Mixed-methods isn't "a survey plus some interviews". It's a deliberate design where qualitative and quantitative strands inform each other in a specific, named structure. Without that integration, a committee will read it as two disconnected mini-studies stapled together.
| Design | Sequence | Best Fit |
|---|---|---|
| Convergent | Qual and quant collected at the same time, merged at interpretation | Comparing or corroborating findings from two angles |
| Explanatory sequential | Quant first, then qual to explain unexpected results | Statistical findings that need contextual explanation |
| Exploratory sequential | Qual first, then quant to test or generalize what emerged | Building/testing an instrument from qualitative themes |
Each of these three designs solves a different problem, which is why naming the wrong one, or picking one out of familiarity rather than fit, creates friction later in the dissertation. A convergent design is chosen when you want two independent angles on the same question at roughly the same time, useful when you want to know whether a self-reported survey measure and an observed or interview-based account of the same phenomenon actually agree. An explanatory sequential design is chosen when you already expect to run a quantitative study but anticipate that the numbers alone won't explain why a pattern exists. The qualitative phase exists specifically to answer the "why" the statistics can't. An exploratory sequential design is chosen when too little is known about a construct to measure it quantitatively yet. The qualitative phase generates the themes or items that the quantitative phase then tests or scales up. Choosing based on which problem you actually have, rather than which design sounds more sophisticated, is the first real methodological decision in a mixed-methods study.
The defining feature of mixed-methods, and the part committees scrutinize most, is the integration point: the specific place where qualitative and quantitative strands are brought together and interpreted jointly, not just presented side by side. This might be a joint display (a table merging qual themes against quant results), a narrative weaving section, or a data transformation (turning qual themes into quant variables, or vice versa).
Four practices separate mixed-methods work that reads as genuinely integrated from work that reads as two studies sharing a title page. First, name your design explicitly: "a convergent parallel design," not a vague description of "combining quant and qual", because the name signals to your committee that you understand the methodological literature you're drawing on and are following its established conventions rather than improvising. Second, justify the mixing rationale using one of the recognized purposes for mixing methods. Triangulation (cross-validating one finding with another method), explanation (using one strand to explain the other), exploration (using one strand to build what the other will test), or development (using one strand's results to develop an instrument or intervention for the other). Tied specifically to your research questions rather than mixing simply because more data feels more thorough. Third, build an explicit integration point that is a real analytic step, not just two results sections placed next to each other in the same chapter. Fourth, address differing sample sizes between strands directly in your methodology. It is entirely normal for a qualitative sub-sample to be much smaller than a quantitative sample, and treating this as a problem to hide rather than a difference to explain is a common, avoidable weakness.
A joint display is the easiest integration tool to defend. A table or matrix showing how a qualitative theme aligns (or doesn't) with a quantitative finding gives your committee a concrete artifact of integration. Far more convincing than a paragraph asserting that the two strands "support each other."
Beyond simply picking convergent, explanatory sequential, or exploratory sequential, a mixed-methods design has to be specific about two further decisions that committees expect stated explicitly: timing and priority. Timing is whether strands are collected concurrently (at the same time, independently) or sequentially (one strand's results shape the next phase). Priority, sometimes written using notation like QUAN → qual or qual → QUAN, signals which strand carries more analytic weight in answering the overall research question, with capitalization indicating dominance. These aren't cosmetic details. A study that collects quant and qual data concurrently but treats the qualitative strand as merely illustrative is functionally a quant-dominant convergent design, and should say so rather than presenting the two strands as equally weighted when the analysis doesn't actually treat them that way. Getting the notation and the actual analytic practice to match is one of the more overlooked forms of rigor in mixed-methods work.
Consider a study on whether a new advising model improves student sense of belonging, using an explanatory sequential design (QUAN → qual). Phase one surveys 300 students using a validated belonging scale before and after the new advising model is introduced, finding a statistically significant but modest increase in belonging scores. That modest, somewhat puzzling effect size is exactly what triggers the qualitative phase: 15 students are purposively selected, a mix of students whose scores rose sharply and students whose scores barely moved, for semi-structured interviews exploring their advising experience. The qualitative analysis reveals that students whose scores rose sharply described advisors who proactively reached out between scheduled meetings, while students whose scores barely moved described advisors who only responded when contacted. The integration point is a joint display: one column with the quantitative belonging-score change per participant, a second column with their qualitative theme, showing a clear alignment between proactive outreach and larger belonging gains. That table, not a separate discussion of "quant findings" and then "qual findings", is the actual integrated analysis a committee is looking for, and it directly explains the modest average effect size the quantitative phase alone couldn't account for.
At the master's level, combining a survey and some interviews with a general explanation is often accepted. At the doctoral level, the design must be named precisely with its timing and priority notation, the mixing rationale must be tied specifically to the research questions (not asserted generically), and, most importantly, the integration itself must produce an insight that neither strand could have produced alone. Doctoral committees, particularly those with a member versed in mixed-methods methodology specifically, will often ask directly: "what did integrating these two strands tell you that one strand alone wouldn't have?" A dissertation that can't answer that question convincingly is, in the committee's eyes, not really a mixed-methods study. Just two studies sharing a cover page.
A named design, justified rationale, and a real integration point, not two studies stapled together.
No. Many mixed-methods studies are deliberately quant-dominant or qual-dominant depending on the research priority. What matters is being explicit about the weighting and the reason for it, not pretending both strands are equal when they aren't.
Yes. This is common. We focus on building the integration point and the joint analysis that ties your existing datasets together, rather than redoing data collection.
Qualitative and quantitative sample sizes follow different logics. Saturation for one, statistical power for the other. We write separate, appropriately justified rationales for each rather than forcing one standard onto both.
That's a legitimate and often publishable finding, not a problem to bury. We help you build a discussion section that interprets the divergence honestly. Sometimes it points to a moderating variable or a limitation in one strand's measurement rather than a genuine contradiction.
It depends on what triggers the second phase: if you need a large-scale finding explained in depth, choose explanatory sequential (quant first); if you need to build or validate an instrument from qualitative themes, choose exploratory sequential (qual first); if you want to compare or corroborate two independent angles on the same question, choose convergent. We help you choose based on your specific questions, not a generic preference.