The part that makes a study mixed methods rather than two studies — where integration happens, how to show it, and how it is judged.
Integration is the point at which the two strands of a mixed methods study meet and produce something neither could produce alone. It is what makes a study mixed methods; two strands reported one after another are two studies in one document, whatever the methodology chapter calls them. Integration can happen at three points — in the design, in the handling of the data, and in the interpretation — and a strong thesis is explicit about which it is using.
The mixed methods literature distinguishes integration at the level of design, methods, and interpretation and reporting. Knowing which you are doing is what lets you describe it precisely rather than asserting that strands were “combined”.
The structure of the study itself connects the strands. In a sequential design, one strand shapes the other: quantitative results decide who is interviewed, or qualitative findings decide what is measured. In a convergent design, both address the same question from different directions by design. In an embedded design, one strand answers a question arising from the other.
This is the level most doctoral studies achieve, because it is built into choosing the design. It is necessary and not sufficient — a sequential design whose phases never actually inform each other has design-level integration on paper only.
Data from one strand is connected to data from the other in a concrete, describable operation. This is where most theses could do more.
Practical forms it takes:
The strands are drawn together into a single argument. The conclusions you reach from the combination — rather than from either strand alone — are what the literature calls meta-inferences, and they are what the thesis is ultimately for.
Two reporting structures are used. Weaving presents quantitative and qualitative findings together theme by theme or question by question. Staging reports each strand in turn and then adds a dedicated integration section. Both are acceptable; what is not acceptable is staging without the integration section, which is where most theses land by default.
The most practical integration tool available, and the one an examiner can look at directly.
A joint display is a table setting the two strands side by side so the comparison is made explicitly rather than gestured at. In its simplest form it has three columns: the quantitative finding, the corresponding qualitative finding, and what the combination means.
| Quantitative finding | Qualitative finding | What the combination shows |
|---|---|---|
| Training had no measurable effect on reported confidence | Participants described the training as useful but impossible to apply under current staffing | The intervention was acceptable but the setting prevented use, so the null result reflects implementation rather than content |
| Confidence was significantly lower in one site | That site had recently reorganised, and accounts centred on unclear responsibilities | The site difference is explained by organisational disruption rather than by the intervention |
Building one is also a diagnostic. If you cannot fill in the third column, the strands have not been integrated — you have two sets of findings sitting next to each other. That is worth discovering while you can still do something about it.
Displays can be organised by research question, by theme, by participant group, or by case where the same people are in both strands. Whichever you choose, include the rows where the strands disagreed. A display showing only agreement invites the question of what happened to the rest.
When you compare findings, the relationship falls into one of a few patterns, and naming them makes the discussion more precise than saying results were “broadly consistent”.
Discordance needs working through rather than resolving. Ask what would explain it: are the strands measuring different constructs; is an average concealing a subgroup; is one mode of data collection inviting a more managed answer; is the instrument operationalising something differently from how participants understand it? Any of those is a finding. Concluding that the divergence is a methodological artefact is also legitimate, provided the reasoning is shown.
What damages a thesis is a discussion that reports agreement in detail and passes over disagreement in a sentence.
Examiners assess mixed methods work on the integration more than on either strand, because either strand alone could have been a single-method study. The questions they ask are predictable:
One structural habit prevents most integration problems: write the integration section before the individual strand chapters are finalised. Doing it last, under deadline, is why it so often becomes a summary of both strands rather than an argument built from them.
Bringing the quantitative and qualitative strands together so the combination produces something neither could produce alone. It can happen at three levels: in the design, where one strand shapes the other; in the handling of the data, through connecting, building, merging or transforming; and in interpretation, where the strands are drawn into a single argument. Integration is what makes a study mixed methods rather than two studies in one document.
A table setting findings from both strands side by side so the comparison is explicit — typically the quantitative finding in one column, the corresponding qualitative finding in the next, and what the combination means in a third. It is the most practical integration tool available and gives an examiner something concrete to assess. It is also a diagnostic: if you cannot complete the third column, the strands have not actually been integrated.
Not only in the discussion. Two structures work: weaving, where findings from both strands are presented together theme by theme, or staging, where each strand is reported in turn followed by a dedicated integration section. Staging without that section is where most theses end up by default, and it is the version examiners criticise. Writing the integration before finalising the strand chapters prevents it becoming a rushed summary.
The conclusions you can draw from the combination of strands that neither strand licenses on its own. They are the point of a mixed methods design, and the practical test of whether you have any is simple: state your main conclusion in one sentence, then check whether it survives deleting either strand. If it does, you have two studies rather than one integrated one.
In full, and interpreted rather than resolved. Work through what would explain it — the strands may be capturing different constructs, an average may be concealing a subgroup, one collection mode may invite a more managed answer, or participants may understand the concept differently from the instrument. Any of those is a finding. Concluding it is a methodological artefact is also legitimate if you show the reasoning. What damages a thesis is reporting agreement in detail and passing over disagreement in a sentence.
This is data transformation, and it is a recognised integration technique, but it needs justifying rather than doing by default. Counting how many participants mentioned a theme imports a frequency logic that most purposive qualitative samples cannot support — the count reflects who you recruited. It is more defensible where the qualitative strand was designed with systematic categorisation in mind, as in content analysis, than where it used an interpretive approach.
If the third column will not fill in, the integration is not there yet — and that is far better found now than in the examination. Send your findings from both strands and a PhD in your field will work through it with you.
Leave an email or a WhatsApp number — whichever you prefer — and tick how we should reach you. We reach out within 30 minutes.
