Embedded Mixed Methods Research Design

The embedded research design in mixed methods research is when the researcher collects and analyzes quantitative and qualitative data within a traditional mixed methods research design. The second data collection may occur at any point of the primary data collection based on the selected design. In this article, we will go through the purpose, selection and how to do an embedded research design.

Introduction

The embedded mixed methods research design integrates qualitative data into predominantly quantitative studies (or vice versa) to answer secondary research questions or provide deeper insights into the primary study. Greene, Caracelli, and Graham (1989), along with Creswell and Plano Clark, played significant roles in shaping the embedded design. They emphasized embedding different types of data to enhance recruitment, understand intervention processes, and explore participant experiences.

Researchers widely adopted the embedded design as one of the core mixed methods frameworks due to its flexibility and ability to meet specific research needs. This design has become particularly prominent in fields like health and social sciences.

The embedded design in mixed methods research integrates qualitative and quantitative data within a primary research framework, allowing researchers to address supplementary research questions. This approach involves embedding a secondary dataset—qualitative within a quantitative framework or vice versa—at any stage of the research process, whether before, during, or after the primary data collection and analysis. In some cases, the secondary data plays a supporting role, such as incorporating qualitative insights into a quantitative experiment to examine specific aspects like intervention processes or participant responses.

The embedded design is particularly useful when a single dataset is inadequate to address all research questions, as it enables researchers to answer diverse questions requiring different types of data. For instance, qualitative data might be embedded in a predominantly quantitative study to improve recruitment strategies or explain participant reactions. The objectives for integrating the secondary dataset often differ from the study's primary purpose, such as assessing the effectiveness of a treatment.

This design differs from convergent designs, where qualitative and quantitative methods are used to answer a unified research question. Instead, the embedded approach complements the primary orientation of the study, making it suitable for researchers with limited resources or less experience in the secondary method. Researchers must also ensure the chosen design aligns with their study's main objectives and consider factors like expertise and resource allocation.

Researchers choose this mixed methods research design when they have the expertise to implement the design rigorously and are comfortable with having the study be driven by a primary orientation. It is also used when the researcher doesn't have prior experience with the supplemental method and doesn't have the resources to place equal priority on both types of data.

The additional data set is secondary and serves as the main methodology. It is not an equal partner but rather contributes supplementary insights, such as contextualizing findings in a quantitative study or exploring specific elements in a qualitative study.

Philosophical assumptions in embedded research design

The philosophical assumptions behind the embedded design in mixed methods research are grounded in the paradigmatic foundation of the primary methodological approach being used. This design aims to enhance the application of a traditional quantitative or qualitative design by incorporating an additional data set that supports the main approach.

The embedded design operates under the philosophical assumptions of the primary approach. For example, a quantitative primary design (e.g., experimental or longitudinal) typically aligns with postpositivist assumptions, focusing on objectivity, measurement, and causal relationships. On the other hand, a qualitative primary design (e.g., phenomenological or ethnographic) often reflects a constructivist paradigm, emphasizing subjective meanings, social contexts, and participants' perspectives.

This design acknowledges that the philosophical underpinnings of qualitative (constructivist) and quantitative (postpositivist) approaches are not inherently incompatible but can be combined strategically. For example, qualitative data may provide depth to quantitative findings, or vice versa, without undermining the coherence of the primary method.

Embedded design often reflects a pragmatic approach to research, prioritizing the research questions over strict adherence to a single paradigm. Researchers choose the secondary data set based on its ability to address specific aspects of the research problem that the primary method alone cannot.

How to do an embedded design?

The embedded design in mixed methods research involves collecting both quantitative and qualitative data, with one type serving as a supplementary component to support the primary data. This design can follow either a one-phase or two-phase approach. In the one-phase approach, both data types are collected simultaneously, but one plays a minor role, while in the two-phase approach, one data type is collected before or after the other to serve a specific purpose, such as informing the study design or explaining results. Key features of the embedded design include the supplemental role of the secondary data, which is embedded within the primary data collection process to enhance understanding rather than stand alone. Additionally, the quantitative and qualitative data address different research questions within the same study, and the supplemental data would lose its significance if it were not connected to the primary data.

1. Establishing a clear research question is essential in any mixed methods research. The research question guides decisions about data collection, analysis, and interpretation. However, embedded designs allow for flexibility, and researchers may adapt their approach as data collection progresses and new insights emerge. While there is no single formula for designing an embedded study, the following steps provide general guidelines for structuring the research process.
2. Decide which type of data will serve as the primary data. The first step in an embedded design is determining whether qualitative or quantitative data will be the primary source of evidence. This decision depends on the research question and the overall goals of the study. If the research requires an in-depth, contextual exploration of a phenomenon, qualitative data may take precedence. Conversely, if the study aims to identify patterns, measure variables, or assess trends, quantitative data may serve as the main dataset. Researchers should also consider their epistemological stance—whether they lean toward post-positivist assumptions, emphasizing measurement and objectivity, or constructivist perspectives, prioritizing meaning and interpretation.
3. Consider how the second type of data will supplement the main data. Once the primary data type is established, the next step is determining how the secondary data will complement it. The secondary data should provide additional insights that enhance or clarify the main findings. For example, in a primarily quantitative study, qualitative data can help explain unexpected results or provide deeper context to numerical trends. In a qualitative study, quantitative data can support themes that emerge by offering numerical evidence or identifying patterns across a larger sample. The secondary data should align with the study’s objectives and enhance the overall understanding of the research problem.
4. Choose a one-phase or two-phase approach to data collection. Embedded designs can follow either a one-phase or two-phase approach, depending on how data is collected. In a one-phase approach, researchers collect qualitative and quantitative data simultaneously. This is often done within a single instrument, such as a survey that includes both closed-ended and open-ended questions. In a two-phase approach, one type of data is collected first, and the second type is gathered based on the initial findings. This sequential approach is useful when qualitative insights help refine a quantitative instrument or when initial statistical patterns guide qualitative follow-up interviews. The choice between these approaches depends on feasibility, timing, and the role of the embedded data in addressing the research question.
5. Collecting and analyzing data. Data collection should align with the chosen structure of the study. Researchers typically begin by analyzing the primary data first, as it forms the foundation of the study. Once the primary data is analyzed, the secondary data is examined in relation to the main findings. For example, if the main dataset consists of survey responses, qualitative interview excerpts can be analyzed to provide deeper insight into specific trends. If the primary data is qualitative, secondary quantitative data can help identify patterns that might support or refine thematic interpretations. Throughout this process, researchers should maintain transparency in how the secondary data informs and interacts with the main dataset.
6. Interpreting results. The final step involves integrating findings from both data sources to construct a comprehensive response to the research question. Interpretation should go beyond merely presenting two sets of results and should focus on how the embedded data enhances or challenges the primary findings. Researchers should consider whether the supplementary data strengthens, refines, or adds nuance to the main conclusions. This stage also involves discussing the implications of the mixed methods approach, including how the embedded design contributed to a deeper understanding of the research problem.

By following these steps, researchers can effectively implement an embedded design that allows for a structured yet flexible approach to mixed methods research.

Design variants

According to Creswell and Plano Clark (2006), the two main variants of the embedded design are the experimental model and the correlational model, both differing in purpose and data integration strategy. The experimental model seeks to understand intervention processes or outcomes, while the correlational model aims to explain relationships between variables. Both models maintain a dominant quantitative focus, with qualitative data playing a complementary, explanatory role.

Embedded experimental model

In this model, qualitative data is embedded within a primarily quantitative experimental design, such as an experiment or quasi-experiment. The quantitative method is dominant, while the qualitative component supports and enriches the quantitative data. It can be implemented in one or two phases. In a one-phase approach, qualitative data is collected during the intervention phase to explore participant experiences or the process of the intervention alongside measuring quantitative outcomes. In a two-phase approach, qualitative data is collected either before or after the intervention. Before the intervention, it helps shape the intervention or select participants. After the intervention, it explains quantitative results or provides follow-up insights. This model is often used to understand the process behind an intervention, explain or interpret quantitative outcomes, and inform the development of measurement tools.

Embedded correlational model

The primary aim of the embedded correlational model is to move beyond numerical relationships and gain a deeper understanding of the factors shaping these associations. In traditional correlational research, findings may indicate a significant relationship between variables, but the reasons behind these associations often remain unclear. By incorporating qualitative data, researchers can explore participant experiences, motivations, or situational factors that contribute to these correlations. This approach is particularly useful when dealing with complex social, behavioral, or psychological phenomena where numerical trends alone may not fully capture the nuances of human interactions. Data collection in this model follows a predominantly quantitative structure, with qualitative data playing a supportive role. Researchers typically begin with quantitative methods, such as surveys or large-scale questionnaires, to identify correlations between key variables.

Strengths

The embedded design is ideal for when a researcher wants to prioritize one data type (quantitative or qualitative) while using the other as a supplementary component. These are other strengths in embedded research design:

• Supplemental data: Adding supplemental data enhances the overall research design by addressing gaps or providing additional context, leading to a more comprehensive and robust study.
• Collaborative environment: This design works well in collaborative research teams, allowing members to focus on specific questions aligned with their expertise and interests.
• Appealing to funding: Funding agencies less familiar with mixed methods research may find this design appealing due to its primary emphasis on a traditional quantitative or qualitative framework, making it more accessible.

Challenges

While the embedded research design comes with time and resource advantages, this design comes with several challenges:

• Expertise requirements: Researchers must possess expertise in both the primary design (quantitative or qualitative) and in mixed methods research. This dual competency is essential for successfully implementing the embedded design.
• Clarifying the purpose of supplemental data: The purpose of including supplemental qualitative or quantitative data must be clearly specified within the context of the larger study. Researchers can articulate these as primary and secondary purposes in their study.
• Timing of qualitative data collection: Determining when to collect qualitative data within an experimental study is critical. This decision should align with the intent of including qualitative data, whether it is to inform the intervention design, explore participant experiences during treatment, or provide follow-up insights after the experimental trial.
• Challenges in integrating results: Since the two methods in an embedded design typically address different research questions, integrating the results can be challenging. However, unlike a convergent design, the embedded design does not aim to merge the data sets. Researchers can keep the results separate and even publish them in distinct papers.
• Risk of treatment bias: In experimental studies that collect qualitative data during the intervention, there is a potential for introducing treatment bias, which could affect the experimental outcomes. Researchers need to carefully consider and mitigate this risk during study design.

Example of an embedded mixed methods research design

Victor, Ross, and Axford (2004) conducted a randomized controlled trial (RCT) to evaluate a health promotion intervention for individuals with osteoarthritis of the knee. In this study, they applied an embedded design, incorporating qualitative data from participant diaries and audiotaped group sessions within the predominantly quantitative experimental framework. While the quantitative component focused on measuring clinical outcomes, such as improvements in physical health and self-management, the qualitative data offered a deeper understanding of participants’ experiences. By collecting this qualitative data during the intervention, the researchers were able to explore how participants engaged with and perceived the program.

Similarly, Evans and Hardy (2002) investigated the effects of a goal-setting intervention on the rehabilitation of injured athletes. Their experimental design measured quantitative outcomes like rehabilitation adherence, self-efficacy, and psychological responses to injury. To enrich these findings, they embedded a qualitative element by conducting semi-structured interviews with participants after the intervention.

Conclusion

As research questions become more complex, there may be a need for designs that go beyond the supportive role of embedded methods. This is where transformative and multiphase designs come into play. The transformative design is especially valuable when research aims to address issues of social justice, equity, or advocacy. It integrates qualitative and quantitative data within a framework guided by a specific theoretical lens, often focusing on marginalized or underrepresented groups. This design not only seeks to understand phenomena but also to inspire change, making it a powerful method for researchers committed to transformative outcomes.

Similarly, the multiphase design extends the possibilities of mixed methods by combining multiple phases of data collection and analysis over time. This approach is ideal for large-scale studies or longitudinal projects where different research questions emerge at various stages. By integrating both qualitative and quantitative components across several phases, the multiphase design allows for a comprehensive exploration of complex issues, capturing changes, developments, and patterns that might be missed in single-phase studies.

As we transition to the next sections of this guide, we will explore these advanced designs in greater detail. Understanding the transformative and multiphase designs will provide researchers with the means to address more nuanced research questions, engage with diverse populations, and contribute to meaningful change in their fields.

References

1. Plano Clark, V. L., & Ivankova, N. V. (2016). Mixed methods research: A guide to the field. Thousand Oaks, CA: SAGE Publications.
2. Creswell, J. W., & Plano Clark, V. L. (2006). Designing and Conducting Mixed Methods Research. Thousand Oaks, CA: Sage
3. Creswell, J. W., & Plano Clark, V. L. (2011). Designing and conducting mixed methods research (2nd ed.). SAGE Publications.
4. Creswell, J. W., & Plano Clark, V. L. (2018). Designing and conducting mixed methods research (3rd ed.). Thousand Oaks, CA: SAGE Publications.
5. Victor, C. R., Ross, F., & Axford, J. (2004). Capturing lay perspectives in a randomized control trial of a health promotion intervention for people with osteoarthritis of the knee. Health Expectations, 7(3), 243–252
6. Evans, L., & Hardy, L. (2002). Injury rehabilitation: A goal-setting intervention study. Research Quarterly for Exercise and Sport, 73(3), 310–319