Why methodology should follow the research question, not researcher habit

1. Intuitive Introduction

Many studies begin to drift not when analysis starts, but earlier, when the researcher has a question and must decide what methodological logic should guide the project. That decision is often made for the wrong reasons. A student may choose a survey because survey software is available. Another may choose interviews because interviews feel richer or more serious. A third may call the project mixed methods simply because using both numbers and quotations sounds stronger than using either one alone. None of these is a defensible basis for design. A methodology is justified only when it fits the kind of answer the research question requires. Creswell and Creswell treat research design precisely as a matter of aligning questions, designs, and evidence, while Hoadley places methodological alignment at the center of sound inquiry.

This blog post therefore makes a simple claim: methodology should not be chosen because it is familiar, fashionable, or convenient. It should be chosen because it gives the study the best chance of producing evidence appropriate to the question. That follows directly from the architecture of this blog posts series. Part I focused on building strong research questions, diagnosing common design mistakes, and showing what good alignment looks like. Part II now turns to core design decisions. The first of those decisions is methodology, because this is where many promising projects either become coherent or begin to weaken.

2. What “Methodology” Actually Means

Researchers often use the words methodology, method, and tool as if they meant the same thing. They do not. Methodology is the overall logic of inquiry. It explains why a certain kind of evidence, generated in a certain way and interpreted under certain assumptions, is appropriate for a particular research question. A method is more concrete: surveying, interviewing, observing, experimenting, analyzing documents, or building a mixed methods sequence. A tool or technique is narrower still: a questionnaire, interview guide, coding frame, regression model, GIS workflow, or joint display.

This distinction matters because many weak designs begin by mistaking procedure for logic. Saying “my methodology is a questionnaire” is like saying that a microscope is the whole logic of a laboratory study. A questionnaire may be part of a method, and that method may sit inside a broader methodology, but the levels should not be collapsed. The same applies to interviews, regressions, thematic coding, and software platforms. Good design begins when researchers stop asking only what they know how to do and start asking what kind of evidence their question actually requires.

3. The Main Decision Rule: Start from the Research Question

The most important rule in choosing methodology is straightforward: start from the research question. Not from the software. Not from the available dataset. Not from departmental habit. Not from the method the researcher already knows best. The key question is this: what kind of answer would count as a credible answer to this research question?

Different questions seek different kinds of answers. Some ask for description: how much, how often, how common. Some ask for comparison: do groups differ. Some ask for association: are variables related. Some ask for interpretation: how do people understand, experience, or negotiate something. Some ask about process: how something unfolds over time or through interaction. Some ask for integration: how numerical patterns and contextual meanings can be brought together in one coherent design. Once the intended answer becomes clear, methodology choice becomes much less mysterious. The methodology is not chosen first and justified later. It is chosen because it is appropriate to the kind of answer being sought.

This is also where the tetrad logic reappears. RQ comes first because it defines the empirical task. M follows because it specifies the logic most suited to answering that task. D comes next because data must then be generated or selected in a form the methodology can use responsibly. RH may matter greatly in some quantitative designs, less in qualitative ones, and differently again in mixed methods studies. But the principle remains stable: the question sets the task, and the methodology should be chosen in response to that task.

4. When a Question Calls for Quantitative Methodology

A question calls primarily for quantitative methodology when the empirical task is to describe distributions, compare groups, examine structured associations, estimate patterns, or, under stronger design conditions, assess effects. These questions require evidence in the form of variables, categories, counts, measurements, scores, frequencies, or structured comparisons.

This does not mean that quantitative methodology is simply “the one that uses numbers.” That shorthand is too crude. The deeper point is that quantitative methodology is appropriate when the study needs bounded, comparable, and structured evidence. If the question is “How prevalent is vaccine hesitancy among first-year university students?” or “Do rural and urban schools differ in standardized mathematics scores?” or “Is financial stress associated with absenteeism among hourly workers?” then quantitative logic is usually a strong fit. The concepts must be translated into observable variables, and the design must support the inferential level of the question. A descriptive question does not become causal simply because the model looks sophisticated. A quantitative design is strongest when the claim remains proportionate to what the variables and design can actually support.

5. When a Question Calls for Qualitative Methodology

A question calls primarily for qualitative methodology when the task is to understand meaning, lived experience, interpretation, process, situated practice, context, or social world. These are not questions best answered by counting alone, even if counting later plays a supporting role.

If the question is “How do first-generation students make sense of belonging in their first year at university?” or “How do clinicians interpret the introduction of AI-assisted decision tools in everyday practice?” or “How do small farmers talk about uncertainty during prolonged drought?” then the researcher is not primarily asking for a distribution or a numerical association. The central task is interpretive. The methodology must therefore generate depth, context, and meaning. This does not make qualitative work less rigorous than quantitative work. It means rigor takes a different form. The design must produce rich enough evidence to illuminate processes, perspectives, and situated understanding rather than merely listing opinions or collecting illustrative quotations.

6. When a Question Calls for Mixed Methods Methodology

A question calls for mixed methods methodology when one kind of evidence is not enough and the design genuinely requires integration between qualitative and quantitative strands. Mixed methods is strongest when the study needs both patterned evidence and contextual understanding, both breadth and depth, or one strand to explain, build, qualify, or extend the other.

This means mixed methods is not justified merely because the topic is complex. A topic may be important and still be fully answerable through a single coherent methodology. Mixed methods is warranted when the research question actually requires integrated inference. A study may need to estimate the prevalence of low belonging among first-year students and also understand how peer networks shape that belonging. Or it may need to identify which groups adopt a health intervention and also understand why implementation differs across contexts. If the second strand is added only to decorate the first, the design is not truly mixed methods. If the strands do not need each other, the rationale for mixing is weak.

7. Same Topic, Different Methodological Paths

A useful way to see the logic of methodology choice is to keep the topic constant while changing the question. Consider the broad topic of trust in AI tools in higher education.

A quantitative version might ask: What proportion of undergraduate students report low trust in AI-assisted feedback tools, and is trust associated with prior experience using them? That question seeks prevalence and association. It therefore calls for structured data and a quantitative methodology.

A qualitative version might ask: How do undergraduate students interpret trust and distrust when using AI-assisted feedback tools in their coursework? This question is about meaning, interpretation, and situated experience. It therefore calls for qualitative methodology.

A mixed methods version might ask: How widespread is distrust in AI-assisted feedback tools among undergraduate students, and how do students’ accounts help explain variation in reported trust across user groups? Here the study requires both patterned evidence and interpretive explanation, and the strands need to inform each other. That is a genuine mixed methods rationale.

The lesson is decisive: methodology is not attached to the topic by nature. It is attached to the question. The same broad area can yield different legitimate studies depending on what the researcher is actually trying to know.

8. How Researchers Choose the Wrong Methodology

Researchers often choose the wrong methodology for reasons that feel practical in the moment and costly later. Familiarity is one of the most common. A researcher who knows how to build surveys may quietly reshape an interpretive question into a surveyable one. Another who is comfortable with interviews may turn a prevalence question into a small interview project and then write as if the conclusions extend far beyond the sample. A third may call a project mixed methods simply because including both kinds of data seems more impressive than choosing one logic and defending it properly.

Departmental habit also matters. In some fields, quantitative work functions as the default sign of rigor. In others, interviews signal depth. In still others, mixed methods has become fashionable enough that researchers feel pressure to add a second strand even when it contributes little. Data availability creates another distortion. Researchers may already possess clickstream logs, administrative records, one accessible site, or a convenient participant pool. The temptation is then to build the design around available evidence instead of beginning with the research question. That is one of the clearest ways that methodology stops serving the question and starts dictating it.

9. What Can Be Fixed Later, and What Must Be Fixed Early

Some methodology problems are easier to fix early than late. Before data collection begins, the researcher can still narrow the question, lower the inferential ambition, change methodological direction, or redesign the data plan. A project that first appeared to require mixed methods may turn out to be much stronger with a focused qualitative design. A study initially framed as explanatory may need to become descriptive or exploratory. At this stage, methodological correction is not failure. It is part of good design discipline.

After data collection, repair becomes harder. If the question was interpretive but the researcher collected only clickstream behavior, the design cannot later be turned into a deep qualitative study through better wording alone. If the study aimed at prevalence but collected only a small set of interviews, no amount of confidence in the conclusion can give those interviews population reach. If a project called itself mixed methods but never linked its strands, integration can sometimes be improved at the analysis stage, but architecture that was never built into the design is difficult to create afterward. This is why methodology choice is a core design decision, not a cosmetic label.

10. Concluding Synthesis

The central lesson of this blog post is simple but demanding. Methodology is not the first thing a researcher should choose, and it is not the same as a favored method or familiar tool. Methodology is the design logic that links the research question to the kind of evidence needed for a credible answer. A study does not become strong because its method sounds modern, advanced, or prestigious. It becomes strong when the methodology genuinely fits the question. The next design decisions in this book will become easier only if this principle is firmly in place. Once the question is clear, methodology must follow it. Only then can data, analysis, and interpretation begin to pull in the same direction.

Mini-Checklist

Before choosing a methodology, ask:

• What kind of answer is my research question actually seeking?
• What kind of evidence would count as a credible answer?
• Does my chosen methodology generate that kind of evidence?
• Am I choosing this methodology because it fits the question, or because it fits my habits, skills, or access?
• If the methodology works perfectly, will it answer the question I actually wrote?
• What would a careful skeptical reader say is missing from my evidence?

A strong study is not the one with the most impressive method. It is the one in which methodology genuinely fits the question.

References

Booth, W. C., Colomb, G. G., Williams, J. M., Bizup, J., & FitzGerald, W. T. (2024). The craft of research (5th ed.). University of Chicago Press. https://doi.org/10.7208/chicago/9780226826660.001.0001

Creswell, J. W., & Creswell, J. D. (2023). Research design: Qualitative, quantitative, and mixed methods approaches (6th ed.). SAGE.

Creswell, J. W., & Plano Clark, V. L. (2018). Designing and conducting mixed methods research (3rd ed.). SAGE.

Fetters, M. D., Curry, L. A., & Creswell, J. W. (2013). Achieving integration in mixed methods designs—Principles and practices. Health Services Research, 48(6 Pt 2), 2134–2156. https://doi.org/10.1111/1475-6773.12117

Hoadley, C. M. (2004). Methodological alignment in design-based research. Educational Psychologist, 39(4), 203–212. https://doi.org/10.1207/s15326985ep3904_2

Maxwell, J. A. (2013). Qualitative research design: An interactive approach (3rd ed.). SAGE.