How to choose a thesis methodology

Start with the question, not the favourite method

Students often choose a methodology because it feels familiar: a survey, interviews, an experiment, a case study, or secondary data analysis. Familiarity helps, but it should not decide the design. The method has to fit the type of answer your question needs.

Ask what you are trying to learn. If you need to understand meanings, experiences, or decision processes, a qualitative design may fit. If you need to estimate relationships, compare groups, or test hypotheses, a quantitative design may fit. If the project needs both explanation and measurement, a mixed design may be justified, but only if the scope and time are realistic.

Common methodology choices

Survey research

Surveys are useful when you need structured responses from a defined group. They work best when the constructs can be measured with clear items and when you can recruit enough respondents. A survey is weaker when the question is exploratory and you do not yet know what the important concepts are.

Experiments

Experiments are useful when you need to compare conditions or test the effect of a stimulus. They require careful control: participants should receive different versions of something, and you should know which participant saw which condition. An experiment can be powerful, but it is not automatically better than a survey.

Interviews

Interviews are useful when you want depth, interpretation, or context. They can explain why people think or act in certain ways. They are less suitable when your thesis needs a large numerical estimate or a statistical test of a relationship.

Case study and document analysis

Case studies are useful when the context itself matters. Document or content analysis can work when the evidence already exists in reports, posts, policies, reviews, or archives. The key is to define selection rules so the evidence is not cherry-picked.

How to defend your methodology choice

A methodology section should explain why the chosen design is appropriate, not only what you will do. Write one sentence that links the design to the research question, one that explains the sample or case selection, one that explains the data collection instrument, and one that explains the analysis. If you cannot write those sentences clearly, the method may not be ready yet.

Also name the tradeoff. Every method has one. Interviews give depth but usually limit generalisation. Surveys can reach more respondents but may miss context. Experiments can compare conditions but require careful control of materials. Secondary data can save recruitment time but may not contain the exact variables you need. Acknowledging the tradeoff makes the method sound more credible, because it shows you understand the limits of your own evidence.

Finally, check that the analysis is not an afterthought. A planned t-test, regression, thematic analysis, or coding framework should follow from the data format. Do not promise analysis you cannot perform or explain.

Methodology checklist

• The design answers the exact type of question you are asking.
• The sample or case selection is defined before data collection.
• The data source is accessible within your deadline.
• Measures, interview topics, or coding categories match the constructs.
• The planned analysis fits the data you will collect.
• Limitations are honest and specific.
• Ethics, consent, and participant-data responsibilities are considered early.

Common mistakes

The first mistake is treating methodology as a set of labels. "Quantitative" is not a complete method. "Interviews" is not a complete method. You need design, participants or cases, data collection, instruments, procedure, and analysis. The second mistake is choosing a method that cannot answer the question. If your question asks about lived experience, a five-item survey may be too shallow. If your question asks whether X predicts Y, three informal interviews may not be enough.

The third mistake is ignoring feasibility. A perfect design that requires 400 respondents, access to a private company database, or advanced statistics you cannot learn in time may not be a good master's thesis design. A defensible, smaller method is often stronger than an ambitious method that remains unfinished.