Mixed methods designs: more than just data collection

    loading  Checking for direct PDF access through Ovid


I recently read the article by Esquibel and Borkan1 which described their study of the experiences of patients with chronic noncancer pain being treated with opioids and that of their providers. The authors clearly articulate the unique contribution of their study to existing knowledge and make note that much of the literature on this phenomenon is based on data obtained through quantitative methods; however, I am concerned with their stated design and methods for data collection and analysis. The authors claim that the study is a mixed methods design and they in fact collect data using various methods: in-depth interviews, several instruments (eg, SF-12), and chart audits. The latter is not mentioned in the data collection section of the article but is presented briefly in both the analysis and results sections.
Missing in the article is any discussion of why a mixed methods design was the appropriate approach for exploring the research questions or how the data were mixed in the analysis phase. It is concerning if the authors believed that the types of data and means for collecting them, alone, constituted mixed methods research. This may be a common misconception among investigators, but there is a body of literature on how to design and implement studies using mixed methods.2–4 In fact, it appears that the only purpose of administering the numerous questionnaires (Table 3) and collecting participants' characteristics (Table 2) was to describe the sample. Had these quantitative data been triangulated with the qualitative data, then mixing would have occurred.
Authors writing about mixed methods designs have stated, “Central to the effectiveness of a mixed methods study is a clear and strategic relationship among the methods in order to ensure that the data converge or triangulate to produce greater insight than a single method could.3” Other authors have discussed a “priority method” and a “complementary method” in mixed methods research and that the sequence of the methods is important.4 This is absent from the article by Esquibel and Borkan and it is unclear to this reader why. One idea may be space limitations in the journal. Another possibility is that the authors may have felt the need to include quantitative data in the study given these are traditionally more desired and accepted in the biomedical and health sciences. I do not believe that the inclusion of the questionnaires aided in achieving the purpose of the study which was “To explore ways in which opioids [sic] medication influence the doctor-patient relationship.1” This aim could have been achieved solely through the qualitative portion of this study.
Given these concerns, I think it is an opportune time to remind investigators, authors, readers, and reviewers that collecting different types of data by different means (and subsequently including them in a single article) does not constitute mixed methods research. In addition, the burden to participants of partaking in research must be considered; completing interviews and multiple questionnaires is time consuming. If these data are not used in a manner that is consistent with a particular method or methodology, then investigators must ask if collecting them was necessary in the first place.
    loading  Loading Related Articles