The one-line version: a methodology chapter is not a description of what you did. It is an argument that what you did was the right way to answer your question, written so precisely that a stranger could repeat it. Everything below serves those two jobs.
The chapter that gets sent back usually fails in the same way: it narrates the project chronologically, including the false starts, and never states why any single choice was the correct one. The ten sections below are in the order a Canadian committee reads them, and each one has a worked sentence you can adapt.
Step 1: Open with one sentence naming the approach
Before any detail, the reader needs to know what kind of study this is. One sentence, no hedging.
This study used a cross-sectional quantitative survey design to test whether workload-reduction interventions are associated with lower reported burnout among secondary-school teachers in two Ontario public boards.
Notice what that sentence contains: design, method, relationship being tested, population and setting. If yours cannot be written in one sentence, the problem is not the writing.
Step 2: Derive the design from the question, and name what you rejected
This is the step that separates a passing chapter from a strong one. Show that the design follows from the question rather than from what you were comfortable with, then dispose of two alternatives in two sentences each.
A randomised design was not available because assignment to workload conditions is made by boards for operational reasons and cannot be manipulated by a researcher. A longitudinal design would have addressed the causal question directly, but the programme’s twelve-month funding cycle placed follow-up outside the study period; the implications of this are addressed in the limitations.
Committees ask “why didn’t you do X?” precisely because the chapter did not say. Answer it in advance and the question disappears.
Step 3: Define the population, the sample and the number
Three separate things, in this order: who is eligible, how you reached them, and how the number was arrived at.
- Eligibility criteria, written as inclusion and exclusion lists, not as prose.
- Sampling strategy named properly — convenience, purposive, stratified, snowball — with the reason.
- The number, justified. For quantitative work this means an a priori power analysis with its parameters reported. For qualitative work it means an argued judgement rather than an assertion.
The justification is the part that gets challenged. Both versions are written out in full, with the software parameters and the current qualitative guidance, in our answer to how many participants a thesis needs. Do not write “a sample of 120 was deemed sufficient”; a committee reads “deemed” as “guessed”.
Step 4: Get the ethics timing right, because it governs everything else
This is a scheduling constraint disguised as a paperwork step, and it is the one that costs Canadian students entire terms.
TCPS 2 (2022), Article 6.11, states:
Researchers shall submit their research proposals, including proposals for pilot studies, for REB review and approval of their ethical acceptability prior to the start of recruitment of participants, data collection, access to data or collection of human biological materials.
Three consequences that students routinely get wrong:
- A pilot is data collection. Sending a draft questionnaire to five classmates “just to test the wording” is inside the rule, not outside it.
- “Access to data” includes secondary data. Analysing an existing identifiable dataset is covered even though you never meet a participant.
- There is a genuine carve-out, and it is narrow. The same article states that REB review is not required for the initial exploratory phase intended to establish research partnerships or to inform the design of a proposal, and that preparatory work such as observing a setting, taking notes or setting up equipment may precede review. If you later want to use anything gathered in that phase, that is a different matter.
Chapter 6 also sets out two levels of review: full REB review is the default requirement, with delegated review available where the research is determined to be of minimal risk. Which one you get changes your timeline by weeks, so ask your REB office early. What TCPS 2 requires of your study overall, and when it does not apply at all, is covered in our guide to REB approval and TCPS 2 in Canada.
Write the clearance into the chapter as a fact: “The study received approval from the [institution] Research Ethics Board on [date], certificate [number].”
Step 5: Describe recruitment exactly as it was approved
Recruitment gets its own short section because it is where the approved protocol and the actual study most often diverge. State the channel, the wording route, the timeframe, and the response outcome.
Participants were recruited between 14 January and 28 February 2026 through an invitation circulated by each board’s professional-learning office. The invitation text was the version approved by the REB. Of 1,240 teachers who received it, 214 opened the survey and 187 completed it, a completion rate of 87.4% among those who began.
Report both denominators. “187 teachers participated” tells a reader nothing about who did not.
Step 6: Name every instrument, its licence and its reliability evidence
For each measure: what it is, who published it, how many items, the response format, the reported reliability in the source, and the reliability you obtained. Then two things students forget.
- State the permission position. Standardized tests, questionnaires, forms and surveys are copyrighted works, and many are licensed commercially. Whether you may reproduce the full item set in an appendix is a separate question from whether you may administer it, and it has to be settled against the licence you agreed to when you obtained the instrument — not at deposit week.
- Report your own reliability, not the manual’s. An alpha from the original validation sample says nothing about your sample.
For an instrument you built yourself, the burden is higher: describe how items were generated, who reviewed them, what the pilot changed, and what evidence of validity you can offer. “The questionnaire was developed by the researcher” is a sentence that invites a difficult afternoon.
Step 7: Write the procedure so a stranger could repeat it
Chronological order, with durations, in the second half of the chapter where a reader expects it.
Each participant completed the survey online in a single session lasting approximately 18 minutes. The consent page appeared first and required an affirmative response before any item was displayed. Items were presented in fixed order within blocks, with block order randomised. No incentive was offered.
The test to apply: could a competent stranger with your materials produce the same dataset? If any sentence would leave them guessing, it needs a number.
Step 8: Say where the data live, who can reach them, and for how long
Canadian committees and REBs now read this section closely, and it is short enough that there is no excuse for leaving it vague. Four facts:
- Storage location and jurisdiction. Name the service and where its servers are. If any participant material is processed outside Canada, say so; it is a fact, not a confession, but it must be disclosed and it must match what the consent form said.
- Access. Who can open the identifiable file — usually you and your supervisor, and no one else.
- De-identification. When the link between identifiers and responses is broken, and where the key is held until then.
- Retention and destruction. The period your REB approved, and the method.
If your study uses Statistics Canada microdata rather than data you collected, this section becomes an access-tier statement instead: which tier you used, what you signed, and what that permits you to publish. The route through public tables, PUMFs, RTRA and the Research Data Centres is set out in our guide to getting Statistics Canada data for a thesis.
Step 9: State the analysis plan before any result appears
The methodology chapter names the tests. The results chapter reports them. Keeping that boundary is the single easiest way to look disciplined.
Analyses were conducted in [software, version]. Descriptive statistics were computed for all variables. The primary hypothesis was tested with a one-way analysis of variance comparing burnout scores across the three intervention conditions, with an alpha of .05. Assumption checks and the treatment of violations are described below. Missing data were handled by [method], affecting [n] cases.
Two things belong here and almost never appear:
- What you will do if an assumption fails, decided in advance. Choosing the test after seeing which one gives a significant result is a different activity with a different name.
- Your missing-data rule, stated before you know how much is missing.
If you have not chosen your software yet, the licensing and portability trade-offs are in our comparison of SPSS, R, jamovi and Stata for a Canadian thesis, and the full run for the most common omnibus test — assumptions, post hoc choice and the finished table — is in our walkthrough of running and reporting a one-way ANOVA in SPSS.
Step 10: Write the deviations paragraph almost nobody includes
Every real study departs from its protocol somewhere. Recruitment stalls and the window is extended; an item is dropped; a site withdraws. The instinct is to write the chapter as though the plan had been followed. Do the opposite.
Two departures from the approved protocol occurred. Recruitment was extended by three weeks after the initial invitation produced 61 responses; the extension was approved by the REB on [date]. One scale item was removed before analysis because 22% of respondents left it blank and free-text comments indicated the wording was ambiguous; analyses are reported with and without the item in Appendix C.
Two reasons this helps rather than hurts. Methodologically, an examiner who finds an undisclosed departure stops trusting the rest of the chapter. Procedurally, TCPS 2 Article 6.15 requires researchers to report to the REB any unanticipated issue or event that may increase the level of risk to participants or that has other ethical implications affecting their welfare — so for some departures, telling someone is not optional.
The order to write it in, which is not the reading order
Write the sections in the order the information becomes stable, then arrange them for the reader:
- Steps 4 to 8 first — ethics, recruitment, instruments, procedure, data management. These are fixed by what you actually did and can be drafted the week you finish collecting.
- Step 9 next, from your analysis notes, while the decisions are fresh.
- Steps 1 to 3 last, because the design argument reads far better once you know how the study actually went.
- Step 10 at the very end, from the running log you should have been keeping all along.
Keep that log from day one: date, what changed, why, who approved it. It takes a minute a week and it writes Step 10 for you.
The chapter also has to hold its shape through several rounds of revision, and methodology chapters are rewritten more than any other because the analysis keeps moving underneath them. Draft your methodology chapter in Tesify — the section structure and references stay consistent through every rewrite while every methodological choice remains 100% yours.
Frequently asked questions
How long should a methodology chapter be?
Long enough for a stranger to repeat the study and no longer. Length is set by your programme and your design; a tightly written 20 pages beats a padded 45 at every defence.
What is the difference between methodology and methods?
Methodology is the justification for the approach; methods are the procedures you carried out. Most Canadian thesis chapters are expected to contain both, with the justification first.
Do I write the methodology chapter in past or future tense?
Past tense in the thesis, because the study has happened. A proposal is written in the future tense and then converted, which is why so many chapters contain both.
Do I need ethics approval before a pilot study?
Yes. TCPS 2 Article 6.11 explicitly includes proposals for pilot studies in what must be submitted before recruitment or data collection begins.
Does secondary data analysis need REB review?
Article 6.11 covers access to data, so analysing an existing identifiable dataset falls inside the rule even though you never meet a participant. Ask your REB office; the answer depends on the dataset.
Where do limitations go?
Design limitations that follow from your methodological choices belong in the methodology chapter, next to the choice that produced them. Limitations discovered in the results belong in the discussion.
Should I include my questionnaire in an appendix?
Include your own instruments. For a published or licensed instrument, check the licence first, because reproducing a copyrighted questionnaire in a publicly deposited thesis is a separate permission question.
How do I justify a qualitative sample size?
By argument rather than by a number, and increasingly not by claiming saturation, which the authors of reflexive thematic analysis now advise against. The current guidance and the wording to use are set out in our article on how many participants a thesis needs.
What if I changed my analysis after seeing the data?
Say so, and say why. An analysis chosen after inspecting the results is exploratory and should be labelled as exploratory; presented as confirmatory, it is a defect a methodologist on your committee will find.
Do I have to name the software version?
Yes. Procedures and defaults change between versions, and the version number is part of what makes the analysis reproducible.
What if recruitment failed and I changed the plan?
Write it into the deviations paragraph with the date and the approval. A documented change is evidence of rigour; an undocumented one is a finding against you.
Who checks that my method is sound?
Your supervisor and supervisory committee carry the scholarly review, while the REB reviews the ethical implications of the design. They are separate reviews with separate criteria, and passing one does not settle the other.