Research Design: Definition, Guide and Methods

“So, talk us through your research design.” That question comes up in almost every supervision meeting, and most students answer it with the name of a tool. They say “an online questionnaire”, then discover halfway through the analysis that the decision which mattered was due months earlier.

A research design is the overall framework for your study. It sets out what you want to find out, where your data comes from, whether that data is qualitative or quantitative, where and how often you collect it, and how many cases you examine. Your method follows from the framework, not the other way round. Work through this guide and you can describe your own design in six decisions and justify every one of them.

📌 Key points at a glance

• The research design is the framework, the method sits inside it.
• Six decisions describe a research design in full.
• Leeds Beckett contrasts mono-method approaches with mixed-methods approaches.
• A cross-sectional study shows association, never cause.
• Markers reward the justification, not the workload.

What is a research design? Definition and purpose

A research design is the methodological plan for an entire empirical study. It describes how the investigation is set up, and it is settled before the first question is drafted and before the first participant is approached.

The Skills for Learning guide at Leeds Beckett University, last updated on 3 August 2026, defines it in a single line: research design is the framework that guides a research project. The same guide lists what the framework covers: approach and methodology, data collection, sampling strategy, analysis techniques, ethical considerations and the reporting of findings.

Framework is the load-bearing word. A framework holds several decisions at once, which is why “I am doing interviews” never answers the question. Interviews fill one slot in the frame and your marker wants to see the other slots filled in too.

What your research design fixes

• The kind of knowledge you are after.
• The source of the data you analyse.
• Whether that data is qualitative, quantitative or both.
• Where, how often and on how many cases you collect it.

What it deliberately leaves open

• The exact wording of your questions.
• The software you collect and analyse with.
• The particular statistical test.
• Formalities such as layout and referencing style.

Keeping the two lists apart pays off in the write-up, because a student who can name the design answers questions about the limits of the study in one sentence.

The design decides what your study is able to show. The method only decides how cleanly it shows it.

Our reading tip: the Leeds Beckett Skills for Learning guide on research design is free to read and covers observation, questionnaires, interviews, focus groups, document analysis, experiments and sampling in one place.

Research design or research method: where is the difference?

A research design answers the question of how your study is set up, while a research method answers the question of how you collect the data. The two do not sit side by side. They sit on top of each other, because the design decides which methods are open to you at all.

Mixing the two levels is the most common reason a methodology chapter comes back with comments. “I am running an online survey” names a method and leaves five of the six design decisions open. Your supervisor then asks about the purpose of the study, the number of measurement points or the selection of cases, and a method on its own answers none of that.

Read the table downwards and every row narrows the row beneath it. A cross-sectional design rules out certain analyses, a qualitative design rules out the standardised questionnaire. Keeping that order is faster than starting with the instrument and inventing a rationale afterwards. An overview of the method level sits in our guide to empirical research methods.

How to create a research design: the six decisions

A research design comes out of six decisions taken in this order: purpose, data source, data type, setting, number of measurement points and number of cases. Each one narrows the next, so skipping a decision costs time later rather than saving it now.

The table below is our own selection, cut down to what actually has to be settled in a dissertation rather than in a funded programme. Treat it as a questionnaire about yourself. Anyone who can answer all six rows has a finished research design, even with no prose written yet. For the wider process around it, see our guide to empirical research.

Decision 1: what is the purpose of your study?

The purpose sets how far your conclusions are allowed to reach. Exploratory studies open up a field that has barely been researched and produce hypotheses, descriptive studies record frequencies and distributions, explanatory studies test a suspected relationship. Choosing one is a commitment rather than a label: exploratory work cannot prove anything at the end, and explanatory work needs testable hypotheses before collection starts. Our guide to exploratory, descriptive and explanatory research separates the terms.

Decision 2: your own data or data that already exists?

The data source saves the most time and gets chosen deliberately least often. Alongside your own fieldwork sits secondary analysis, which means working with data other people collected. For the UK, the UK Data Service holds those collections at three access levels: open, safeguarded and controlled. Students below PhD level are advised to limit themselves to safeguarded End User Licence data for their dissertations, and Secure Access collections are explicitly not suitable for them.

Decision 3: mono-method or mixed-methods?

The data type governs everything downstream, from sample size to analysis. Leeds Beckett puts the choice as mono-method, meaning quantitative or qualitative on its own, against mixed-methods, meaning a combination of both. Quantitative work runs on numbers, standardised instruments and larger samples, qualitative work on text, open questions and few cases studied in depth. Mixed-methods needs a reason beyond thoroughness. The full comparison sits in our guide to qualitative and quantitative research.

Decision 4: laboratory or field?

The setting governs how much control you have over interference. A laboratory gives you a controlled situation and lets you vary conditions on purpose, at the cost of everyday realism. Fieldwork reaches people where they normally are, with every influence that comes with it. Student projects are almost always fieldwork, and an online questionnaire counts as fieldwork too. What matters is presenting the setting as a choice rather than as something that simply happened.

Decision 5: one measurement point or several?

The number of measurement points separates cross-sectional from longitudinal designs and decides whether you may talk about change. A cross-sectional study collects data once and shows a state, a longitudinal study collects repeatedly and shows development. The Millennium Cohort Study at UCL is the British textbook case: it follows around 19,000 people born across England, Scotland, Wales and Northern Ireland in 2000 to 2002, across eight main sweeps from nine months of age to 23. In a dissertation the cross-section is the norm, because the deadline rarely leaves room for a second wave.

Decision 6: how many cases do you study?

The number of cases runs from a single one to the whole population. A case study examines one case, perhaps one school or one company, in real depth. A sample is a selection from the population and carries conclusions back to the whole. A census questions everyone, which outside a national count or a small, clearly bounded group is rarely realistic. How to choose procedure and size is set out in our guide to sampling.

Research design examples from two dissertations

A research design becomes concrete as soon as the six decisions are filled in for a real question. The two examples below come from ordinary dissertation projects and differ in every single row, although both sit under the same broad topic.

Put them next to each other. On the left a student who wants to know how widespread a behaviour is, on the right a student who wants to know why people behave that way. The same curiosity produces two designs with almost nothing in common.

Both designs are complete and neither is the better one. Ten interviews are not worth less than 200 questionnaires, they answer a different question. Fill in the left-hand column for your own project and the structure of your methodology chapter is already settled.

💡 Tip

Write your six rows on half a page and send them to your supervisor before the proposal deadline. A design can be corrected in a ten minute conversation. Fieldwork that has already started cannot.

Writing up: what belongs in your methodology chapter

The research design appears in your write-up as its own chapter between the literature review and the results, and it explains in a few pages what you did and why. Marks go to the justification far more often than to the description.

British terminology is worth getting right here. At undergraduate and taught master's level the write-up is called a dissertation, while thesis is kept for doctoral work, and the UK Data Service uses the words that way. The viva belongs to the doctorate, and taught programmes are assessed on the written submission alone.

The order inside the chapter is much the same across disciplines because it follows the route the study took. Your department handbook overrules it, since some programmes want the hypotheses in the literature review instead.

1. Research question and hypotheses. What the study follows from.
2. Purpose and design type. The six decisions in prose.
3. Choice of method. With a reason, not just a name.
4. Sample and access. Selection procedure, size, response rate.
5. Instrument and pilot. Structure, scales, changes after piloting.
6. Analysis technique. Which technique, applied to which question.
7. Limitations of the design. What this setup cannot show.

The gap between a good mark and a middling one sits almost always in point 3 and point 7. Four questions turn a report into an argument: why this method and not another, why these cases, why this analysis, and what each decision costs you.

⚠️ Warning

Naming your own limitations gains marks rather than losing them. Writing that a cross-section cannot support a causal claim shows methodological understanding. Leaving it out means the examiner raises it instead.

Where the chapter sits in the shape of the whole submission is covered in our guide to the empirical dissertation. For the instrument itself, use our guide to questionnaire design.

Common mistakes in research design

Most trouble in a methodology chapter comes not from missing knowledge but from an early decision nobody revisited. Four mistakes turn up again and again.

All four are cheap to fix in the first few weeks and impossible to fix after data collection. Timing is the reason a research design belongs before you register the project rather than after.

The method is fixed before the question is

Starting with “I am running an online survey” means hunting for a question that fits it. Markers notice, because the methodology chapter then contains no justification: no alternatives were ever weighed. Turn the order round and narrow the research question until a purpose falls out of it. The method follows almost by itself after that.

The design does not fit the submission deadline

A longitudinal study across two semesters, a census of a whole city, thirty interviews with transcription: interesting on paper, undeliverable in four months. Budget the fieldwork realistically, because responses do not arrive on day one and transcription takes roughly five times the recording length. A small design carried out properly beats a large one handed in unfinished.

A cross-section turns into a causal claim

The most common factual error in empirical dissertations shows up in the results chapter, as a sentence of the form “X leads to Y”. A cross-sectional study only shows that two characteristics occur together, not which one produces the other. Report such findings as association and say in the discussion which design a causal claim would have needed.

The design is stated but never justified

Many submissions name the design correctly and say nothing about why it was chosen. Examiners then cannot tell whether the decision was deliberate or accidental. Write one sentence with because for each of the six decisions, even where it sounds obvious. Four sentences like that change how the whole chapter reads.

Conclusion

A research design is more predictable than the term suggests: six decisions, each with two or three usual answers, and the rest of the project follows almost by itself. Take them deliberately and in order before you build the first questionnaire. Write a because next to each one, since that is what your methodology chapter is marked on.

Where to go next

• Still missing the overview of the whole research process? Fundamentals of empirical research
• Choosing between qualitative and quantitative? Qualitative vs. quantitative compared
• Hypotheses not written yet? How to write a hypothesis
• Planning your own fieldwork? Running a survey for your dissertation

Design settled, fieldwork still to do?

With empirio.ai, an online survey tool from Germany with hosting in the EU, you build the questionnaire that matches your design and export the responses straight into your analysis.

How to build the questionnaire for your research design