Many research projects begin with a familiar question: “Should I use quantitative or qualitative methods?” It is a convenient question, but it often comes in the wrong order. A method is not a label for a topic; it is a way to generate the kind of evidence required to answer a clearly defined question. When researchers begin with software, a questionnaire, or an intention to interview, they can easily collect a great deal of data without having the evidence needed to defend the argument they want to make.
Choose a method only after identifying what must be known, who can provide the evidence, and what would make that evidence credible.
Start with the research decision, not the tool
A practical issue becomes a research problem only when we can identify what has not been explained, measured, or understood well enough to inform a decision. For example, a company sees customers leaving after their first purchase. If the question is “how many customers leave and which factors are associated with leaving?”, behavioural data and a survey may fit. But if the question is “which experience do customers interpret as untrustworthy, and why do they not give feedback?”, aggregate figures are insufficient: the study needs access to customers’ experiences and language.
Calling a project quantitative or qualitative before clarifying the question removes the most important reasoning step. Write a brief description: what phenomenon is occurring, who is involved, which decision needs improvement, and which part of the mechanism remains unclear. That description helps identify the necessary evidence rather than forcing the question to follow a familiar tool.
Three kinds of evidence that are often confused
Evidence about magnitude answers “how many”, “how prevalent”, or “whether groups differ”. It requires clearly defined variables, suitable measures, an appropriate sample size, and matching analysis. Evidence about mechanisms answers “why”, “through which pathway”, or “under what conditions a relationship changes”. It may require theory, longitudinal data, comparative design, or controls for competing explanations. Evidence about meaning answers “how people in the situation understand this”, “what they prioritize”, or “which language and context make the behaviour reasonable”. Interviews, observation, documents, and thematic analysis are often useful here.
No kind of evidence is inherently superior to another. What needs to be controlled is the fit between the claim and the evidence. A coefficient may show an association, but it does not by itself explain why participants chose as they did. A rich interview may clarify a decision process, but it does not by itself establish how widespread that process is in a population. When a conclusion goes beyond the capacity of the data, the study becomes weaker even if its technique is sophisticated.
Example
A university wants to understand why students rarely use career counselling. A survey can estimate awareness, convenience, and usage intention. Interviews can reveal that students fear being judged, do not understand which problem the service solves, or believe it is only for final-year students. If the aim is both to understand the scale and design a response, the two kinds of evidence can be combined in sequence: explore first, then measure.
When is a mixed-methods design genuinely necessary?
Mixed methods does not mean “do both to be safe”. It is useful only when each source of data addresses a different part of one argument and connecting them adds value. A design may begin with interviews to identify how users describe an experience, then use those findings to develop variables, measures, or survey hypotheses. It may also begin with a surprising quantitative result and then interview contrasting groups to explain the pattern. If the two parts merely sit alongside each other without an integration point, research costs increase without a matching increase in contribution.
A checking sequence before committing to a method
First, write the research question with an appropriate verb: estimate, compare, explain, explore, interpret, or evaluate. Next, state the conclusion the study should be entitled to make. Then list the minimum evidence required and the main risk of bias. Only then consider the approach, unit of analysis, sampling, instrument, and analytical technique. This sequence avoids a common mistake: taking an available questionnaire and then revising the research question to justify it.
For quantitative research, check the fit among construct, indicator, and unit of analysis before discussing a model. For qualitative research, define who has relevant experience, how they will be approached, the stopping criterion for data collection, and the interpretive audit trail. For mixed methods, state the connection in advance: which finding from the first phase will design, sample, test, or explain the second phase. These are design decisions, not decorative sections of a proposal.
Conclusion
Choosing a method is a decision about the quality of an argument. A good project does not need to be more complicated than necessary; it needs to generate the right kind of evidence, make it sufficiently credible, and interpret it within the right scope. Starting from the question and the evidence needed helps researchers build a leaner design, defend their choice more clearly, and turn results into usable knowledge.
References
- Creswell, J. W., & Creswell, J. D. (2018). Research design: Qualitative, quantitative, and mixed methods approaches (5th ed.). SAGE.
- Maxwell, J. A. (2013). Qualitative research design: An interactive approach (3rd ed.). SAGE.
- Venkatesh, V., Brown, S. A., & Bala, H. (2013). Bridging the qualitative-quantitative divide: Guidelines for conducting mixed methods research in information systems. MIS Quarterly, 37(1), 21–54. https://doi.org/10.25300/MISQ/2013/37.1.02
