Methodological choice is frequently made on temperament — some researchers like numbers, others like interviews — and then justified afterward. The justification is usually visible as such.
The question decides. Work backward from the evidence that would actually settle it.
Quantitative answers how much, how many, how often, and whether a relationship holds
If your question concerns magnitude, frequency, distribution, or whether an association survives controls, you need numbers on enough cases to support inference.
Quantitative work requires that the construct can be measured with reasonable validity, and that you can reach a sample large enough to detect the effect you expect. Both are constraints to check before committing, not after.
What it will not tell you is why. A significant coefficient establishes that something moves together; it is silent on mechanism.
Qualitative answers how, why, and what it is like
If your question concerns process, meaning, interpretation, or the experience of a phenomenon from the inside, you need depth rather than breadth.
Qualitative work requires access to participants who can speak to the phenomenon, and enough analytic discipline to derive themes from data rather than from expectation. It will not tell you how common something is, and a well-designed qualitative study does not attempt to.
The most frequent error is treating a small qualitative sample as a weak quantitative one — reporting that "seven of twelve participants mentioned X" as if it were a prevalence estimate.
Mixed methods answers a question that genuinely has two parts
Mixed methods is not a way of strengthening a study by adding a second dataset. It is appropriate when the question contains two components requiring different evidence, and the design specifies how the strands relate.
Three common designs:
- Explanatory sequential — quantitative first, then qualitative to explain the pattern. Use when you expect a result you will need to interpret.
- Exploratory sequential — qualitative first, then quantitative to test what emerged. Use when the constructs are not yet well defined in your context.
- Convergent — both in parallel, compared at interpretation. Use when you want corroboration from independent angles.
Say which you are using and why. A study that collects both without specifying the relationship produces two thin studies rather than one strong one.
The feasibility filter
Mixed methods roughly doubles the work: two instruments, two analyses, two literatures on method, and an integration argument. It is often the right choice for a funded team and the wrong one for a single candidate on a three-year timeline.
Choosing a well-executed single-method study over a rushed mixed-methods one is a judgement examiners respect — and one you should be able to state plainly in the viva.
The test
Write the sentence you want in your conclusion. Then ask what evidence would make a sceptical reader accept it.
If the answer involves a number, you are quantitative. If it involves an account, you are qualitative. If it genuinely involves both, and you can say how they connect, you are mixed — and you now have the justification paragraph your methodology chapter needs.
We do this work every week — for MPhil and PhD candidates, faculty, and whole departments.
Start a conversation
