Mixed methods research: 4 designs, examples and when to use it

Mixed methods research combines qualitative and quantitative data collection in the same study. The qualitative component provides depth and context — it explains why. The quantitative component provides breadth and precision — it measures how much. Neither approach alone can answer certain research questions; mixed methods is appropriate when both dimensions of understanding are necessary. The 4 main designs are: sequential explanatory, sequential exploratory, concurrent triangulation, and embedded.

What is mixed methods research?

Mixed methods research integrates qualitative and quantitative data within a single study to investigate a research question that neither approach could answer adequately on its own. The rationale is not simply to use both types of data, but to integrate them so that each informs and deepens the other.

In a widely cited analysis, Alan Bryman examined 232 mixed methods studies published in social science journals and found that the most common justification for combining methods was that qualitative research explained quantitative findings that the numbers alone could not account for. The second most common justification was that qualitative work helped design a subsequent quantitative phase by identifying the right variables and categories to measure.

Mixed methods research is used across disciplines — education, health sciences, psychology, organisational research, policy analysis, and social work — as well as in applied contexts including market research, user research, and programme evaluation.

How does mixed methods research differ from a study that simply uses two methods?

The defining feature of mixed methods research is integration, not just combination. A study that runs a survey and separately reports interviews without connecting them is not genuinely mixed methods — it is two parallel studies reported in the same paper.

Integration means that the findings from one strand inform, explain, or extend the findings from the other. The quantitative results raise questions that the qualitative data answers. The qualitative themes suggest variables that the quantitative analysis measures. The two strands triangulate: they approach the same question from different directions, and the convergence (or divergence) of their findings produces a richer understanding than either could alone.

What are the 4 main mixed methods designs?

1. Sequential explanatory

The quantitative phase comes first, followed by a qualitative phase designed to explain the quantitative results.

How it works: You run a survey and find, for example, that employee satisfaction scores are significantly lower in one department than in all others. The survey cannot explain why. You then conduct a series of interviews in that department to understand what is driving the divergence.

When to use it: When you have quantitative findings that are surprising, puzzling, or require contextual explanation. The qualitative phase is explicitly designed to answer the "why?" behind the numbers.

Example: A university surveys graduate employability outcomes across programmes. Programme X consistently shows lower employment rates. The sequential explanatory design adds interviews with graduates from that programme to understand what is happening.

2. Sequential exploratory

The qualitative phase comes first, followed by a quantitative phase that tests or extends the qualitative findings.

How it works: You conduct interviews to understand a phenomenon you know little about. The interview analysis surfaces themes and generates hypotheses. You then design a survey instrument to test whether those themes and relationships hold at scale, across a larger population.

When to use it: When you are building new theory or measurement tools in an area where existing instruments do not fit. The qualitative phase ensures that what you subsequently measure reflects the lived experience of participants rather than researcher assumptions.

Example: A health service is developing a patient experience questionnaire for a condition where no validated measure exists. Interviews with patients identify the dimensions of experience that matter most. The survey is then built around those dimensions and administered to a large sample.

3. Concurrent triangulation

Both qualitative and quantitative data are collected simultaneously, and findings are compared to assess convergence or divergence.

How it works: You collect survey data and interview data at the same time, analyse them separately, and then compare the findings. Where the two strands agree, confidence in the findings is strengthened. Where they diverge, the divergence itself becomes analytically interesting — why do participants say one thing in interviews and behave differently as measured by the survey?

When to use it: When you want to cross-validate findings through different data types, and when you have the resources to run both strands simultaneously without one informing the design of the other.

Example: A study of remote work effectiveness collects productivity metrics (quantitative) and conducts interviews about the experience of remote work (qualitative) at the same time. The analysis compares what the productivity data shows with what workers report feeling.

4. Embedded design

One type of data is nested within the other — qualitative data collected within a primarily quantitative study, or quantitative data collected within a primarily qualitative study.

How it works: The secondary strand provides a supplementary function rather than an equal one. A clinical trial (quantitative) might embed a qualitative component to understand how participants experience the treatment — not as the primary outcome, but as context that enriches interpretation.

When to use it: When one strand is clearly primary and the other provides supporting depth or context, without requiring full integration of findings.

Design	Sequence	Primary strand	Best for
Sequential explanatory	Quant → Qual	Quantitative	Explaining surprising quantitative findings
Sequential exploratory	Qual → Quant	Qualitative	Building theory, developing measurement tools
Concurrent triangulation	Both simultaneous	Equal	Cross-validating findings
Embedded	Nested	One dominant	Adding supplementary depth to a primary study

When should you use mixed methods research?

Mixed methods is appropriate when:

Neither approach alone can answer your research question. If you want to understand both the prevalence of a phenomenon and the meaning it holds for those who experience it, you need both measurement and interpretation.

You want to cross-validate findings. Triangulating qualitative and quantitative findings builds confidence in results — and surfacing divergence between the two strands produces interesting analytical questions.

Your study involves developing a measurement instrument. Before you can survey at scale, you need to know what to measure. Qualitative research identifies the relevant constructs; quantitative research then measures them.

One method is insufficient on its own. A survey showing that programme satisfaction is low tells you something. Interviews with dissatisfied participants tell you what to do about it.

What are the challenges of mixed methods research?

Requires breadth of skills. A researcher needs competence in both qualitative analysis and quantitative methods. In practice, mixed methods research often involves teams with complementary expertise.

Integration is harder than it looks. Collecting both types of data is straightforward. Genuinely integrating the findings — letting each strand inform the other in a meaningful way — requires careful design and deliberate analysis.

More time and resources. Two methodological strands, two analytical processes, and the integration work between them add up. Mixed methods is not the right choice when one approach would suffice.

Publishing challenges. Mixed methods papers are longer and more complex than single-method papers. Some journals specialise in one methodological tradition and are resistant to publishing mixed methods work. This is changing, but it remains a practical consideration in academic research.

How do you analyse mixed methods data?

Each strand is analysed using its appropriate methods first: quantitative data with statistical techniques, qualitative data with thematic analysis, grounded theory, or another qualitative approach. Integration happens at the level of interpretation.

Common integration strategies include:

Narrative integration: The write-up weaves together findings from both strands, showing where they converge and where they diverge.
Data transformation: Quantitative data is converted into qualitative data (by describing patterns in narrative terms) or qualitative data is converted into quantitative data (by counting how often themes appear across participants). Integration then happens at a single level.
Joint display: Both strands' findings are presented side by side in a matrix or table, making comparisons visible.

For large datasets involving qualitative components — multiple rounds of interviews across different sites, for example — AI-assisted analysis can substantially reduce the coding and synthesis burden on the qualitative strand. See how to analyse interview transcripts for practical guidance.

Beyond analysis of existing qualitative data, AI-powered tools have introduced a newer pattern that changes the qual/quant boundary itself: qualitative research at scale. Skimle Ask conducts AI-powered conversational interviews — following up on interesting answers, probing when responses are vague, and adapting to each participant — making it possible to gather interview-depth data from hundreds of participants rather than the 10-30 typical of conventional qualitative research. For mixed methods researchers, this particularly changes the embedded design: the qualitative strand no longer has to be a small nested component inside a large quantitative study; it can stand alongside it at comparable scale, with genuine thematic depth across a large sample.

Quantitative understanding can also flow in the other direction — out of qualitative data, using metadata variables. By tagging each interview or open-ended response with participant attributes — role, region, product line, tenure — Skimle lets you cross-tabulate themes against those variables and identify which findings are universal and which are concentrated in a specific subgroup. This produces a form of mixed-methods integration that happens inside the qualitative analysis itself: thematic findings with a quantitative distribution behind them, without requiring a separate survey strand.

Academic researchers working in health, education, or social policy research are the most likely to encounter formal mixed methods design requirements. For market researchers and HR teams, a lighter version of the sequential explanatory approach — survey to understand the scale, interviews to understand the cause — is a common and practical pattern.

Frequently asked questions

Is mixed methods research qualitative or quantitative?

It is both. Mixed methods research integrates qualitative and quantitative data rather than choosing between them. The mix does not have to be equal: many mixed methods studies have a dominant strand (usually quantitative) with a supporting qualitative component, or vice versa.

How do you write up mixed methods research?

Structure your write-up around the integration of findings, not just the presentation of two separate analyses. In the methods section, explain the design (sequential, concurrent, embedded), why you chose it, and how the strands relate. In the results section, present each strand and then explicitly show how they connect. The integration — where findings converge or diverge — is the methodological contribution.

What is the difference between triangulation and mixed methods?

Triangulation is one mixed methods design (concurrent triangulation), in which findings from qualitative and quantitative strands are compared to assess convergence. Mixed methods research is the broader category that includes all designs that combine qualitative and quantitative approaches, of which triangulation is one.

Does mixed methods research require a larger sample size?

The quantitative strand requires a sample large enough for statistical analysis, which is typically much larger than qualitative samples. The qualitative strand follows qualitative sampling logic — purposive selection, typically 10-30 participants. The overall study therefore tends to have a much larger total sample than a purely qualitative study, driven by the quantitative requirements.

Analysing qualitative data as part of a mixed methods study? Try Skimle for free — systematic analysis of interview and open-ended data that integrates with your quantitative findings.

Related reading:

About the authors

Henri Schildt is a Professor of Strategy at Aalto University School of Business and co-founder of Skimle. He has published over a dozen peer-reviewed articles using qualitative methods, including work in Academy of Management Journal, Organisation Science, and Strategic Management Journal. His research focuses on organisational strategy, innovation, and qualitative methodology. Google Scholar profile

Olli Salo is a former Partner at McKinsey & Company where he spent 18 years helping clients understand the markets and themselves, develop winning strategies and improve their operating models. He has done over 1000 client interviews and published over 10 articles on McKinsey.com and beyond. LinkedIn profile