Mixed Methods Research Explained: Designs, Uses and Pitfalls

Some questions are too big for one kind of evidence. If you want to know whether a mentoring program improved nurse retention, numbers can show whether it did. To find out why, and what the nurses made of it, you need to hear from them. Mixed methods research combines both in a single, planned study.

It can produce fuller answers, and it can also produce a project that is twice as much work and half as clear. This guide explains when mixed methods is worth it, the main designs, how to integrate the strands and how to avoid the common traps.

What mixed methods is, and is not

A mixed methods study has three features:

1. Both types of data. Quantitative data, such as survey scores, and qualitative data, such as interview transcripts.
2. A purpose for combining them. Each strand does something the other cannot.
3. Integration. The results are brought together in analysis, interpretation or both.

It is not two separate studies placed side by side, and it is not a survey with a few open comments added to look thorough. If the strands do not speak to each other, it is not mixed methods in the full sense.

Why combine methods

Researchers mix methods for several reasons.

• Explanation: numbers show a pattern, and interviews explain it.
• Exploration: interviews reveal what matters, and a survey measures how widespread it is.
• Triangulation: two kinds of data confirm, or challenge, the same finding.
• Completeness: one method covers breadth, and the other adds depth.
• Development: early qualitative work helps build better survey questions or interventions.

If you cannot state in one sentence what each strand adds, reconsider.

Single method or mixed?

Our guide to qualitative vs quantitative research covers the two strands separately. Mixed methods rests on being able to do both well, or having a supervisor who can guide you.

The main designs

Researchers commonly describe a few core designs.

Convergent parallel

You collect quantitative and qualitative data at about the same time, analyze them separately and then compare the results to see whether they agree. This suits questions where you want two views of the same issue, and it can save time because the strands run together. The challenge is integrating results that may not line up neatly.

Explanatory sequential

You collect and analyze quantitative data first, then use qualitative data to explain or deepen the results. For example, a survey finds that students in one program report lower satisfaction, and follow-up interviews explore why. This is often the easiest design for students, since the second strand has a clear job.

Exploratory sequential

You begin with qualitative data to explore a topic, then use the findings to build a quantitative instrument or intervention, which you test in a second stage. For example, interviews with new teachers identify the main sources of stress, and these themes shape a survey that measures them in a larger group. It suits new topics where measures do not yet exist, but takes the longest.

Embedded

One strand sits inside a larger study of the other type. For instance, an experiment includes a small set of participant interviews to understand how the intervention was experienced. The embedded strand supports the main one.

Choosing a design

Ask yourself:

1. What is the main question, and which strand answers it?
2. Does one strand need to come before the other?
3. How much time do I have for each?
4. Do I have access to participants twice, if the design is sequential?
5. Can I analyze both types of data competently?
6. What will integration look like?

If in doubt, explanatory sequential is usually the most manageable choice for a student project.

Sampling in mixed methods

You may need two kinds of samples: a larger one for the quantitative strand and a smaller, purposeful one for the qualitative strand. In sequential designs, the second sample is often drawn from the first, such as interviewing survey respondents who volunteered. Be clear about how the samples relate, because it affects how you can combine the findings. Our guide to sampling methods explained covers the options.

Integration: the heart of mixed methods

Integration is what makes the study more than the sum of its parts. It can happen at several points.

• At design: the questions and instruments are planned to connect.
• At data collection: one strand informs the other, as in a sequential design.
• At analysis: results are compared, merged or used to explain each other.
• At interpretation: the discussion draws conclusions from both.

Useful ways to show integration include a joint display, a table that places quantitative results beside related qualitative themes, and narrative explanation that moves between the two. For example, a table might list each survey scale with its average score on one side and illustrative quotations on the other.

Reporting a mixed methods study

Structure the write-up so that readers can follow both strands and see how they connect.

1. Introduce the question and explain why mixed methods suits it.
2. Describe the design and its sequence, perhaps with a simple diagram.
3. Report each strand's methods, including sampling, instruments and analysis.
4. Report results by strand, or by question.
5. Integrate in a joint display or a dedicated section.
6. Discuss what the combined evidence shows, and where the strands agree or differ.
7. Note limitations of both strands and of the integration.

Make sure the methodology chapter explains why each strand is needed, using the language of the design you chose. Our guide to the methodology chapter covers the standard sections.

A short worked example

A researcher wants to understand why some first-year students leave a university before the second year. She uses an explanatory sequential design.

1. Quantitative strand: she analyzes institutional data and a survey of 400 first-year students on engagement, finances and sense of belonging. She finds that low sense of belonging is the strongest predictor of intention to leave.
2. Qualitative strand: she interviews 12 students with low belonging scores to explore what contributes to the feeling.
3. Integration: she presents a joint display that links each survey factor to interview themes such as feeling invisible in large lectures, and discusses implications.

(The example is illustrative.) Each strand has a clear job, and the interviews explain the pattern the numbers revealed.

Pitfalls

• No clear rationale for mixing.
• Underestimating the workload, especially for two stages of data collection.
• Weak integration, so the strands remain separate.
• Weak execution of one strand, often the qualitative one added on to look rigorous.
• Mismatched samples that cannot be compared.
• Overlong write-up that tries to report everything.
• Poor timing, such as discovering that the second round of access is not allowed.
• Skills gaps in either method.

Ethics and approvals

Mixed methods studies often need ethics approval for each stage, and sequential designs may require consent for follow-up contact at the start. Plan for this in your application, and explain how data from both strands will be stored and linked. Our guide to research ethics and ethics approval covers the process.

A quick checklist

• Can I state what each strand adds?
• Have I chosen a design that fits my question and timeline?
• Are the samples and their relationship clear?
• Do I have a plan for integration, such as a joint display?
• Do I have the skills, or support, for both strands?
• Is my ethics approval in place for each stage?
• Can I finish both strands in the time available?

Final thoughts

Mixed methods is a powerful approach when a question truly needs both numbers and narratives, and a risky one when it is chosen for appearances. Pick the simplest design that fits, give each strand a clear job and plan integration from the start. A well-designed mixed study answers questions that no single method could, and a well-designed single-method study is still better than a muddled mixed one.