistants and requires every use to be documented in the thesis; Hamburg's lab requires any use of generative AI to be transparent and documented. If your department has a rule like this, ask your supervisor before you use Esmond on your method section, and record the use the way your rules ask.
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 method section, Esmond can flag what a psychology reader would question in the text you made: participant numbers that don't add up or don't match the degrees of freedom elsewhere in the draft, a power analysis without its inputs, reliability reported only from another study, a hypothesis variable with no measure, an exclusion rule with no reason. The sample report shows the format of such a review, and how Esmond works explains the steps. For the other chapters, see 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 doesn't rerun your power analysis or your statistics, doesn't verify your design, and doesn't check your ethics approval or your department's template. You decide which points to act on.
In the guides quoted on this page: participants (with recruitment, exclusions, and a sample-size justification), the design, the measures or materials, the procedure, and the analysis plan. Waterloo's honours handbook lists participants, design, the task or test, testing procedures, and measures. A dissertation with several studies repeats this for each study.
Several of the guides on this page ask for a justification of the sample size, with a power analysis as the preferred or named form: Bochum, Hamburg, and SWPS University among them. If your sample size was fixed by circumstance, report that and state what effect size the sample could detect. Your department's guidelines decide what is required.
From your own data, with the original value as background if you like. Reliability depends on the sample, so the value from a validation study doesn't show how the scale performed in yours. McDonald's omega is an alternative some departments prefer.
In the method section, next to the part of the plan that changed, with the reason. The University of Graz's differential psychology group requires deviations to be indicated in the master's thesis and the revised preregistration to be added to the appendix.
An example shows the order of subsections and the level of detail. Your own section has to describe your own study, so the useful check is whether a reader could repeat your study from your text. The made-up fragments above show the gaps a reader tends to find.
Read on November 8, 2026.
Get your paper evaluated before you submit