What does mixed-methods research look like in practice?
A mixed-methods study uses quantitative and qualitative evidence together to answer a combined research question. A useful example shows the question, the design, who contributes, how each strand is analyzed and what is learned by bringing the findings together.
This guide contains three fictional worked examples: training transfer, member services and employee listening. All organizations, sample sizes and results below are invented for teaching. They are not customer results or published studies. Use them to understand the method, then adapt the design to your own question and constraints.
The examples illustrate three different relationships between evidence. One uses interviews to investigate an earlier numerical result. One compares findings collected during the same period. One uses interviews to help develop a later survey. For the broader choice, see mixed-methods research designs.
Scroll horizontally to see all columns →
| Example | Question | Design | What integration contributes |
|---|---|---|---|
| Training transfer | Why is reported use of a new skill uneven? | Explanatory sequential | Interviews investigate a pattern in follow-up ratings |
| Member services | How useful are services, and what makes access difficult? | Convergent | Ratings and accounts distinguish service quality from access |
| Employee listening | What does useful manager support mean in this workplace? | Exploratory sequential | Interviews help develop more relevant survey questions |
Example 1: explain an uneven training result
The question and design
A training team wants to know whether people use a new planning technique at work and what helps or prevents its use. It selects an explanatory sequential design: analyze a follow-up survey first, then interview a small, deliberately varied group to investigate the pattern.
The team enrolls 100 learners. Six weeks after training, 70 answer the follow-up survey. Of these, 42 say they used the technique at least once: 60% of respondents. This is not a 60% rate for all learners; the experience of the 30 nonrespondents remains unknown.
Quantitative finding
Among 40 respondents who report an opportunity to use the technique, 32 used it: 80%. Among 30 who report no clear opportunity, 10 nevertheless report some use: 33.3%. That apparent mismatch needs investigation. The questions may capture different interpretations of opportunity, or learners may have created informal ways to try the skill.
The team also checks whether the groups differ in role, workload and access to manager support. These are descriptive findings from respondents, not proof that providing an opportunity would cause the same increase for everyone.
Qualitative investigation
With appropriate consent, the team invites interviews across four combinations: opportunity and use, opportunity without use, use without a clear opportunity, and neither. Twelve learners participate. The sample is chosen for explanation and variety, not to estimate how common each barrier is across the whole cohort.
Interview accounts suggest three issues to examine: some roles lack a relevant task, some learners consider a small personal experiment to be “use,” and some wait for manager permission before applying a new technique. The analyst preserves these distinctions instead of coding all non-use as poor motivation.
Integrated finding
Scroll horizontally to see all columns →
| Evidence | Interpretation | Next step |
|---|---|---|
| Reported use differs by reported opportunity | Task context may matter, alongside skills and confidence | Review role-specific opportunities with managers |
| Some learners report use but no opportunity | The two questions may be interpreted differently | Clarify what counts as practice and opportunity |
| Interviewees describe waiting for permission | A practical barrier may sit outside the course itself | Pilot a manager-supported practice task and review uptake |
The combined evidence changes the response. Instead of immediately adding more training, the team tests a relevant practice task and clearer manager expectations. It records the intervention, owner and next review date. Follow-up should examine both use and the opportunity to use the skill, with nonresponse and other changes noted.
This study supports a practical hypothesis. It does not establish the training's causal effect or show that the interviewed barriers explain every learner's experience. Continue with measuring behavior change after training.
Example 2: compare member ratings and accounts of access
The question and design
A professional network wants to understand how useful members find a service and whether access differs across local groups. During one reporting period, it collects a short survey and interviews selected members. It analyzes each strand before comparing them in a convergent design.
The shared survey core defines the service, period, usefulness scale and member-tenure groups. Chapters can add relevant local questions. The analysis combines only fields that share definitions; a local question about event quality is not silently treated as a rating of advisory support.
Quantitative finding
There are 120 valid usefulness ratings: 80 from established members and 40 from newer members. Sixty established members give a favorable rating, or 75%. Twenty newer members do so, or 50%. The overall favorable share is 80 of 120, or 66.7%.
The 25-percentage-point group difference raises a question. Before interpreting it, the team checks recruitment, missing answers and service use. If newer members who never found the service were less likely to answer, the observed results may miss an important part of the access problem.
Qualitative finding
The team conducts 12 interviews selected to include different tenure groups and experiences. Some established members describe informal referral routes. Some newer members describe uncertainty about whom to contact. Several accounts distinguish helpful advice from the time and effort required to obtain it.
These interviews are not necessarily with the survey respondents. Therefore, the analyst does not attach an interview explanation to a particular rating or report interview themes as percentages of all members. The integration is at the group and service-experience level.
Integrated finding
A joint display places each group's usefulness results beside relevant interview themes and records an interpretation. The emerging issue is that useful advice may coexist with uneven access. A high usefulness score does not rule out delays or confusing entry points.
The team tests clearer entry instructions and a named first contact. It retains the usefulness item and adds an appropriately tested access question for the next cycle. Because this changes the instrument, it records the new question's start date rather than pretending earlier cycles measured the same thing.
One shared data dictionary supports comparison while local teams keep questions appropriate to their work. The next analysis can distinguish whether access became clearer, whether usefulness changed and who remains missing from the evidence. See the stakeholder question guide for planning that shared core.
Example 3: develop an employee survey from interviews
The question and design
A growing business wants to improve manager support but finds its existing question, “My manager supports me,” too broad to guide action. It chooses an exploratory sequential design: interview employees to understand the concept, then develop and test a short survey.
The purpose is workplace improvement. Before inviting employees, the team explains who will access the material and how findings will be reported. It considers small-team identification risks and avoids promising anonymity that the process cannot deliver.
Qualitative development
Fifteen interviews cover different roles, locations and working arrangements. The analyst develops themes through close reading and review. For this fictional example, three recurring aspects of support emerge: clarifying priorities, removing practical obstacles and helping people make time for development.
These themes help define possible questions; they are not yet a validated measurement scale. The team checks whether important perspectives are missing and tests draft wording with employees who were not involved in the original interviews.
Build and test the quantitative strand
Instead of one broad item, the draft asks separately whether the manager helps clarify competing priorities, helps resolve work obstacles and supports time for agreed development. Each item specifies a relevant reference period and offers a suitable option for people without that experience.
A pilot produces 60 responses. Fifty-four answer the priorities item; 42 give a favorable answer, or 77.8% of valid answers. Forty-eight answer the development item; 24 are favorable, or 50%. These are separate item results with different denominators. The team should not average them into a single “support score” without a sound measurement rationale.
Integrated finding
The interviews explain why the survey separates these dimensions. The pilot suggests development time may deserve investigation, while wording feedback identifies any misunderstood items. The next step is to refine the instrument and plan a broader collection, rather than announcing that half the workforce lacks support.
If the organization wants a formal scale or intends to make high-stakes decisions, further measurement expertise and validation may be needed. A practical pilot can improve questions without establishing reliability, validity or representativeness for every use.
What makes these examples mixed methods?
Each has a deliberate connection. The training survey guides interview sampling and questions. The member study compares two strands around a common experience. The employee interviews shape a later instrument. In each case, integration changes the design or interpretation.
This follows the broader distinction between integration through connecting, building and merging evidence described by Fetters, Curry and Creswell. Simply collecting comments and scores without analyzing their relationship does not demonstrate that connection.
A useful study plan names the combined question, each strand's role, sample relationship, timing, analysis method and point of integration. It also states what the evidence will not establish. For a step-by-step treatment, see mixed-methods data analysis.
Keep the recurring work manageable
Repeated studies create maintenance work: question versions, group definitions, source records, coding decisions and permissions. Plan these early. Store the context needed for each analysis, rather than imposing personal identification on every form. Group-level research and anonymous feedback can be appropriate.
Sopact's relevant role is bringing collection, analysis and governance into a workflow an operational team can maintain. Test how a reviewer moves from a combined finding to the underlying authorized evidence, how local questions coexist with a shared core and what happens when a new period arrives.
AI assistance can help organize material and draft codes, but it can misread context or create unsupported summaries. Review its output, preserve source material and record significant corrections. A quotation attached to a finding is a starting point for verification, not a guarantee of accuracy.
To present the findings, use the impact report guide and report examples. Keep the question, methods, integrated interpretation and limits visible to the reader.
How Sopact reduces coding and reporting work
For a one-off study, a bounded manual workflow may be appropriate. Repeating these designs across cohorts changes the workload: every cycle brings new coding, revisions and integration.
A workflow with repeated manual work
- Define from an initial sampleRead material and agree on the codebook.
- Apply it across the datasetCode responses and check the result.
- Revise a definitionReturn to affected material and recode it.
- Reconnect the numbersReconcile coded results with ratings and context, then rebuild the view.
The Sopact workflow
- Your team owns the definitionsDecide what each code means and improve it as you learn.
- Apply coding across the eligible dataAutomate application; people review quality and exceptions.
- Reprocess after a definition changesReapply the revised definition across the configured scope instead of recoding each response by hand.
- Ask across coded text and numbersKeep the response, rating and relevant record context connected; inspect the evidence behind the result.
This compares workflow patterns, not a claim that every research tool requires manual coding or separate files. Some already automate parts of this work; compare the complete cycle.
For this codebook-based workflow, the main saving is repeated application and reconnection—not the removal of human judgment. A changed definition can be reapplied across the configured data while reviewers concentrate on quality, exceptions and interpretation. Coded text stays connected to the relevant ratings and context.
Count the recurring work in ownership cost. Include setup, coding, recoding after revisions, source reconciliation, review and reporting, plus your actual platform and processing expenses. A worked scenario of four cycles of 4,000 responses illustrates 272 fewer annual staff hours; it is an assumption-based example, not a customer benchmark. Existing automation, review needs and implementation effort can substantially change the result.
Adjust the workload assumptions and compare total effort →
A reliable assistant should calculate from the selected records and let a reviewer open the supporting evidence. Check the data scope, definition, denominator and access permissions. Reproducible arithmetic does not make every AI interpretation correct.
Watch: Why Qualitative Analysis Stays Small — And How to Scale It
See why revising a codebook creates repeat work, and how connected coding and quantitative analysis change that workload.
Watch: combining qualitative and quantitative evidence
This companion video introduces mixed-methods research. Use the three designs above to decide what combining evidence would contribute to your own question.
Frequently asked questions
Are these real research studies?
No. They are fictional teaching examples with invented samples and results. They demonstrate research planning and interpretation, rather than reporting customer outcomes or published findings.
What is a simple mixed-methods research example?
A training team surveys later skill use, then interviews a deliberately varied group to investigate the pattern. The survey findings guide the interviews, and the two strands support an integrated explanation with clear limits.
Do both strands need the same sample size?
No. Each sample should fit its purpose. A larger survey and a smaller purposive interview sample can work together, provided the sampling relationship and limits are explained.
Does every study need a persistent personal ID?
No. Some studies integrate findings at a group or conceptual level. Appropriate record keys help when individual responses or waves must be linked, but personal identification should serve a clear purpose.
Can interview themes establish population percentages?
Not simply because they can be counted. The sampling and analysis must support that interpretation. In these purposive examples, theme counts describe the interviews and do not estimate prevalence across the wider population.
How do I choose between the three designs?
Use explanatory sequential work when an earlier quantitative result needs investigation, exploratory sequential work when qualitative learning should shape later measurement, and a convergent design when both strands can contribute to the same question during a comparable period.
Choose your next step
ASME
Explore how quantitative and qualitative stakeholder evidence supports a membership organization. The examples above are fictional, not results from this story.
Read the customer story →