A methodology chapter is not a procedure manual. Describing what you did is necessary but not sufficient. Every choice needs a stated reason tied back to your research questions, or a committee member will ask "why this design and not another?"
| Decision | Question It Must Answer |
|---|---|
| Research design | Why this design (e.g. case study, survey, experimental) fits these specific questions? |
| Population & sampling | Why this population, and why this sampling strategy and size? |
| Instrumentation | Why this instrument, and is it validated/reliable for this population? |
| Data collection procedure | How will data actually be gathered, step by step? |
| Analysis plan | What specific analysis will answer each research question? |
| Ethical considerations | How are participant rights and data protected? |
The single most common methodology weakness is a mismatch: an exploratory "how do people experience X" question paired with a rigid quantitative survey, or a "does X predict Y" question paired with a handful of interviews that can't establish a statistical relationship. We start every methodology chapter by checking that the design can actually answer the questions as worded. Sometimes that means revising the questions slightly, not just the design.
Write your analysis plan before you collect data. Naming the exact statistical test or coding approach for each research question, before data collection, both strengthens your proposal and prevents costly mismatches discovered after the data is already in.
Committees read Chapter 3 (or Chapter 2 in some proposal formats) looking for one thing above all: a straight line from your research questions to a defensible answer-generating process. That line runs through seven sections, and a weakness in any one of them is usually what triggers a "revise and resubmit." Here is what each section needs to do, and why skipping the reasoning behind it, not just the content, is what gets chapters sent back.
Even though the design was likely introduced in Chapter 1, the methodology chapter restates it in more technical language and defends it directly against the research questions. This is not repetition. It's the section where you show the design was chosen deliberately, after weighing alternatives, rather than picked because it was familiar or convenient. A strong version of this section names the design precisely (e.g. "a multiple case study design" rather than "a qualitative approach") and explains the specific features of that design that make it suited to answering your questions.
This section answers three questions your committee will always ask, whether or not they're stated explicitly: where will data come from, who exactly qualifies, and how many are enough? The setting needs enough description that a reader unfamiliar with your context understands why it's an appropriate site for the study. The sample section needs explicit inclusion and exclusion criteria, not just "working nurses" but "registered nurses with at least two years of acute-care experience, currently employed in a hospital of 200+ beds". Plus the sampling strategy (purposive, stratified random, convenience, snowball) and a justified sample size, whether that justification comes from a power analysis or a saturation argument.
Every instrument named, a survey, an interview protocol, a standardized scale, an observation checklist, needs its origin explained. If you're using an existing validated instrument, cite the validation study and report its established reliability (e.g. Cronbach's alpha) rather than assuming reliability transfers automatically to your population. If you're building a new instrument, this section needs to describe how it was developed, whether it was piloted, and what changes the pilot produced. A common weakness here is listing instruments without ever stating why they were the right tool for measuring these specific constructs.
This is the most literal section in the chapter. A step-by-step account detailed enough that another researcher could replicate it. Recruitment method, consent process, where and how data collection actually happens, how long it takes, what happens to the raw data immediately afterward (transcription, secure storage). Vague procedural language ("participants will be interviewed") is one of the fastest ways to draw committee questions, because it leaves the exact mechanics, and therefore the study's feasibility, unclear.
Here you name the exact analytic procedure for each research question individually, not a general description of "qualitative analysis" or "statistical analysis" for the study as a whole. For quantitative work that means naming the specific test (see our quantitative dissertation help guide) and the software you'll use. For qualitative work that means describing the coding approach step by step (see our coding and thematic analysis guide). Writing this section before data collection, not after, is one of the single highest-leverage things a doctoral candidate can do, because it exposes any mismatch between question and analysis while it's still cheap to fix.
This section covers informed consent, confidentiality and data protection, how vulnerable populations (if any) are protected, and your institution's IRB or ethics board approval status. It should name the specific safeguards used, de-identification procedures, secure storage location, who has access to raw data, rather than a generic statement that "ethical guidelines will be followed." Our IRB approval guide covers this process in more depth, including timing it correctly against your proposal defense.
Limitations are constraints outside your control (a self-selected sample, single-site data collection, self-report bias); delimitations are boundaries you chose deliberately (excluding a subgroup, limiting the timeframe). Conflating the two is a common and easily fixed mistake. Naming limitations honestly, and explaining what you did to mitigate each one where possible, reads as methodological maturity, not weakness. A chapter that claims no limitations at all is usually read as naive rather than rigorous.
Consider a doctoral candidate researching whether a structured mentoring program affects first-year retention among community college students. The research question is: "What is the relationship between participation in a structured mentoring program and first-year retention rates among first-generation community college students?" Working through the seven sections above: the design is correlational (predictive, non-experimental, since students weren't randomly assigned to mentoring); the setting is two community colleges with an established mentoring program; the sample is first-generation students in their first year, drawn via stratified sampling to balance mentored and non-mentored groups, with sample size set by a power analysis targeting a medium effect size at 80% power; the instrument is institutional retention records plus a validated belonging scale; the procedure pulls de-identified institutional data with IRB approval and administers the scale at two time points; the analysis plan is logistic regression predicting retention (yes/no) from mentoring participation, controlling for GPA and financial aid status; the ethical considerations center on de-identifying institutional records before analysis; and the limitations acknowledge that non-random assignment to mentoring means causation can't be claimed, only association. Every section traces back to that one research question, which is exactly what a committee is checking for.
Master's-level methods sections often get away with describing a standard, well-trodden design competently. Doctoral methodology chapters are held to a higher bar: originality of contribution matters, so the design has to be defended not just as workable but as the best available option given genuine trade-offs, and the chapter is expected to anticipate and pre-empt the specific objections a committee in your field is likely to raise. Doctoral committees are also more likely to include a methodologist whose sole role is to stress-test this chapter, which is why a methodology chapter that would pass at the master's level can still stall at the doctoral proposal defense without changing a single procedure, simply because the justification wasn't deep enough.
Design, sampling, instrumentation, and analysis. Each one justified, not just described.
Usually the fix is explicit comparison: briefly note one or two alternative designs you considered and explain specifically why they fit your questions less well than your chosen approach. A justification in isolation reads weaker than one made by contrast.
Yes. For quantitative and mixed-methods studies we include power analysis or established sampling guidelines appropriate to your design; for qualitative studies we justify sample size against saturation principles specific to your chosen approach.
Yes, this is usually included as part of the methodology chapter. See our dedicated IRB approval guide for more detail on that specific section.
Yes. Our research design guide focuses specifically on choosing and justifying between experimental, quasi-experimental, correlational, and descriptive designs. This guide covers the full methodology chapter that follows once that choice is made: sample, instruments, procedures, analysis plan, ethics, and limitations.
The seven-part structure above is the most common convention, but institutions vary in labeling and ordering. We always draft to your program's required template or dissertation handbook first, then make sure each of these underlying elements is present regardless of what the sections are called.