How Do You Connect Samples Across Phases in Sequential Mixed-Methods Research?

How Do You Make Participants From Different Research Phases Meaningfully Connected?

You have completed a quantitative survey and identified students whose learning engagement patterns deserve further investigation. You now want to interview selected students, but how will you connect their interview accounts to the original survey findings?

Perhaps your survey was anonymous. Perhaps only some respondents agreed to further contact. Or perhaps the qualitative and quantitative phases involve different participants, even though their findings are intended to address the same research problem.

These are not merely recruitment difficulties. They concern whether the relationship between samples supports the integration your study promises.

Connecting samples across phases requires a defensible sampling rationale, appropriate procedures for identifying or relating cases, and clarity about the conclusions that can be drawn from the resulting evidence.

Connect Samples According to the Relationship Your Research Question Requires

Building Defensible Sampling Connections Between Research Phases

What Does Connecting Mean in Mixed-Methods Methodology?

Connecting is a methods-level integration strategy in which information from one research component informs sampling in another. Fetters, Curry, and Creswell (2013) distinguish connecting from building, merging, and embedding as different approaches to mixed-methods integration.

In a typical explanatory sequential study, quantitative findings may identify participants for qualitative follow-up. For example, survey results could identify students whose engagement and performance profiles differ in theoretically important ways. Researchers then purposively recruit students representing those profiles for interviews.

Connecting does not necessarily mean that all participants appear in both datasets. It means that the sampling relationship is established deliberately to serve the research purpose.

It is also useful to distinguish sampling connection from record linkage. Sampling connection concerns how one phase informs participant selection. Record linkage concerns whether observations from different phases can be associated with the same identifiable or pseudonymized case.

A study may establish a meaningful sampling connection without linking every individual record. Conversely, a researcher may link records successfully without having a defensible reason for selecting those participants.

Which Sampling Relationships Can Be Used?

Several arrangements are possible. The appropriate choice depends on whether the study requires case-level explanation, group-level interpretation, or development of a subsequent instrument or intervention.

These arrangements are not interchangeable. A researcher investigating why particular survey respondents obtained certain scores may need linked individual records. A researcher developing a questionnaire from teacher interviews may appropriately evaluate it using an independent quantitative sample.

The broader question of whether the same participants should be used in both phases should be resolved before deciding how their data will be connected.

How Does Connecting Work in Explanatory Sequential Research?

Explanatory sequential research commonly begins with quantitative analysis and uses selected results to guide qualitative follow-up.

For example, a study may identify students with high recorded online participation but relatively low assessment performance. Researchers could then select interview participants representing this profile and relevant contrasting profiles.

Ivankova, Creswell, and Stick (2006) illustrate how participant selection and qualitative follow-up can establish meaningful connections between sequential phases.

The connection should involve more than inviting available survey respondents. Researchers need to specify which quantitative findings warrant investigation, how the sampling criteria represent those findings, and why selected participants are likely to provide relevant evidence.

Purposive selection can use typical cases, contrasting profiles, extreme values, or cases that differ from model-based expectations. The choice should reflect the explanatory purpose rather than an automatic preference for high and low scores.

Further consideration of using quantitative results to select interview participants is important when measurement uncertainty or subgroup composition affects the selection.

How Does Sampling Work in Exploratory Sequential Research?

Exploratory sequential research commonly begins with qualitative inquiry that informs subsequent quantitative development or investigation.

For example, interviews with teachers may identify relevant dimensions of responsible generative AI use. The findings could inform a questionnaire administered to a different group of teachers.

The relationship between samples is developmental rather than necessarily individual.

The qualitative sample should provide appropriate evidence for the intended conceptual development. The quantitative sample should support the measurement evaluation or statistical questions of the subsequent phase.

Researchers should examine whether the qualitative participants' contexts are sufficiently relevant to the intended quantitative population. A questionnaire developed exclusively from experiences in one specialized institution may omit practices important elsewhere.

However, a separate quantitative sample does not automatically establish generalizability. The selection of that sample and the evidence obtained from it remain consequential.

What Is the Difference Between Linking Participants and Linking Findings?

Participant linkage allows researchers to associate information from different phases with the same individual or case. Findings integration involves developing interpretations from the relationship between quantitative and qualitative evidence.

These are related but distinct activities.

For example, a researcher may assign each consenting survey respondent a secure study identifier. The identifier allows the researcher to associate a subsequent interview with that respondent's quantitative profile.

That linkage makes case-level comparison possible. It does not automatically establish that the interview explains the numerical result.

The researcher must still analyze the qualitative evidence and examine its relationship with the original quantitative findings.

Similarly, a study using different samples may integrate findings at the classroom or institutional level without linking individual records.

The appropriate level of integration should be established before interpreting the combined evidence.

How Can Researchers Plan Individual-Level Linkage?

When individual-level linkage is necessary, the research protocol should establish how participants can be identified across phases without unnecessarily exposing their identities.

One approach is to assign pseudonymous study identifiers and maintain any necessary identity-to-identifier mapping separately from the analytical datasets.

The research team should specify who can access the mapping, how follow-up invitations will be managed, where linkage information will be stored, and when identifiable information will be deleted or otherwise handled according to the approved retention plan.

Participants should be informed accurately about whether their responses can be linked and whether they may be contacted for follow-up.

These procedures should be reviewed under applicable institutional ethics requirements and data protection obligations.

For sensitive data, including academic performance or identifiable subgroup characteristics, researchers should minimize the information exposed during recruitment and analysis.

What If the First-Phase Survey Was Anonymous?

Genuinely anonymous survey responses ordinarily cannot be linked to identifiable respondents after collection because the necessary connection was never retained.

Researchers may recruit volunteers for interviews through a separate contact form, but this does not automatically allow their interview accounts to be matched with their original responses.

For example, a survey may display a separate invitation to provide contact details for an interview. If no linkage token or equivalent mechanism connects the two submissions, the researcher cannot reliably determine which survey response belongs to a volunteer.

Researchers should not attempt to reconstruct identities from combinations of demographic or behavioral characteristics when doing so would violate participants' expectations or the approved research protocol.

If individual-level integration is essential, the linkage procedure should be planned before data collection and described accurately during consent.

How Do You Build a Sampling Matrix From Quantitative Results?

A sampling matrix can translate quantitative findings into transparent recruitment categories.

Suppose a study investigates the relationship between learning platform participation and academic performance. The researcher may identify participants across combinations of relatively high and low participation and performance.

The categories should be substantively justified. Arbitrary median splits can discard information, so researchers should consider the original continuous values, measurement uncertainty, and the intended comparison.

The matrix might identify relevant profiles, eligible participants, recruitment priorities, and replacement procedures.

Palinkas and colleagues (2015) discuss purposeful sampling strategies in mixed-method implementation research, providing a useful basis for selecting information-rich cases.

The sampling matrix should remain a tool for recruitment rather than a substitute for evaluating the meaning and quality of the quantitative findings.

How Do You Decide How Many Participants to Connect?

There is no universal percentage of first-phase participants who must participate in the second phase.

For explanatory follow-up, qualitative sample adequacy depends on the question, relevant variation, analytical approach, and depth of information required.

Malterud, Siersma, and Guassora (2016) propose information power as one approach to considering qualitative sample adequacy. Its relevance depends on the specificity of the sample, study aim, theoretical background, dialogue quality, and analytical strategy.

Quantitative sample requirements should be justified separately according to the intended statistical analyses and inferential purpose.

Matching sample sizes is not a methodological requirement. The central issue is whether each component has sufficient evidence to support its contribution and whether their sampling relationship permits the planned integration.

What Happens When Participants Decline Follow-Up?

Sequential studies may encounter recruitment losses between phases. Participants with particular quantitative profiles may decline interviews, become unavailable, or withdraw permission for further contact.

These losses can affect which experiences are represented in the qualitative component.

Researchers should establish recruitment alternatives that preserve the intended sampling rationale. For example, if a selected participant declines, another eligible participant with a comparable profile may be invited where available and appropriate.

However, a replacement should not be described as equivalent when important characteristics differ.

Researchers should document invitation numbers, participation, refusals where known, and the characteristics relevant to the sampling plan without compromising confidentiality.

When possible, they may examine whether participants who completed follow-up differ on relevant measured characteristics from those who did not. Such comparisons can identify potential selection concerns but cannot eliminate every source of nonresponse bias.

How Does Attrition Affect Integrated Conclusions?

Attrition may narrow the range of quantitative profiles represented in the qualitative sample.

For example, if students with low engagement are less likely to accept interview invitations, the qualitative findings may disproportionately represent more engaged students.

This does not necessarily invalidate the qualitative analysis. It does, however, limit what can be concluded about experiences associated with the missing profiles.

Researchers should distinguish the original quantitative sample from the subset that contributed qualitative evidence.

The final report should explain how recruitment outcomes affected the planned connections and whether any intended comparisons could not be completed.

Can Different Stakeholder Samples Be Connected?

Yes, when the research question supports integration across stakeholder groups or analytical levels.

For example, quantitative student outcomes may be interpreted alongside interviews with teachers responsible for implementing an educational intervention.

The connection may operate at the course or institutional level rather than the individual student level.

However, teachers' accounts cannot automatically explain why particular students obtained particular scores. The researcher must establish which contextual relationships are supported by the evidence.

Care is needed to avoid ecological or individualistic inference errors when findings concern different levels of analysis.

How Should Connected Samples Be Represented in the Analysis?

A joint display can make the relationship between quantitative profiles and qualitative findings visible.

For example, rows might represent selected participant profiles, with columns presenting quantitative characteristics, relevant qualitative themes, and integrated interpretations.

Guetterman, Fetters, and Creswell (2015) discuss joint displays as a means of facilitating mixed-methods integration.

Researchers should avoid including identifiable information in published displays. Small or unusual combinations of characteristics may require aggregation or suppression to protect confidentiality.

The display should support an analytical comparison, not merely place two datasets beside each other.

How Should the Sampling Connection Be Reported?

The methodology should describe the relationship between phase-specific samples, the criteria used for selecting second-phase participants, and any linkage procedures.

For explanatory sequential studies, researchers should state which quantitative findings guided recruitment and how the resulting qualitative sample represented relevant profiles.

For exploratory sequential studies, they should explain how the qualitative sample informed development and why the subsequent quantitative population was appropriate.

Recruitment losses, deviations from planned selection, and limitations of linkage should be disclosed.

The final interpretation should distinguish what is supported at the individual level from what can only be inferred at a broader contextual level.

Connecting Student Survey Profiles With Follow-Up Interviews

Suppose the interviews suggest that students use the platform for different purposes and that some undertake substantial learning activities outside the recorded environment.

The integrated findings might indicate that platform activity measures do not fully represent meaningful engagement.

Because the interviews are linked to individual quantitative profiles, the researcher can examine the correspondence between particular students' recorded activity and their accounts.

Nevertheless, the purposive qualitative sample does not support population prevalence estimates, and the interviews do not establish that particular learning practices caused the observed performance differences.

Common Mistakes When Connecting Mixed-Methods Samples

How to Plan the Sampling Connection Before Data Collection

Begin by identifying whether the research question requires individual-level explanation, development of a subsequent instrument, or integration across groups or contexts.

Then choose the sampling relationship that supports that purpose and establish any necessary recruitment and linkage procedures before the first phase begins.

Keep a clear record of sampling decisions, recruitment outcomes, and the relationship between the final phase-specific samples.

The strongest sampling connection is not necessarily the one with the greatest participant overlap. It is the one that supports the intended integrated inference without overstating what the evidence can establish.

Check Whether Your Samples Can Be Connected Defensibly

Frequently Asked Questions About Connecting Samples Across Phases

Connect Samples in the Way Your Intended Inference Requires

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.