Mixed Methods Applications: Illustrations
Integration means bringing the quantitative and qualitative strands of a mixed methods study together for analysis and comparison, or for one strand of the data collection and analysis to inform the other. Often investigators discuss the integration in the text of manuscripts but joint displays of the data may provide another way to integrate the study strands.
A joint display brings quantitative data together with themes or analysis derived from qualitative data.
How the "mix" of quantitative and qualitative "strands" is configured in a particular study must be justified by the goals of the study and the questions to be answered by the research. Sequential designs (exploratory and explanatory sequential designs) have data collection performed in sequence, so that the results of one strand influence the data collection for the next strand. In concurrent designs the data collection for the strands are embedded in one another, often with one method being primary.
In an exploratory sequential design, qualitative data is collected prior to the collection of quantitative data. This type of design is likely to be familiar to many investigators as it is commonly used as a strategy to design an instrument or questionnaire or to gather information to guide the content of an intervention (e.g. focus group: a guided discussion that systematically investigates what a diverse group of people think of a set of research questions).
An alternative to the exploratory sequential design is the explanatory sequential design in which the quantitative data collection comes first, followed by qualitative data collection. Investigators who select this design often do so because there is some need for an explanation of the quantitative findings from study participants.
In some cases, the research questions or circumstances of the research settings require that the data collection is essentially concurrent, often referred as “embedding” one strand in another, typically with one strand dominating. For concurrent designs, less emphasis is placed on how one strand informs the next, and more on making inference or drawing conclusions from the concurrent strands.
Basic mixed methods designs that deploy qualitative and quantitative methods sequentially or concurrently (Adapted from (J. W. Creswell & Plano Clark, 2011)).
Mixed methods may be valuable throughout the development and testing of interventions - in the development of the intervention, during the evaluation of the intervention, and after the follow-up and assessment of outcomes is completed. Qualitative approaches are most frequently used to develop an instrument, to understand strategies for successful recruitment, to find areas for intervention adaptation, to understand the processes of an intervention, to evaluate fidelity and other implementation factors, to explain outcomes, to provide feedback to improve intervention, and to understand mediators and moderators.
Traditional efficacy and effectiveness clinical trials focus on improving individual-level clinical and functional outcomes. For example, the National Institute of Mental Health (NIMH) is leading the efforts to move towards an "experimental medicine approach" that generates knowledge about 'mechanisms' underlying a disorder or a service use pattern (Insel, 2014). An emphasis is on understanding mechanisms and designing studies in such a way that even if an intervention has minimal effects, it will be possible to inform future improvements or modifications to the intervention (O'Cathain, Murphy, & Nicholl, 2007). Mixed methods are essential to achieving these objectives as it is no longer tenable for an investigator to only answer the question, "does this work?" The investigator must also be prepared to address the questions of "why didn't this work?," "why didn't this work for this group?," and "why didn't this intervention reach the people for whom it was intended?"
More detailed information can be found in Gallo, J.J. & Lee, S.Y. Mixed Methods in Behavioral Intervention Research, in Gitlin, L.N. & Czaja, S.J. (Editors). Behavioral Interventions: Designing, Evaluating and Implementing. Springer Publishing Company, New York, in press.
The role and timing of using quantitative and qualitative approaches in randomized clinical trials (adapted from Sandelowski (Sandelowski, 1996)).