What a methodology chapter actually has to establish, why philosophy is in there at all, and how to choose a design your question can carry.
Research methodology is the reasoning that connects your question to the way you answer it. It is not a list of the techniques you used — those are your methods. The chapter has to show why this question, asked in this field, is best answered by this design, using this data, analysed this way, and what that combination can and cannot establish. An examiner is reading for the chain of justification, not the inventory.
This is the most common confusion in doctoral writing, and it costs marks quietly rather than obviously.
Methods are what you did: semi-structured interviews, an online survey, multiple regression, thematic analysis. Methodology is why any of that was the right thing to do — the argument that runs from the question you asked, through your assumptions about what counts as evidence for it, to the design that follows, and on to what your findings are therefore entitled to claim.
A chapter that lists methods reads like a recipe. It answers what and leaves the examiner asking why this and not something else, which is the question the viva will open with. A chapter with a methodology in it has already answered that before it is asked.
The practical difference shows up in the sentences. “Thirty semi-structured interviews were conducted” is a method. “Because the question asks how clinicians reconcile competing guidance in the moment, the study needed accounts in participants’ own terms rather than counts of behaviour, which is why interviews were chosen over a survey” is a methodology.
A methodology chapter is a sequence of decisions where each one is justified by the one before it. Break a link and everything after it is left unsupported, however well it is executed.
The chain runs roughly like this:
When a supervisor says a methodology chapter “does not hang together”, this chain is nearly always what they mean. The pieces are present; the joins are not argued.
Most researchers find this the least comfortable part of the chapter, usually because it is taught as vocabulary to be recited. It is easier once you see that it is only answering two plain questions about the thing you are studying.
Ontology asks what you take that thing to be. Is “organisational culture” something that exists out there, stable enough to be measured the same way in two companies? Or is it something continuously produced by people talking to each other, so that measuring it flattens the thing you wanted to study?
Epistemology asks what would count as knowing something about it. Numbers from a validated instrument? Accounts from people inside it? Both, treated differently?
The named positions are just common combinations of those two answers.
The trap is asserting a position in one paragraph and then designing a study that contradicts it — declaring interpretivism and then reporting frequencies of themes as if they measured something, or declaring positivism and then generalising from a purposive sample of nine. Examiners notice this quickly, because it is the one part of the chapter where the claim and the practice can be checked against each other on the same page.
You do not need to resolve a centuries-old debate. You need to say where you stand, in a paragraph or two, and then not contradict it for the next two hundred pages.
Deductive work starts from theory, derives expectations from it, and tests them. Inductive work starts from data and builds an explanation upwards. Abductive work moves between the two: you notice something the existing theory does not explain well, and work towards the explanation that would best account for it.
Most real doctorates are more abductive than they admit in writing. Being honest about that is usually stronger than claiming a purity the study did not have — particularly at the point where you explain why the framework changed after fieldwork.
The design follows from what the question is asking for. It is worth being blunt with yourself about what each one can establish, because that is the boundary your conclusions have to stay inside.
| If your question asks… | The design that usually fits | What it will not establish |
|---|---|---|
| How common something is, or how it differs between groups | Cross-sectional survey | Causation, or change over time |
| Whether X causes Y | Experiment or quasi-experiment | Much about why it happens, or how it feels from inside |
| How something unfolds over time | Longitudinal design | Anything, if attrition is not planned for from the start |
| How people understand or account for an experience | Interviews, analysed qualitatively | Prevalence, or a claim about a population |
| How something works inside its real setting | Case study | Statistical generalisation — the claim is analytic, to theory |
| What is going on in a culture or community | Ethnography | Anything on a short timescale |
| Both the size of an effect and the reason for it | Mixed methods | Either, well, if the strands are never actually integrated |
If two designs both seem to fit, the question is usually still too broad. Narrowing it is faster than agonising over the choice.
Every methodology chapter has to say why the findings should be believed. Which vocabulary you use depends on the kind of study, and using the wrong one is a small error that signals a larger confusion.
Reliability is consistency — whether the instrument gives you the same answer under the same conditions. Validity is whether it measures the thing you say it measures, and splits into several questions an examiner may ask separately: does it cover the whole construct, does it behave as theory predicts, does it agree with other measures of the same thing, and does it stay distinct from measures of different things.
Reporting a reliability coefficient and stopping there is common and not enough. A scale can be highly consistent and still be measuring something other than what you named it.
The parallel criteria are credibility (do the findings genuinely represent what participants meant), transferability (could a reader judge whether this applies in their setting — which is why thick description matters), dependability (could the process be followed), and confirmability (are the interpretations traceable to the data rather than to you).
What makes these convincing is evidence in the appendices: an audit trail, coded extracts, memos showing how a theme developed. Asserting trustworthiness in a paragraph convinces nobody.
Write it as an argument in past tense, in the order of the chain: question, assumptions, approach, design, participants and sampling, instruments or protocols, procedure, analysis, quality, ethics, limitations. Each section should end having earned the next one.
Two habits make it noticeably stronger. First, state what you rejected and why — a paragraph explaining why a survey would not have answered this question does more for your credibility than three pages on what interviews are. Second, write it so that somebody else could repeat the study from it. That is the working test of whether the chapter is complete, and it is roughly the test a reviewer applies too.
It is the reasoning behind how you answered your question — why this design, this data and this analysis were the right way to answer it, and what the answer is therefore entitled to claim. The methods are the techniques themselves. Methodology is the justification that holds them together.
Methods are what you did; methodology is why that was the right thing to do. “A questionnaire was distributed to 220 nurses” is a method. “Because the question asks how widely a practice has spread, the study needed breadth rather than depth, which is why a questionnaire was chosen over interviews” is methodology. Chapters that only ever do the first kind of sentence read as descriptive.
In most social science, business, education and health doctorates, yes — and it is usually expected explicitly. In some science and engineering disciplines it is not conventional at all. Check recent theses passed in your own department; that tells you more than any general guide. Where it is expected, a couple of clear paragraphs that then govern the rest of the design beat several pages of textbook summary.
It varies widely by discipline and university, and your department’s own recent theses are the only reliable guide. Length is rarely the real question, though. A chapter is finished when someone else could repeat the study from it and when every choice in it has a stated reason — not when it reaches a word count.
It depends on the kind of study. Quantitative work justifies it in advance from the effect you would consider meaningful, the power you want and the analysis you plan to run. Qualitative work justifies it by what the design needs and by what happened during collection — when new material stopped changing the developing account. Either way, say what determined the number rather than presenting it as given, and be straightforward about constraints that shaped it.
Often, yes, and it happens more than people admit — though it needs your supervisor and, where relevant, your ethics approval to keep pace with it. Changes made before data collection are cheap. Changes after it usually mean adjusting the claim rather than the method: a study that cannot support the conclusion you first wanted can often still support a narrower one, argued well.
Design problems are cheap to fix before data collection and expensive afterwards. Describe your question and where you have got to, and a PhD in your field will tell you honestly whether the design will carry it.
Leave an email or a WhatsApp number — whichever you prefer — and tick how we should reach you. We reach out within 30 minutes.
