Embedded Multiphase Mixed Methods Design
Also known as: embedded multi-phase mixed methods, nested multiphase design, multiphase embedded MMR, embedded phased mixed design
Embedded multiphase mixed methods is a research design in which a secondary data strand (qualitative or quantitative) is nested within a primary, dominant strand across two or more sequential study phases. Each phase builds on the prior one, while the embedded strand enhances understanding of specific sub-questions that the dominant strand alone cannot answer. This design is suited to complex, longitudinal, or program-evaluation research problems requiring sustained inquiry across stages.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+1 more
When to use it
Use this design when a complex problem requires staged investigation across time — for example, developing and testing an intervention, or evaluating a program through formative, implementation, and summative phases. It is appropriate when one data type dominates but targeted sub-questions within each phase benefit from a secondary strand. Do not use it when resources are limited: the design demands sustained team capacity across phases. Avoid it when a single embedded or a simpler sequential design would answer the research questions — multiphase structures add coordination overhead that must be justified by the complexity of the inquiry.
Strengths & limitations
- Matches the structure of real-world program evaluation and intervention research, which naturally unfolds in phases.
- The embedded strand adds depth to the dominant strand without requiring equal resources for both data types.
- Iterative phase-by-phase learning improves the quality of each subsequent phase's design.
- Produces cumulative, multi-level evidence that is more convincing to policymakers and funders than single-phase studies.
- Flexible regarding which strand is dominant and which is embedded — the configuration can shift between phases if needed.
- Highly resource-intensive: multiple phases with two concurrent strands demand significant time, funding, and research team expertise.
- Integration across phases is complex; each phase generates its own integration challenge, and the cross-phase meta-inference adds a further layer of analytical difficulty.
- The embedded strand, by design, occupies a secondary role — questions requiring equal-weight qualitative and quantitative inference are better served by a concurrent triangulation design.
- Long study timelines increase the risk of participant attrition, contextual shifts, or changes in research team personnel that threaten coherence.
Frequently asked
How is this different from a simple multiphase mixed methods design?
In a standard multiphase design, each phase may use either QUAN or QUAL data, and phases are connected sequentially but each strand stands on its own. In the embedded multiphase variant, each phase contains a primary dominant strand with a secondary strand explicitly nested within it — so there are two layers of structure: the phase-to-phase sequence and the within-phase embedding. This adds analytical depth within each phase but also increases coordination complexity.
Does the dominant strand have to stay the same across all phases?
No. The dominant strand can shift between phases if the research logic requires it — for example, a qualitative-dominant formative phase followed by a quantitative-dominant efficacy phase. What must remain consistent is the explicit specification of which strand is dominant in each phase and the rationale for the embedded strand's role.
How do I report integration in a multiphase embedded study?
Report integration at two levels: first, within each phase, describe how the dominant and embedded strand findings were brought together (e.g., a joint display or narrative summary). Second, at the end of the paper or report, present a cross-phase meta-inference that synthesizes what was learned across all phases and explains how earlier phases informed later ones. The meta-inference is the most novel and important contribution of this design.
Can this design be used in a single-researcher dissertation?
It is possible but demanding. A two-phase embedded design with modest data collection in each phase is feasible for a doctoral dissertation given sufficient time. Three or more phases typically require a research team and external funding. A single-researcher dissertation should carefully scope the number of phases and the size of the embedded strand, and pilot the coordination demands before committing.
What software supports analysis in this design?
No single software manages the full design, but a combination of tools is common: statistical packages (R, SPSS, Stata) for the quantitative dominant strand, qualitative software (NVivo, ATLAS.ti, MAXQDA) for qualitative data, and spreadsheet or matrix tools for joint-display integration. MAXQDA has native mixed methods integration features that are particularly useful for within-phase and cross-phase joint displays.
Sources
- Creswell, J. W., & Plano Clark, V. L. (2018). Designing and Conducting Mixed Methods Research (3rd ed.). SAGE Publications. ISBN: 9781483344379
- Tashakkori, A., & Teddlie, C. (Eds.). (2010). SAGE Handbook of Mixed Methods in Social and Behavioral Research (2nd ed.). SAGE Publications. ISBN: 9781412972666
How to cite this page
ScholarGate. (2026, June 3). Embedded Multiphase Mixed Methods Design. ScholarGate. https://scholargate.app/en/research-design/embedded-multiphase-mixed-methods
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Concurrent Embedded Mixed Methods DesignResearch Design↔ compare
- Explanatory Sequential Mixed Methods DesignResearch Design↔ compare
- Exploratory Sequential Mixed Methods DesignResearch Design↔ compare
- Multiphase Mixed Methods DesignResearch Design↔ compare