Qualitative or quantitative? The decision sits at the front of every empirical study and shapes everything that follows: the research question, the data you collect, the analysis you run, the kind of claim you can defend. Treat it as a real choice with consequences, not a label to attach later.
This guide explains what each approach is, when to use them, how they differ in practice, and how mixed-methods designs combine the two without diluting either.
Need help choosing your research method?
ThesisAI helps you match your research question to a method, weigh the trade-offs, and design a study you can actually defend.
Start WritingWhat is Qualitative Research?
Qualitative research investigates meaning, experience, and process. It collects rich, often textual or visual data - interview transcripts, observation notes, documents, images - and analyses it interpretively to understand how people make sense of something.
Common qualitative methods include:
- Semi-structured interviews - in-depth conversations around a flexible topic guide.
- Focus groups - small-group discussions that surface shared and contested meanings.
- Ethnography - sustained observation of a setting, often over months.
- Case studies - detailed analysis of a single instance (person, organisation, event).
- Document and discourse analysis - examining texts for meaning, framing, or rhetoric.
Typical analysis approaches: thematic analysis, grounded theory, narrative analysis, discourse analysis, interpretive phenomenological analysis.
What is Quantitative Research?
Quantitative research measures variables and tests relationships using numerical data. It operationalises concepts as measurable indicators, collects structured data, and uses statistical analysis to estimate effects or test hypotheses.
Common quantitative methods include:
- Surveys - structured questionnaires delivering numerical or categorical responses.
- Experiments - random assignment to conditions to test causal effects.
- Quasi-experiments - comparison across naturally occurring groups when random assignment is not possible.
- Secondary data analysis - using administrative, archival, or large-scale survey data.
- Content coding - converting documents or behaviours into countable categories.
Typical analysis approaches: descriptive statistics, regression, ANOVA, factor analysis, structural equation modelling, time-series analysis.
The Core Differences (Side-by-Side)
Most differences trace back to one root: qualitative work asks how and why; quantitative work asks how much, how often, and under what conditions.
| Dimension | Qualitative | Quantitative |
|---|---|---|
| Goal | Understand meaning, experience, process | Measure variables, test relationships |
| Question form | How? Why? In what ways? | How much? How often? Does X cause Y? |
| Data type | Words, images, observations | Numbers, structured categories |
| Sample size | Small, purposive (5-50 typical) | Larger, often probabilistic (often 100+) |
| Sampling logic | Information richness | Statistical representativeness |
| Analysis | Interpretive, iterative | Statistical, hypothesis-driven |
| Claim type | Transferable, contextual | Generalisable, estimated |
| Researcher role | Acknowledged influence, reflexive | Aim for objectivity, replicable procedure |
| Reporting | Narrative, with quotes | Tables, figures, p-values, effect sizes |
When to Use Qualitative Research
Qualitative is the right fit when:
- The research question is about meaning, experience, or process.
- The phenomenon is new, under-theorised, or hard to operationalise as variables.
- You need to understand how something works, not just whether it works.
- The setting is bounded - one organisation, one community, one policy - and the goal is depth, not breadth.
- You expect the relevant categories to emerge from the data rather than from pre-existing theory.
Example research questions that suit qualitative work
How do first-year medical students manage the transition from classroom to clinical placement?
In what ways do remote workers reconstruct boundaries between work and home life?
How do school leaders interpret and adapt national accountability policies in practice?
When to Use Quantitative Research
Quantitative is the right fit when:
- The research question is about magnitude, frequency, or causal effect.
- The relevant concepts can be operationalised with valid measures.
- You can collect data from a sample large enough to estimate the effect of interest with acceptable precision.
- You want to generalise beyond your immediate sample.
- You are testing predictions derived from theory or prior work.
Example research questions that suit quantitative work
Does a four-week mindfulness intervention reduce self-reported anxiety in undergraduates?
How is parental education associated with university completion rates across OECD countries?
Do hybrid work arrangements affect engineering team productivity over a three-year window?
Side-by-Side: The Same Topic, Two Designs
The same research interest can be approached either way; what changes is the question, the data, and the kind of claim you end up able to make.
Topic: Smartphone use and adolescent wellbeing
Quantitative design
Question: Is daily smartphone screen time associated with self-reported wellbeing scores among 14-17 year-olds, controlling for sleep duration, family income, and gender?
Method: Survey of 1,800 adolescents across 12 secondary schools; multilevel regression with screen-time as predictor and the WEMWBS score as outcome.
Claim type: Estimated effect size and confidence interval, generalisable within the target population, with limits on causal inference owing to cross-sectional design.
Qualitative design
Question: How do 14-17 year-olds describe the role of their smartphone in their daily emotional life, and what practices do they develop to manage it?
Method: Twenty-four semi-structured interviews with adolescents from three secondary schools, analysed thematically.
Claim type: Transferable account of how smartphone use is woven into everyday emotional regulation, not a statistical estimate; useful for theory-building and intervention design.
Neither design is "better". They answer different questions. The quantitative version tells you the size of the average association; the qualitative version tells you how the relationship is lived. If your literature has plenty of the first and little of the second, the second is the higher-value contribution - and vice versa.
Mixed-Methods: When and How to Combine Them
Mixed-methods designs use both approaches deliberately, not as window dressing. They are most valuable when one method genuinely cannot answer the full question alone.
Common Mixed-Methods Designs
- Sequential exploratory (QUAL then QUANT): Start qualitative to identify the relevant concepts, then build and test a survey or model. Useful when the field lacks validated instruments.
- Sequential explanatory (QUANT then QUAL): Start quantitative to identify the patterns, then use interviews to interpret unexpected findings. Useful when the numbers raise more questions than they answer.
- Concurrent triangulation: Collect both kinds of data in parallel and compare. Useful when convergence (or divergence) is itself the finding.
- Embedded designs: One method plays a supporting role inside a study dominated by the other.
What Mixed-Methods Is Not
Mixed-methods is not "I did some interviews and also a survey". To count as a mixed-methods design, the two strands must be planned together, address connected questions, and be integrated in the analysis or discussion - not just reported back-to-back.
Common Mistakes (And How to Fix Them)
Mistake 1: Choosing Method by Familiarity, Not by Question
Problem: "I always do interviews, so this will be a qualitative study." If the research question is about effect size, that is the wrong fit.
Fix: Write the research question first. Then ask: does answering it require numbers, words, or both?
Mistake 2: Treating Qualitative as a Small Quantitative Study
Problem: "Six of the twelve participants said X, which means 50% of people think X." Qualitative samples are not designed for that arithmetic.
Fix: Report patterns interpretively ("most participants framed X as..."), explain what their accounts illuminate, and avoid percentage claims that read as quantitative.
Mistake 3: Treating Quantitative as Self-Justifying
Problem: "We ran a regression, so the findings are objective." Statistical procedures only produce meaningful claims when the measures and design are sound.
Fix: Justify the measures, the sample, the model, and the assumptions. A weak measure in a strong model still yields a weak claim.
Mistake 4: Bolting on Method to Tick a Box
Problem: Adding three interviews to a survey study just to call it "mixed methods", without integrating the findings.
Fix: Use mixed methods only when one of the recognised designs above genuinely fits your question. Otherwise, do one method well.
Mistake 5: Confusing Sample Size Rules
Problem: "My qualitative study had only fifteen participants, so the sample is too small." Or "My quantitative study has fifty participants, that's plenty."
Fix: Qualitative sample size is judged on information richness and saturation, not statistical power. Quantitative sample size is judged on a power calculation tied to the effect size you expect to detect. Use the correct logic for the method you chose.
How to Decide: A 5-Step Process
Step 1: Write the Research Question in One Sentence
If it starts with "how" or "in what ways", lean qualitative. If it starts with "how much", "how often", or "does X cause Y", lean quantitative.
Step 2: Identify the Type of Claim You Need to Make
Estimated effect, generalisable pattern, or causal claim? Quantitative. Interpretive understanding, process account, or theory-building? Qualitative.
Step 3: Check Whether Valid Measures Exist
If your concepts have validated, established instruments, quantitative is more feasible. If they do not, you may need a qualitative study first - or you need to develop and validate measures.
Step 4: Check Feasibility
Can you reach a sample large enough for the quantitative claim you want to make? Can you sustain the depth and access needed for the qualitative claim you want to make? Match ambition to what is actually possible.
Step 5: Decide on a Single Method or a Justified Mixed Design
Default to a single method done well. Choose mixed methods only if a recognised mixed design genuinely fits the question.
Quality Criteria for Each Approach
The criteria differ because the goals differ.
Quantitative Quality Criteria
- Validity: Are the measures measuring what they claim to measure?
- Reliability: Are the measures consistent across raters, items, time?
- Generalisability: Does the sample allow inferences about the target population?
- Replicability: Could another team reproduce the result with the same materials?
Qualitative Quality Criteria
- Credibility: Are the interpretations grounded in the data?
- Transferability: Is the context described well enough for a reader to judge fit to their own setting?
- Dependability: Is the analytic process documented and traceable?
- Reflexivity: Has the researcher's positionality been considered and mitigated?
FAQs About Qualitative vs Quantitative Research
Is qualitative research less rigorous than quantitative?
No. They are judged by different criteria. Rigorous qualitative work is as demanding as rigorous quantitative work; weak versions of either are weak.
Is quantitative always more objective?
No. Quantitative researchers also make consequential decisions - what to measure, how to operationalise it, which variables to include. Objectivity is a goal of method, not a property of numbers.
Can a thesis use both methods?
Yes, in a mixed-methods design. The two strands must be clearly justified, planned together, and integrated in the discussion.
How big should my qualitative sample be?
Most interview-based studies sit in the 8-30 range, with saturation as the practical guide. Ethnographic and case-study work may use far fewer cases analysed in more depth.
How big should my quantitative sample be?
It depends on the expected effect size and the test you plan. A power calculation - usually for 80% power to detect your minimum effect of interest - is the standard way to justify a sample size.
What if my topic has both quantitative and qualitative literature?
Read both, then pick the method that addresses the gap you find most informative. The literature is asking you to extend, qualify, or challenge it - not just replicate.
The choice between qualitative and quantitative is not a worldview decision. It is a fit-for-purpose decision: start with the question, choose the method that answers it, and defend the trade-offs honestly.