Sampling Strategy for Mixed Methods Research

Sampling in mixed methods research means making two sets of decisions instead of one. Your quantitative strand needs a sample that supports statistical inference. Your qualitative strand needs a sample that supports depth and meaning. And you need to decide how those two samples relate to each other.

Getting sampling wrong undermines the entire study. A qualitative sample that doesn't connect to your quantitative findings produces insights you can't integrate. A quantitative sample that's too small for the subgroup comparisons you planned limits what your qualitative phase can target.

Here's how to get it right.

The Core Decision: Same Participants or Different?

Before selecting specific participants, answer one question: Will the same people contribute to both strands, or will you draw separate samples?

Same-Participant Design

Every person in your study provides both quantitative and qualitative data.

Advantages:

• Strongest foundation for integration, you can link individual survey responses to interview themes
• Easier to explain to stakeholders ("We surveyed them and then interviewed a subset")
• Supports individual-level joint displays

Disadvantages:

• Increases participant burden (especially if both strands require significant time)
• Can introduce order effects (the survey may prime interview responses, or vice versa)
• Requires participants who are willing to engage in both activities

Best for: Sequential explanatory design, where survey respondents are selected for follow-up interviews; embedded design, where both data types come from the same instrument.

Different-Participant Design

Separate groups contribute to each strand. Both groups are drawn from the same population but don't overlap.

Advantages:

• No participant burden from completing both strands
• No order effects or priming between methods
• Can optimize each sample independently for its strand

Disadvantages:

• Integration happens at the group level, not the individual level
• Harder to trace specific connections between quant and qual findings
• Requires clear definition of the shared population

Best for: Concurrent triangulation and convergent design, where strands run independently.

Quantitative Sampling

The quantitative strand follows standard sampling principles. The goal is representativeness and statistical power.

Key decisions:

• Sample size. Calculate based on the analyses you plan to run. For a straightforward survey, 200-400 respondents often suffice. For conjoint or MaxDiff studies, sample requirements depend on the number of attributes and levels. For subgroup comparisons, you need enough participants in each subgroup.

• Sampling method. Probability sampling (random, stratified, cluster) supports generalizability. Convenience or panel samples are faster but limit how broadly you can apply your findings. Stratified sampling is particularly useful in mixed methods because it ensures you'll have enough participants in each segment for both quantitative analysis and qualitative follow-up.

• Over-recruit for qualitative follow-up. If you're using a sequential design and plan to interview survey respondents, include a consent question in your survey ("Would you be willing to participate in a 30-minute follow-up interview?"). Expect that 30-50% will say yes and that only 50-70% of those will actually schedule. Plan accordingly.

Qualitative Sampling

The qualitative strand uses purposeful sampling, selecting participants deliberately based on specific criteria, not randomly.

Common Purposeful Sampling Strategies

Maximum variation sampling: Select participants who represent the widest range of experiences or characteristics. Useful when you want your qualitative findings to cover the breadth of your population.

Extreme/deviant case sampling: Select participants at the extremes of your quantitative distribution (highest and lowest satisfaction scores, heaviest and lightest product users). Useful in sequential explanatory design when you want to understand outliers.

Typical case sampling: Select participants who represent the most common pattern in your quantitative data. Useful when you want to understand the "average" experience in depth.

Confirming/disconfirming case sampling: Select participants whose experiences appear to confirm or challenge a pattern you've seen in the quantitative data. Useful for testing the strength of your quantitative findings.

Homogeneous sampling: Select participants who share a key characteristic (same segment, same usage pattern, same satisfaction level). Useful for focus groups where group discussion benefits from shared context.

Qualitative Sample Size

There's no universal formula. Guidelines by study type:

For mixed methods specifically, qualitative sample sizes tend to be on the lower end of these ranges because the qualitative phase is typically focused (targeted by quant findings) rather than broadly exploratory.

Sampling Across Mixed Methods Designs

Sequential Explanatory

Quant sample: Full probability or stratified sample (200-500+). Qual sample: Purposefully selected subset of quant respondents (10-25). Selection criteria come directly from quantitative findings, you interview the people whose responses you need to understand.

Key consideration: Build in the recruitment path during the quantitative phase. Collect contact info and consent for follow-up.

Concurrent Triangulation

Quant sample: Full probability or stratified sample. Qual sample: Separate purposeful sample from the same population, or the same participants if burden is manageable.

Key consideration: Both samples should be drawn from the same defined population, or integration at the comparison stage won't be valid.

Embedded Design

Primary sample: Designed for the primary strand (usually quantitative). Embedded sample: Usually the same participants, since the embedded data comes from the same instrument (e.g., open-ended questions within a survey).

Key consideration: If the embedded strand involves separate data collection (e.g., interviews with a subset), apply purposeful sampling from the primary sample.

Convergent Design

Quant sample: Probability or stratified. Qual sample: Purposeful, from the same population.

Key consideration: Sample sizes should be proportional to each strand's analytical needs. Don't shortchange the qualitative sample just because the quantitative sample is large.

Getting Sampling Right in Practice

The operational challenge of mixed methods sampling is tracking participants across strands. In sequential designs, you need to identify survey respondents, contact them for interviews, and link their survey data to their interview data. In concurrent designs with the same participants, you need to manage scheduling across two data collection processes.

Quali-Fi simplifies this because participant management for surveys, focus groups, and IDIs happens in one system. Flag survey respondents for qualitative follow-up based on their answers, and their data stays linked automatically. No spreadsheet matching required.

For guidance on analysis after sampling, see our mixed methods data analysis guide.

Plan your mixed methods sampling on Quali-Fi

FAQs

Do I need the same participants for both strands of a mixed methods study?

Not always. Same-participant designs support stronger integration at the individual level, but different-participant designs are valid and sometimes preferable, especially in concurrent designs where you want each strand to be independent. The right choice depends on your mixed methods design and integration goals.

How many participants do I need for the qualitative strand?

For interviews, 10-30 is typical; for focus groups, 3-6 groups. In mixed methods, qualitative samples are often on the smaller end of these ranges because the qualitative phase is usually focused on explaining specific quantitative findings rather than broad exploration. Recruit until you stop hearing new themes.

What is purposeful sampling in mixed methods?

Purposeful sampling means selecting participants deliberately based on specific criteria rather than randomly. In mixed methods, the criteria often come from your quantitative findings, you interview people who represent certain patterns, outliers, or segments in the quantitative data.

How do I recruit survey respondents for follow-up interviews?

Include a consent question in your survey asking whether participants are willing to do a follow-up interview. Collect their contact information at that point. Expect about 30-50% to agree, and about half of those to actually participate when contacted. Over-recruit to account for dropout.