Why using both qualitative and quantitative data is not enough if the study never truly integrates them

Opening: having two strands is not the same as having one study

A study does not become mixed methods merely because it contains both numbers and narratives. Researchers often assume that if a project includes a survey and some interviews, or a dataset and a few focus groups, the design is automatically stronger and more comprehensive. But mixed methods is not defined by coexistence alone. It is defined by integration, the deliberate bringing together of qualitative and quantitative components so that their combination yields insight that neither strand would provide on its own. Fetters, Curry, and Creswell (2013) make this point explicitly, arguing that integration is the core of mixed methods design at the levels of design, methods, and interpretation.

That is why this mistake matters so much. When mixed methods are “mixed in name only,” the study often looks ambitious but remains internally split. One strand answers one question, the other answers something adjacent, and the final discussion simply places the results side by side without showing what was learned through their combination. Guetterman, Fetters, and Creswell (2015) note that meaningful integration remains elusive in practice and that joint displays can help researchers move from parallel reporting to genuinely integrated analysis.

Why researchers commonly make this mistake

Researchers usually do not make this mistake because they reject mixed methods principles. More often, they make it because mixed methods carries a strong promise: breadth plus depth, numbers plus meaning, measurement plus context. That promise is attractive across fields such as education, marketing, health, and agriculture. But it also creates a temptation to add a second strand late in the design process simply to make the study look more complete. Instead of asking, “Why do these two forms of evidence need each other here?” researchers ask, “How can I include both?” That is a very different design logic. As Creswell and Plano Clark’s mixed-methods design framework has long emphasized, the rationale for mixing must be explicit and purposeful, not decorative.

A second reason is training. Many researchers are educated primarily in one tradition and only later add elements from the other. A quantitatively trained researcher may add interviews to “illustrate” survey findings without planning how the qualitative material will shape analysis. A qualitatively trained researcher may add a small questionnaire to provide “supporting data” without a clear role for those numbers in the argument. Hesse-Biber (2010) argues that methodological assumptions shape how mixed methods is practiced, often limiting integration when researchers remain anchored in one primary tradition.

A third reason is practical workflow. It is relatively easy to collect two datasets separately. It is much harder to design the study so that one strand informs sampling, instrument development, analysis, interpretation, or inference in the other. Fetters et al. (2013) describe several integration approaches, connecting, building, merging, and embedding, and their article makes clear that integration requires planned links, not just the accumulation of multiple data sources.

Dominant design context: mixed methods

This mistake belongs most clearly to mixed methods research, because integration is the defining characteristic of the approach. Recent mixed-methods literature continues to stress that integration is not optional or secondary. The dominant logic here is M > RQ > D. Methodology ranks first because the main failure is architectural: the study is labeled mixed methods, but the design does not specify how the strands relate. Research question comes second because mixed methods questions should either require both strands or require one strand to elaborate, explain, build, or test findings from the other. If the question does not require integration, the study easily becomes a pair of parallel mini-studies. Data come third because, once integration is weak at the design stage, the resulting qualitative and quantitative data often remain disconnected throughout collection, analysis, and interpretation. Fetters et al. (2013) and Guetterman et al. (2015) both support this logic by showing that integration must be designed into the study rather than added at the end.

Where the failure occurs in the RQ–RH–D–M chain

At the RQ level, the problem often begins with a question that sounds broad enough to justify mixed methods but is not actually structured to require it. For example, a study may ask whether a health intervention is effective and then add interviews without clarifying whether they are meant to explain outcome variation, improve implementation understanding, develop measures, or capture patient experience as a distinct analytic dimension. Without that logic, the second strand becomes an add-on rather than a necessary component. Creswell and Plano Clark’s design typology, convergent, explanatory sequential, exploratory sequential, and more advanced forms, exists precisely to prevent this vagueness.

At the M level, the failure becomes decisive. The study may describe itself as convergent, explanatory sequential, or exploratory sequential, but the actual procedures do not connect the strands. Sampling may not be linked. One dataset may not build the instrument for the other. Results may be analyzed separately and only compared superficially in the discussion. Fetters et al. (2013) identify connecting, building, merging, and embedding as concrete ways to integrate, which means that a study lacking such links is mixed in label more than in design.

At the D level, the consequences become visible in the evidence itself. The quantitative data may show patterns, and the qualitative data may provide narratives, but neither is positioned to transform or deepen the interpretation of the other. Guetterman et al. (2015) show that joint displays can make these linkages visible by bringing the strands together in one analytic representation. When no such integration occurs, the study often produces two valid but separate outputs rather than one genuinely mixed-methods inference.

There may also be an RH issue in some mixed-methods studies, especially when hypotheses guide the quantitative strand. But in this mistake, RH is usually secondary. The main failure is not hypothesis formulation; it is the failure to build a design in which the strands genuinely need and inform one another.

How this mistake distorts findings and conclusions

When mixed methods are mixed in name only, the study often produces a false sense of completeness. The reader sees both statistics and quotations and assumes the findings have been triangulated, deepened, or cross-validated. But unless the strands were meaningfully integrated, that assumption is unwarranted. The project may simply offer two descriptions of related issues, not one coherent mixed-methods answer.

In Education, a researcher may survey teachers about digital learning adoption and also conduct interviews with a small group of teachers, but if the interviews are not used to explain surprising survey patterns, refine constructs, or challenge quantitative assumptions, they remain illustrative rather than integrative.

In Marketing, a study may analyze consumer preference data and collect focus groups, but if the focus groups are not linked to segment patterns, measure development, or interpretation of contradictory results, the design remains parallel rather than mixed.

In Health/Wellbeing, a study may report clinical outcomes and patient interviews in the same paper, yet never show how patient experience helps explain outcomes, implementation, adherence, or subgroup variation. Fetters et al. (2013) describe “fit” as the extent to which qualitative and quantitative findings cohere, and that notion is especially useful here: mixed-methods studies often fail because they never establish or test that fit.

The damage is therefore not just procedural. It is inferential. The study invites readers to believe that stronger or more nuanced conclusions are warranted because two forms of evidence are present. But presence is not the same as integration.

How to avoid the mistake before collecting data

The strongest prevention is to decide why the study needs mixed methods before deciding how to collect both kinds of data. A mixed-methods design should begin with a rationale such as explanation, exploration, development, comparison, expansion, or implementation insight. If the researcher cannot say what the second strand is needed for, the design may not require mixed methods at all. The literature on mixed methods design has consistently emphasized this rationale-first logic.

A second preventive step is to plan integration at multiple stages. Fetters et al. (2013) show that integration can occur through connecting, building, merging, or embedding. That means the design should specify where the strands will meet: in sampling, instrument development, analytic comparison, data transformation, interpretation, or visual integration such as joint displays. Guetterman et al. (2015) further show that joint displays are not just presentation devices; they are analytic tools that help generate new insights from combined evidence.

A third preventive step is to write the mixed methods question explicitly. The study should not only have a quantitative question and a qualitative question; it should also have a mixed-methods question or at least a clearly stated integrative purpose. Recent mixed-methods scholarship continues to stress that explicit integration planning is essential rather than optional.

What can still be repaired after data collection

After data collection, some repair is possible, but only up to a point. If both strands are already collected, researchers may still improve integration at the analysis and interpretation stage. They can use joint displays, compare convergence and divergence explicitly, examine how one strand qualifies or reframes the other, or generate meta-inferences that neither dataset alone would justify. Guetterman et al. (2015) and Fetters et al. (2022) both point to the value of joint displays for making integrated reasoning more transparent.

However, not everything can be repaired. If the datasets were collected for unrelated purposes, if sampling was not linked where it needed to be linked, or if the qualitative and quantitative strands are effectively answering different questions, then post hoc integration may remain superficial. In such cases, the most honest solution may be to reframe the paper as a multimethod or parallel design rather than a truly integrated mixed-methods study. That is a partial salvage, not a full rescue.

Brief cross-field illustrations

In Education, a study may use survey data to estimate student engagement and interviews to ask teachers about classroom motivation. Both datasets may be individually useful, but unless the design links them through a shared construct, subgroup logic, or explanatory sequence, they do not yet form an integrated mixed-methods study.

In Marketing, a project may combine purchase intention scores with focus groups about brand image. If the focus groups are not used to explain segment differences, inform measure construction, or illuminate contradictions in the quantitative findings, the study is not truly mixed, it is doubled.

In Agriculture, a study may report crop-yield outcomes and farmer narratives about climate adaptation. Those two sources can be powerful together, but only if the design clarifies whether narratives are explaining variation in outcomes, helping define adaptation practices, or reframing what “success” means from the farmer’s perspective. Otherwise, the strands simply coexist.

Across all three examples, the lesson is the same: mixed methods is not about having two voices in the same article. It is about making those voices answer each other.

Short takeaway checklist

Before collecting data, ask:

• Why does this study truly need both qualitative and quantitative evidence?
• Where, exactly, will the two strands be connected, built, merged, or embedded?
• Am I planning two separate components, or one integrated design?
• If I removed one strand, what would the study lose?
• Can I explain in plain language what new insight will emerge only from integration?

A good mixed-methods study does not merely contain two kinds of data. It produces a stronger inference because those data have been deliberately brought together.

References

Creswell, J. W., & Plano Clark, V. L. (2018). Designing and conducting mixed methods research (3rd ed.). SAGE.

Fetters, M. D., Curry, L. A., & Creswell, J. W. (2013). Achieving integration in mixed methods designs—Principles and practices. Health Services Research, 48(6 Pt 2), 2134–2156. https://doi.org/10.1111/1475-6773.12117

Fetters, M. D., Creswell, J. W., & Morris, M. A. (2022). Joint displays of integrated data collection in mixed methods research. International Journal of Qualitative Methods, 21. https://doi.org/10.1177/16094069221104564

Guetterman, T. C., Fetters, M. D., & Creswell, J. W. (2015). Integrating quantitative and qualitative results in health science mixed methods research through joint displays. The Annals of Family Medicine, 13(6), 554–561. https://doi.org/10.1370/afm.1865

Plano Clark, V. L. (2010). Applying three strategies for integrating quantitative and qualitative approaches in a mixed methods study. Field Methods, 22(3), 274–287. https://doi.org/10.1177/1525822X09357174

O’Cathain, A. (2019). Mixed methods research: The issues beyond combining methods. Journal of Advanced Nursing, 75(3), 499–501. https://doi.org/10.1111/jan.13877

Hesse-Biber, S. (2010). Qualitative approaches to mixed methods practice. Qualitative Inquiry, 16(6), 455–468. https://doi.org/10.1177/1077800410364611