How to write a research methodology: steps and example

The methodology tells readers exactly what you did to answer your research question, and why you did it that way. It has two jobs. It lets readers judge whether your results can be trusted, and it gives another researcher enough detail to repeat the study.

methods and methodologytwo words, one section

Methods are the tools and steps you used: a questionnaire, a weighing scale, a statistical test. Methodology is the reasoning behind them: why this design, this sample and these tools suit your question. A good section gives both.

Who reads it changes how much you write. A video guide from the UK's National Centre for Research Methods advises PhD students to picture their examiners before writing the methodology chapter: what will they need explained and justified? It also points out that in a research paper you may have only a few hundred words for the same job.

• In a thesis, the methodology is usually a full chapter. You explain your choices, the options you rejected, and the problems you met along the way.
• In a journal paper, the methods section is shorter and more factual. It sits after the introduction, as our guide to the structure of a research paper explains.

Begin the section by restating your research question in a sentence. Every choice that follows should make sense against it. If your question is still vague, fix that first with our guide on how to write a good research question.

Then name your overall approach and your design, and say why it fits.

Abhishek's question had two halves. How much of the meal do children eat on millet days compared with rice days? And why? The first half needs measurement; the second needs people's own words. So he named his design as mixed methods and gave that reason in one sentence. "Now the reader knows why there are scales and interviews in the same study," Dr. Singh said.

Next, say where and when the study took place, and who took part. Give enough for a reader to picture it: the district, the type of school or hospital, the months of data collection.

Then explain how you chose your participants. Name the sampling method, and do not call a sample random unless it was.

Add your inclusion and exclusion criteria, your sample size and how you decided it, and how many people actually took part. Abhishek wrote that he listed all government primary schools in one block of Varanasi district, picked four by lottery, and included children of classes 3 to 5 who were present on the study days. Children who brought food from home were excluded, and he said so.

This is where most first drafts are too thin. Describe what you collected, with what tools, and in what order, so that someone else could do the same.

1. The tools.Name each questionnaire, test or instrument. If you used an existing questionnaire, cite it; if you made your own, say how you built and checked it.
2. The equipment.Give the make or type and the precision, for example a digital kitchen scale accurate to 1 gram.
3. The pilot.If you tested your tools first, say what you changed afterwards.
4. The procedure.Who collected the data, when, where and in what order. A reader should be able to follow it like a recipe.
5. Translation.If you used Hindi or another language with participants, say who translated the tools and how you checked the translation.

Abhishek's new version had a paragraph for each tool. On two millet days and two rice days in each school, he weighed each child's serving before the meal and the leftovers after it. Children rated the taste on a five point smiley scale read aloud in Hindi. Afterwards he interviewed the cooks and two teachers in each school.

Readers should be able to see which analysis answers which part of your question. Match them one to one.

• the software you used, with its version
• how you cleaned the data and handled missing values
• the statistical tests for each comparison, and why they suit your data
• the significance level you chose, such as p < 0.05
• for interviews: how you transcribed them, the method of analysis, such as thematic analysis, and who did the coding

Do not report results here. "We compared the amount eaten on millet and rice days with a paired test" belongs in the methodology; "children ate less on millet days" belongs in the results.

Abhishek had written "appropriate statistical methods". Dr. Singh circled the words. "Appropriate according to whom? Name the test, and tell me why it fits." He rewrote it as two sentences: one naming the test and the software, one explaining that each child was measured on both kinds of day.

If your study involves people, the methodology must say how you protected them. In India, the ICMR's National Ethical Guidelines for Biomedical and Health Research Involving Human Participants (2017) apply to biomedical, social and behavioural research for health. They say all research proposals must be submitted to an ethics committee, and that the researcher must obtain voluntary written informed consent from participants.

• Name the committee and the approval. Give the ethics committee's name and, if your journal or university asks, the approval number and date.
• Describe consent. Say how participants were told about the study and how consent was recorded.
• Children need two things. The ICMR guidelines ask for consent from parents and assent, the child's own agreement, from children of 7 to 18 years of age.
• Clinical trials must be registered. Registration with the Clinical Trials Registry, India (CTRI) has been mandatory since 15 June 2009 for trials under the Drugs and Cosmetics Act and its rules, and voluntary for other health research.

End the methodology with its limits. Every study has some: a small sample, one district, a season, a tool that cannot capture everything. Say what they are and how they affect what your results can show. Abhishek wrote that four schools in one block could not represent the whole district, and that weighing leftovers missed food children shared with friends. Naming limits does not weaken a study. It shows the reader you understand it.

Write the methodology of a finished study in the past tense: "we selected", "children rated". Use subheadings, and keep them in the same order as your results, so readers can move between the two. In a proposal, where the study has not happened yet, you write in the future tense instead; our guide on how to write a research proposal covers that.

Many fields have reporting guidelines that list exactly what a methods section should include. The EQUATOR Network collects them for health research.

Here is the outline of Abhishek's rewritten methods section. The details are simplified, but the shape is the one reviewers expect.

Before you submit, read your methodology as a stranger would and ask: could I repeat this study from these words alone? Abhishek gave his new version to a friend in another department. She found one gap: he had not said what time of day he weighed the plates. He added it. Dr. Singh read the section twice and returned it with only one word in the margin: "good".