03 · What You Need to Know
How Sampling Can Connect Quantitative and Qualitative Evidence
What Is Sampling Integration in Mixed-Methods Research?
Sampling integration occurs when the selection of participants, cases, settings, or other units in one research strand is intentionally related to information or sampling decisions in another strand.
The relationship may be established through participant-level linkage, selection based on quantitative findings, coordinated recruitment from a shared population, or other purposeful arrangements.
For example, a researcher may administer a student engagement questionnaire and then select interview participants according to their engagement profiles.
Alternatively, a study may recruit students for quantitative surveys and teachers for qualitative interviews to examine complementary perspectives on an institutional intervention.
Both arrangements can support mixed-methods inquiry, but they offer different opportunities for integration.
The central question is what relationship between the samples is necessary to answer the research question.
How Is Sampling Integration Different From Simply Using Two Samples?
Collecting quantitative data from one group and qualitative data from another does not automatically establish sampling integration.
The researcher must explain why the samples are related and how their relationship contributes to the mixed-methods design.
Parallel sampling
Quantitative and qualitative samples are selected for their respective purposes without one strand directly determining selection in the other.
Connected sampling
Information from one strand informs selection of participants or cases in another, establishing an explicit relationship between the samples.
Parallel sampling may be entirely appropriate when the strands address complementary dimensions of a phenomenon. Connected sampling becomes particularly valuable when the research seeks to explain specific quantitative patterns through qualitative follow-up.
When Is Sampling Integration Especially Important?
Sampling integration is particularly consequential when the research question requires explanation of particular results, examination of contrasting cases, or participant-level comparison.
For example, a researcher may want to understand why students with similar access to online learning resources report different levels of engagement.
Recruiting interview participants based on relevant survey profiles can establish a direct connection between the quantitative patterns and the qualitative accounts used to investigate them.
Without that connection, independently recruited interview participants may provide useful contextual insights, but their accounts cannot automatically be treated as explanations of the specific survey respondents' experiences.
The sampling decision should therefore follow the intended integration purpose rather than the convenience of recruitment.
What Sampling Arrangements Are Possible?
Mixed-methods studies may use several relationships between quantitative and qualitative samples.
| Sampling relationship |
Description |
Potential use |
| Identical samples |
The same participants contribute quantitative and qualitative evidence. |
Examine participant-level relationships between measured characteristics and personal accounts. |
| Nested samples |
A qualitative sample is selected from a larger quantitative sample, or another clearly defined subset relationship is established. |
Investigate selected quantitative profiles through detailed qualitative follow-up. |
| Overlapping samples |
Some participants contribute to both strands, while others participate in only one. |
Combine linked case-level evidence with broader or complementary perspectives. |
| Distinct but related samples |
Different participants or groups are recruited from related contexts or populations. |
Examine complementary perspectives, such as student survey responses and teacher interviews. |
| Sequentially connected samples |
Results from an earlier strand inform subsequent selection criteria. |
Select contrasting, unusual, or otherwise analytically informative cases. |
These arrangements are not mutually exclusive. A nested qualitative sample may also be sequentially connected because participants are selected after quantitative analysis.
The researcher should describe the actual relationship rather than assume that a design label fully explains the sampling procedures.
How Does Sampling Integration Work in Explanatory Sequential Research?
Explanatory sequential designs ordinarily begin with quantitative data collection and analysis, followed by qualitative inquiry intended to investigate selected quantitative findings.
Sampling integration often occurs when the quantitative results guide the selection of qualitative participants.
For example, researchers may identify students whose engagement scores are unexpectedly high or low relative to their reported institutional support.
They may then invite eligible participants from these profiles for interviews exploring their experiences.
This arrangement can provide qualitative evidence directly related to the quantitative cases of interest.
However, the interviews should not be designed merely to confirm the researcher's preferred explanation. Participants may reveal contextual factors or interpretations that challenge the initial quantitative assumptions.
What Is Extreme or Deviant Case Sampling?
Extreme or deviant case sampling involves deliberately selecting cases that exhibit unusual, unusually high, unusually low, or otherwise distinctive characteristics relevant to the research question.
In mixed-methods research, quantitative results may help identify such cases.
For example, a researcher studying online student engagement might identify students with very high engagement scores despite reporting limited institutional support.
Interviewing selected students could help explore experiences that are not readily explained by the broader quantitative pattern.
However, unusual cases should not automatically be treated as representative of the wider population.
Their value lies in examining the circumstances surrounding distinctive observations, potentially revealing limitations or alternative explanations relevant to the original model.
What Is Maximum Variation Sampling in Mixed-Methods Research?
Maximum variation sampling seeks cases representing a purposeful range of characteristics or experiences relevant to the study.
Quantitative information can help identify participants across dimensions such as engagement levels, institutional contexts, or technology use patterns.
For example, a researcher might select students from different combinations of engagement and support scores rather than interviewing only those at the highest and lowest ends.
This may help illuminate how experiences differ across contrasting profiles.
However, maximum variation sampling is not equivalent to statistical representativeness. The qualitative sample is selected to support analytic understanding rather than to reproduce the population distribution.
Can Qualitative Findings Inform Quantitative Sampling?
Yes. Sampling integration is not limited to quantitative-first designs.
In exploratory sequential research, qualitative findings may identify participant characteristics, contextual conditions, or subgroups that should be represented in a subsequent quantitative phase.
For example, interviews may reveal that online students experience academic belonging differently depending on their opportunities for peer interaction.
Researchers may use these findings to define relevant sampling strata or ensure that the quantitative study includes students from contrasting learning arrangements.
However, qualitative findings alone do not establish the population prevalence of the identified experiences.
Subsequent quantitative sampling must still be appropriate to the intended statistical claims.
Does Convergent Mixed-Methods Research Require Identical Samples?
No. Convergent designs commonly collect quantitative and qualitative evidence during a broadly similar period and integrate the findings afterward.
Identical samples may be useful when the study requires participant-level comparisons. However, distinct samples may be appropriate when the strands examine related dimensions of a phenomenon.
For example, a university might survey students about online learning support and interview teachers about how that support is provided.
The strands address related institutional practices but represent different perspectives.
Integration can be meaningful if the researcher explains how the findings relate and avoids attributing teachers' accounts to individual surveyed students.
The sampling relationship should support the intended level of interpretation.
How Does Sampling Integration Affect the Unit of Analysis?
The unit of analysis is the entity about which the research intends to draw conclusions. It may be an individual, classroom, institution, program, or another defined unit.
Sampling integration should preserve clarity about this level.
Suppose quantitative data describe average engagement across universities, while qualitative interviews examine experiences of selected students.
The researcher may use interviews to contextualize institutional patterns, but individual accounts cannot automatically establish why an institution has a particular average score.
Likewise, institutional-level associations should not be interpreted as though they necessarily apply to every student.
These problems concern cross-level inference and can arise when researchers integrate evidence without considering the units represented by each strand.
Should the Qualitative Sample Be Statistically Representative?
Not necessarily. Qualitative sampling often prioritizes relevance, depth, variation, and the opportunity to examine particular experiences or processes.
In an explanatory sequential design, participants may be selected because their quantitative profiles are especially informative, not because they reproduce the statistical distribution of the larger sample.
However, researchers should avoid making population-level prevalence claims from a purposively selected qualitative subset.
Statistical representativeness and qualitative information richness serve different purposes.
The integration strategy should recognize these differences rather than assuming that one sampling logic can replace the other.
How Many Participants Should Be Included in Both Strands?
There is no universal numerical requirement for overlap between quantitative and qualitative samples.
The appropriate number depends on the research question, design, sampling relationship, intended integration, and practical constraints.
A qualitative follow-up sample may be relatively small when the purpose is detailed examination of carefully selected cases. Other designs may require broader coverage of contrasting profiles or contexts.
Researchers should justify sample adequacy separately for each strand while also explaining whether the relationship between samples is sufficient for the intended integration.
A large quantitative sample does not automatically compensate for an inadequate qualitative sample, and extensive qualitative interviews do not make a nonprobability quantitative sample statistically representative.
What Is the Difference Between Sampling Integration and Analytic Integration?
Sampling integration establishes relationships through participant or case selection. Analytic integration examines relationships between evidence or findings.
A study may use quantitative scores to select interview participants, thereby integrating during sampling.
However, the researcher must still determine how the interview findings will be related to the quantitative results.
For example, the qualitative accounts may be organized by quantitative profile and compared using a joint display.
The sampling connection creates an opportunity for integration, but it does not automatically produce an integrated interpretation.
The subsequent procedures for integrating evidence during analysis should be planned according to the research question.
When Should Sampling Integration Be Planned?
Sampling relationships that depend on participant linkage, recontact, or access to particular cases should be considered before the relevant data collection begins.
For example, if survey respondents may be invited to qualitative interviews based on their responses, researchers need a recruitment arrangement that permits this connection.
An irreversibly anonymous survey cannot later be linked to identifiable respondents unless a separate, appropriately designed mechanism was established.
Researchers may use voluntary recontact forms or other privacy-preserving procedures, but these arrangements should be consistent with informed consent and institutional requirements.
This illustrates why planning mixed-methods integration before data collection can be necessary even when the actual selection criteria will be finalized after initial analysis.
What Ethical Issues Arise When Linking Samples?
Participant-level linkage can increase privacy risks because combining records may reveal information that is not identifiable within either dataset alone.
Researchers should establish how identifiers will be generated, stored, separated from substantive data, and accessed.
Consent procedures should accurately explain any intended follow-up contact or linkage, subject to applicable institutional and legal requirements.
Participants should not be led to believe that their responses are irreversibly anonymous when the study retains a mechanism for identifying them.
Researchers should also consider whether selecting participants based on sensitive responses could expose information about their circumstances.
Where necessary, ethics review or amendments should be obtained before changing recruitment or linkage procedures.
What If Selected Participants Decline Qualitative Follow-Up?
Nonparticipation is a practical and methodological possibility in connected sampling.
Participants selected because of unusual or important quantitative profiles may decline interviews, be unreachable, or withdraw consent.
This can limit the range of cases represented in the qualitative follow-up.
Researchers should consider whether alternative participants with relevant profiles can be recruited and document how nonresponse affects the intended comparisons.
However, replacement participants should not be treated as interchangeable merely because they share one quantitative characteristic.
Other contextual differences may matter, and the final report should acknowledge relevant limitations.
Can Sampling Integration Introduce Bias?
Yes. Selecting qualitative participants from a quantitative dataset may introduce selection effects, particularly when only certain respondents agree to follow-up.
For example, students with strong opinions about institutional support may be more willing to participate in interviews than those with less distinctive experiences.
Similarly, selecting only extreme quantitative cases may overemphasize unusual circumstances.
These choices may be appropriate for the research question, but their implications should be recognized.
Researchers should distinguish purposeful selection from claims that the resulting qualitative accounts represent all members of the quantitative population.
How Can a Joint Display Support Sampling Integration?
A joint display can organize quantitative profiles alongside qualitative findings from the corresponding participants or cases.
For example, rows may represent engagement profiles, while columns contain quantitative indicators, interview themes, and integrated interpretations.
This structure helps researchers examine whether qualitative accounts illuminate the characteristics used to select participants.
However, a joint display should not imply participant-level linkage where none exists.
Researchers should establish whether each row represents an individual, group, or broader analytic category and ensure that the evidence is compared at an appropriate level.
The design of a joint display for mixed-methods research should reflect the actual sampling relationship.
How Should Sampling Integration Be Reported?
The methods section should explain the sampling procedures for each strand and the relationship between them.
Researchers should identify whether participants were selected from an earlier sample, recruited independently, or chosen using findings from another strand.
When quantitative results informed qualitative selection, report which characteristics or patterns were used and why they were analytically relevant.
The report should also explain recruitment limitations, participant overlap, and any differences between intended and achieved samples.
In the integrated interpretation, researchers should distinguish conclusions supported by linked cases from broader contextual comparisons involving distinct samples.
Watch Out
Do not assume that interviews with independently recruited participants explain the experiences of specific survey respondents. Without an appropriate sampling or evidentiary connection, qualitative accounts may contextualize quantitative patterns but cannot automatically be attributed to the individuals represented in those patterns.
