Were the Quantitative and Qualitative Samples Appropriately Connected?

How should the quantitative and qualitative samples relate to each other?

Mixed methods studies often involve two samples that look very different. A quantitative phase may include hundreds or thousands of participants, while a qualitative phase includes only a few dozen. Sometimes the interview participants come directly from the quantitative sample. Sometimes they do not.

Neither arrangement is automatically wrong. The important question is whether the relationship between the samples fits what the researchers are trying to learn.

This becomes especially consequential when researchers use one component to explain another. If the qualitative participants bear little relationship to the people who produced the quantitative pattern, for example, their accounts may not be able to explain that pattern in the way the researchers claim.

The samples should be connected in a way that matches the mixed methods purpose

Sample connection is part of the logic of integration

Connecting is a recognized form of mixed methods integration

Mixed methods integration does not occur only when researchers combine findings at the end. It can begin with sampling.

Fetters, Curry, and Creswell describe connecting as an integration strategy in which one dataset links to the other through sampling. A common example occurs in an explanatory sequential design: researchers analyze quantitative results and use those results to select participants for subsequent qualitative inquiry.

This means that evaluating whether a study integrated its quantitative and qualitative components may require examining participant selection long before you reach the integrated findings.

The samples do not have to be identical

A mixed methods study does not ordinarily require every participant to contribute both quantitative and qualitative data. Several sampling relationships can be methodologically defensible.

Mixed methods methodological literature describes sampling relationships including identical, nested, and separate samples. The appropriate relationship depends on what is being integrated and what inference researchers intend to make.

Start by asking what the qualitative sample is supposed to accomplish

Suppose a quantitative survey identifies an unexpected group of participants: people who report high access to educational technology but unusually low use. Researchers then want interviews to understand why.

A purposive qualitative sample drawn from that particular quantitative profile makes methodological sense. Randomly interviewing participants from the entire survey sample could dilute the very phenomenon the second phase was designed to investigate.

This is one reason quantitative results may be used strategically rather than merely administratively when selecting qualitative participants. Research on integration through connecting has demonstrated how participant profiles can be constructed from quantitative findings and then used to select cases representing theoretically useful patterns, including cases that converge with or diverge from an expected model.

The sampling decision should therefore follow from the role assigned to the component. That role should already be visible when evaluating whether each component addresses a meaningful part of the research question.

A smaller qualitative sample is not automatically a weakness

Comparing sample sizes numerically can be misleading because quantitative and qualitative sampling often serve different purposes.

A survey may require a relatively large sample to estimate population parameters, examine associations, compare groups, or fit statistical models with adequate precision. Qualitative inquiry commonly uses smaller purposively selected samples because its analytic purpose may involve depth, variation, processes, meanings, or case-level explanation rather than statistical estimation.

The relevant question is therefore not whether 20 interviewees can somehow “match” 1,000 survey respondents. Ask whether those 20 participants were selected appropriately for the qualitative question and whether researchers restrict their conclusions accordingly.

Connection should preserve the cases that matter

In explanatory sequential research, researchers may want qualitative data precisely because the quantitative results contain something requiring explanation: an outlier, an unexpected association, contrasting profiles, subgroup differences, or cases that do not fit an anticipated pattern.

Participant selection should preserve access to those cases.

For example, if researchers want to understand why some high-performing students nevertheless report low academic belonging, the qualitative sample should include students who actually exhibit that combination. Interviewing only “typical” students would weaken the explanatory connection.

Attrition can break an otherwise sensible connection

A sampling strategy can look appropriate on paper and deteriorate during recruitment.

Imagine that researchers identify four contrasting quantitative profiles and plan to interview participants from each. Nearly everyone from three profiles agrees to participate, but very few members of the fourth respond. The final qualitative sample no longer represents the contrasts on which the integration strategy depended.

This does not necessarily invalidate the qualitative component. It does mean that the intended connection and the achieved connection are different. Researchers should disclose the difference and moderate claims that rely on the missing group.

Separate samples can still be appropriate

Sometimes the two components do not need person-level correspondence.

Researchers evaluating implementation of a university program might analyze student outcome data while interviewing instructors and administrators. These are deliberately different samples because the study seeks evidence from different positions within the same system.

Likewise, one component may characterize a population while another examines institutional processes using participants who were not members of the quantitative sample.

The crucial question is whether the integrated inference requires the participants to be the same. If the researchers claim that interviews explain why particular survey respondents answered as they did, separate samples would create a serious problem. If the interviews instead illuminate organizational conditions surrounding the quantitative pattern, different samples may be entirely appropriate.

Sampling connection and finding integration should work together

Good sample connection creates opportunities for stronger analysis, but it does not finish the job.

If researchers purposively select interviewees from contrasting quantitative profiles, the subsequent qualitative analysis should make use of those profiles. The researchers might compare narratives across groups, investigate anomalous cases, or construct case-level analyses linking quantitative characteristics with qualitative accounts.

If the quantitative profiles disappear completely once interviews begin, the study may have connected its samples without fully exploiting that connection analytically.

Published methodological work on explanatory sequential research illustrates this point well: quantitative participant profiles can support not only participant selection but tailored interview guides and later analysis of convergence and divergence across components.

Sampling should follow the explanation the study is seeking

Common mistakes when evaluating mixed methods samples

Trace the sampling logic from question to conclusion

When evaluating a mixed methods study, reconstruct why each sample exists and how participants moved, or did not move, between components. Do not assume that overlap is automatically good or separation automatically bad.

Sampling should ultimately make the intended integration possible. If the study later claims that one method explains, expands, or contradicts the other, the samples must provide a defensible basis for making that comparison.

Check whether the sampling relationship supports the study's claims

Questions about sampling in mixed methods research

The right connection depends on what the samples need to accomplish together

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.