Three mixed methods designs
Chapter 13: Qualitative and Mixed Methods — Video 3
https://ghrbook.com/videos/three-mixed-methods-designs/

[Slide 1] Mixed methods research systematically integrates quantitative and qualitative data collection and analysis, to provide a more comprehensive understanding than either approach alone.
[Slide 2] Numbers can tell us what and how much, but we often need qualitative insight to understand why and how. Mixed methods research isn't just doing a quantitative study and a qualitative study in parallel and stapling them together in the discussion section, though I've seen people try. True mixed methods involves integration: bringing the qualitative and quantitative strands together in ways that create insights neither strand alone could provide.
[Slide 3] Mixed methods designs vary in complexity, but three basic designs cover most situations you'll encounter. They differ in timing, which method comes first, and in priority, which method is emphasized. Think of them as blueprints for how we'll structure a study: convergent, explanatory sequential, and exploratory sequential.
[Slide 4] In a convergent design, we collect quantitative and qualitative data simultaneously, or close to it, analyze each separately, then bring the results together to compare and contrast them. It's like running two parallel investigations of the same phenomenon, then merging them to see where they converge and where they diverge. We choose it when we want to validate or corroborate findings across methods, or when we want different but complementary data about the same phenomenon.
[Slide 5] Integration happens in joint displays: tables, matrices, or visual displays that bring the quantitative and qualitative findings side by side for direct comparison. Then we look for convergence, where both strands support the same conclusion, divergence, where they suggest different conclusions, and expansion, where one strand provides deeper understanding of patterns identified in the other.
[Slide 6] The vaccination governance study in northwest Syria from our last video used a convergent approach. The researchers collected quantitative vaccination coverage data while simultaneously conducting 14 key informant interviews and observations about governance processes. Integration happened when they created joint displays comparing vaccination coverage data against stakeholder reports of governance effectiveness.
[Slide 7] The joint displays revealed crucial discrepancies. Despite positive stakeholder perceptions of program effectiveness, the quantitative records revealed a lack of demographic data and an over-reliance on paper-based systems. This divergence prompted deeper investigation that neither strand alone would have revealed.
[Slide 8] In an explanatory sequential design, we start with quantitative data collection and analysis, then use qualitative methods to explain or elaborate on the quantitative findings. The quantitative phase identifies patterns, associations, or outcomes. The qualitative phase explores why those patterns exist or how the mechanisms work. We choose it when we have quantitative results that need explanation, like surprising findings, significant differences between groups, or outlier cases.
[Slide 9] A study of perioperative care in three teaching hospitals in southern Ethiopia is a clean example. The quantitative phase used the WHO and Harvard Surgical Assessment Tool to measure capacity across five domains. The survey identified specific gaps. The surgical, obstetric, and anesthesia specialist workforce was just 0.58 per 100,000 population. Blood products were available only 25 to 50 percent of the time, and perioperative management protocols were used in just 1 to 25 percent of cases.
[Slide 10] These numbers raised the obvious question: why? Twenty semi-structured interviews with surgeons, anesthetists, and nurses formed the qualitative second phase. Protocols were underused, and clinicians also described a poor patient-safety culture, fragmented interprofessional communication, and frustration that repeated quality-improvement projects produced no tangible change. Blood scarcity was a logistics failure and also a downstream effect of declining donations, broken collection practices, and unresolved national-level policy issues.
[Slide 11] Three themes emerged that the survey instrument could not have produced at all: the influence of national governance, sociopolitical turmoil, and global market volatility on local perioperative care. The team integrated the two strands through a joint display table, mapping each quantitative finding alongside the qualitative themes that explained it.
[Slide 12] In an exploratory sequential design, we start with qualitative exploration to identify important variables, develop measures, or understand a phenomenon, then follow with quantitative data collection to test patterns or generalize to a larger population. It's particularly valuable in global health research, where Western-developed measures or theories may not apply in different cultural contexts.
[Slide 13] My team developed a perinatal depression screening tool in rural Kenya using an exploratory sequential design. We started with the suspicion that screening instruments designed in high-income settings, like the Edinburgh Postnatal Depression Scale and the PHQ-9, might miss how Kenyan women actually describe perinatal distress.
[Slide 14] We assembled 365 items from 17 existing screening tools and pulled out 171 unique cover terms. Groups of pregnant women and new mothers, and groups of community health volunteers in Bungoma County, ranked which of those terms matched local experience, and they generated new local idioms of distress that no Western tool included. A panel of eleven Kenyan mental health professionals reviewed the results and signed off on a blended pool of 60 items.
[Slide 15] The quantitative phase then validated this blended tool against a gold standard. A random sample of 193 pregnant women and new mothers completed the Edinburgh scale, the PHQ-9, and the new pool of 60 items, along with a structured clinical interview administered by trained Kenyan counselors. Item analysis and combinatorial optimization across more than 600,000 possible scale combinations narrowed the pool to a Perinatal Depression Screening tool with just 9 items. We saw this tool's cutoff in the measurement chapter.
[Slide 16] In this population, the new tool outperformed both established scales, with sensitivity and specificity of 0.90 and 0.90, compared to 0.70 and 0.72 for the Edinburgh scale and 0.70 and 0.73 for the PHQ-9. The qualitative phase ensured the tool was grounded in how Kenyan women actually talk about distress, and the quantitative phase confirmed that grounding paid off in diagnostic accuracy.
[Slide 17] The defining feature of mixed methods is integrating qualitative and quantitative research, and merely doing both isn't enough. Ask yourself: how will I bring these strands together? What insights emerge from integration that couldn't emerge from separate analyses? If you can't answer these questions, maybe you're doing two separate studies rather than true mixed methods.
[Slide 18] Integration happens through joint displays; through data transformation, converting one type of data to the other; through narrative weaving, writing that explicitly connects findings across strands; and through meta-inferences, interpretations that draw on both quantitative and qualitative insights. The best mixed methods studies make integration explicit and central.
[Slide 19] Global health faces complex challenges that do not fit neatly into single methodological boxes. Understanding whether an intervention works calls for randomized trials. Understanding how it works calls for qualitative inquiry into mechanisms. Understanding whether it will work in a new context calls for mixed methods implementation research. And understanding how to adapt it calls for participatory approaches that center community knowledge.
[Slide 20] The most interesting global health research is increasingly comfortable moving across these approaches, as a strength and not as a compromise. As you develop your own work, resist the pressure to pick one methodological camp and stick to it. Let the question lead. Sometimes that means quantitative work, sometimes qualitative, often mixed methods, and increasingly participatory approaches that share research power with communities. The goal is evidence that is rigorous, contextually grounded, and useful for improving health and advancing equity.