+254 701 066 845

Qualitative and Mixed-Methods Research

From voices to themes: qualitative and mixed-methods research support in Kenya

A clear guide to transcription, coding, interpretation and full project support, so you know exactly which help you need and which work must remain yours.

14 min read · Published 29 October 2026

COST BARRIERTRUST IN STAFFDISTANCEACCESSTHEME

Why qualitative work needs more than a transcript

Every year, Kenyan graduate students, public health teams, education researchers and NGO evaluators collect hours of interviews and focus group discussions. The recordings feel like progress, and they are. Yet many projects stall at exactly this point, because a folder of audio files is not yet a finding. Someone must convert speech into text, build a way of organising it, notice patterns, and explain those patterns to a reader who was not in the room.

This guide explains that journey in plain language. It also draws firm lines between four different kinds of help: transcription, coding, interpretation and full project support. Those lines matter for quality, and they matter for academic integrity. If you search for qualitative data analysis in Kenya, you will find offers that blur all four together. A careful researcher separates them, uses support where it is legitimate, and keeps authorship of the argument.

A transcript records what was said. A theme explains why it matters. Only the researcher can stand behind that explanation.

Four stages, four different jobs

Think of qualitative research as a pipeline. Each stage has a different purpose, a different skill set and a different level of researcher ownership. Scroll sideways on a phone to see all four.

  1. 01

    Transcription

    Turns audio or video into accurate written text, with speakers labelled and language choices recorded.

    Does not label, interpret or judge the content.

  2. 02

    Coding

    Applies short labels to segments of text, guided by a codebook that you can defend and revise.

    Does not decide what the findings mean.

  3. 03

    Interpretation

    Groups codes into themes and explains what they say about your research questions. This stage is yours.

    Cannot be handed to anyone else if the work is to remain your own.

  4. 04

    Full project support

    Coordinates the whole journey: planning, tools, methodology guidance, feedback on drafts and structure.

    Does not replace the researcher as author of the conclusions.

Transcription is largely technical and can be done by a trained listener. Coding is analytical but guided by your codebook, so a specialist can apply it under your direction and then you review every decision. Interpretation is where your judgement, theory and context come together. Full project support surrounds all of this with planning, tool training and feedback, without taking over the thinking.

Interview and focus group transcription in Kenya

Good interview transcription in Kenya begins before anyone presses play. The quality of your transcript depends on the quality of your recording, the languages spoken and the clarity of your instructions.

Verbatim, clean verbatim or translated

A verbatim transcript keeps every pause, false start and filler. It suits conversation analysis and some narrative studies. A clean verbatim transcript removes fillers and repetitions but keeps the meaning and the speaker voice, and it suits most thematic studies. If your participants spoke Kiswahili, Sheng or a mother tongue such as Kikuyu, Luo or Kalenjin, decide whether you need the original language, an English translation, or both side by side. Back translation by a second person is a common quality check when the translated text will be quoted in a thesis.

Focus group transcription needs speaker discipline

Focus group transcription is harder than one to one transcription. Voices overlap, people interrupt, and side comments carry meaning. A good transcript labels speakers consistently (for example Participant 1, Participant 2, Moderator), marks inaudible sections with timestamps and notes laughter or silence when it affects meaning. Ask the moderator to note the seating order or voice features right after the session, because this makes later labelling far more accurate.

Prepare your recordings

  • Record in a quiet room and place the device near the centre of the group.
  • Test the device before every session and keep spare batteries.
  • Name files consistently, for example by site, date and participant code, never by real names.
  • Write down any specialised terms, place names and abbreviations that the transcriber should expect.
  • State your preferred format, such as speaker labels, timestamps every minute and your template.

Remember that transcription alone is not analysis. A well formatted transcript is a faithful record, and it is the foundation on which everything else rests.

Coding qualitative data and developing a codebook

Coding qualitative data means reading the transcript closely and attaching short labels to meaningful segments. A code is not a theme. A code is a small, specific handle such as ACCESS_COST. A theme is a broader pattern of meaning built from several codes.

Inductive, deductive or both

In inductive coding, codes emerge from the data itself, which suits exploratory questions. In deductive coding, you begin with codes drawn from a framework, such as a health behaviour model or an education policy framework, and test how well they fit. Many Kenyan theses use a hybrid approach: a small set of framework codes plus new codes that arise from participants. Your methodology chapter should state which approach you used and why.

What good codebook development looks like

Codebook development is the discipline that makes coding transparent. Each code needs a name, a definition, rules for inclusion and exclusion, and an example quote. When two people code the same data, the codebook is what allows them to compare decisions and resolve differences. Here is a short sample drawn from a fictional study on access to maternal health services.

Sample codebook (fictional study, for illustration only)
CodeDefinitionInclusion and exclusionExample
ACCESS_COSTParticipant mentions money, fares or fees as a barrier to using a service.Include: clinic fees, matatu fare. Exclude: general complaints about prices with no link to use.“I counted my fare before I counted my pain.”
STAFF_TRUSTParticipant expresses confidence or doubt in staff behaviour or competence.Include: respect, rudeness, confidentiality. Exclude: staff shortages alone.“The nurse greeted me by name.”
DISTANCEParticipant refers to travel time, terrain or distance to a facility.Include: walking hours, flooded roads. Exclude: waiting time inside the facility.“The river cuts us off in the rains.”
FAMILY_DECISIONParticipant describes decisions shaped by relatives or elders.Include: permission, advice. Exclude: financial support only (use ACCESS_COST).“My mother in law decides first.”

Notice that the exclusion rules do as much work as the definitions. They stop codes from swallowing each other and keep the analysis consistent from the first transcript to the last.

A worked example: from an excerpt to a theme

The excerpt below is entirely fictional. It was written to show the logic of coding and theme building, and it does not come from any real participant or study.

Fictional interview excerpt

Interviewer: What made it difficult to attend the clinic during your pregnancy?

Participant 7: The fare was about half of what I earn in a day, and the road floods when it rains, so sometimes we walk for hours. When I finally arrived, the nurse knew my name and did not shout at me, so I went back.

  • Code: ACCESS_COST
  • Code: DISTANCE
  • Code: STAFF_TRUST

Candidate theme: “Access is a daily trade off between cost, distance and respect.”

Look at what happened. The transcriber produced the text. A coder, working from your codebook, applied three labels. But the sentence in the last box is an interpretation. It claims that these three codes belong together and that respect can outweigh hardship. Whether that claim is supported depends on the other transcripts, on negative cases and on your theoretical framework. That is the stage where you must think for yourself, and where a supervisor expects to see your voice.

Thematic analysis, content analysis and open-ended survey responses

Thematic analysis help is among the most requested forms of support, and the method is widely taught. Braun and Clarke describe a sequence that moves from familiarisation, to initial coding, to searching for themes, reviewing them, defining and naming them, and finally writing up. You can read more about the journal literature on this approach through Taylor and Francis Online, where the original methodological papers are indexed.

Thematic analysis versus content analysis

Thematic analysis looks for patterns of meaning and invites interpretation. Content analysis often counts how frequently categories appear, which can be useful when you need to compare documents, policies, radio transcripts or social media posts. Some studies use both, reporting counts alongside rich quotations. Choose the method that matches your research question, not the one that looks most impressive.

Open-ended survey responses

Surveys frequently end with a box that reads “Any other comments?” In a large Kenyan household or school survey, that box can contain hundreds of short answers. They deserve systematic treatment. Clean the responses, remove personal details, develop a short codebook, code in a spreadsheet or in software, and then report categories with representative quotations. Short answers have less context than interviews, so claim less and stay close to the words.

Qualitative rigour does not come from software or from numbers of participants. It comes from transparent decisions. Keep an audit trail: dated notes on why codes were merged, split or dropped. Consider member checking where it is ethical and practical, and look for cases that challenge your emerging themes.

NVivo and ATLAS.ti support

Software does not analyse your data. It helps you store, organise, retrieve and visualise it. Two of the most widely used tools are NVivo and ATLAS.ti. Universities in Kenya and the wider region often provide licences or training for one or both, so confirm with your department before you purchase anything.

What NVivo support can cover

NVivo support typically includes setting up a project, importing transcripts, survey data and documents, creating a code structure, running queries and building matrices that show how codes appear across groups. Many students need help with practical problems: organising cases and attributes, merging work done by two coders, or exporting a codebook for an appendix.

What ATLAS.ti support can cover

ATLAS.ti support follows a similar pattern: document groups, quotations, code groups, memos and network views that show relationships between codes. Network views are especially popular for presenting theme maps, though you should explain every link in words rather than leaving the diagram to speak alone.

Support is not substitution

Legitimate software support means training, troubleshooting, project set up, quality checks and feedback on your coding scheme. It does not mean that someone else builds your themes and writes your discussion. If you want to learn the tool yourself, consider tutoring through the data analysis training and tutoring service, which is designed to build your own confidence.

Mixed methods research: bringing numbers and voices together

Mixed methods research combines quantitative and qualitative strands within one study. It is common in Kenyan public health, education and community development, where a survey can show how widespread a problem is and interviews can show why. The value lies in integration. Two separate studies placed in one document are not mixed methods unless the findings speak to each other.

ConvergentQUANQUALMERGEExplanatory sequentialQUANqualNumbers first, thenvoices explain themExploratory sequentialQUALquanVoices first, then ameasure is built and tested
Three core designs. Capital letters indicate the dominant strand, and arrows show the order of the work.

Convergent design

You collect both kinds of data at roughly the same time, analyse them separately and then merge the results to see where they agree or differ. A joint display table, with survey findings on one side and illustrative themes on the other, is a helpful way to show this.

Explanatory sequential design

You begin with a survey, notice something surprising, such as low uptake of a programme among one group, and then interview selected participants to explain it. Your sampling for the second phase should follow logically from the first.

Exploratory sequential design

You begin with interviews or focus groups to understand a poorly studied issue, then use the themes to build a questionnaire or scale, which you test with a larger sample. Instrument development needs careful pilot testing before use.

If your quantitative strand needs statistical support, the data analysis and statistics service covers that side, while the broader methodological discussion is covered in resources from SAGE Publishing, a major publisher of research methods texts.

Research ethics: consent, anonymisation and secure handling

Qualitative data is personal. People describe illness, grief, conflict, money and family life, often to a researcher they met a few days earlier. Ethical practice is therefore not an optional extra. In Kenya, most studies need approval from an institutional review board and, for many projects, a research permit from the national regulator. Follow the conditions of your approval at every stage, including when you involve outside help.

Consent

Informed consent should explain the purpose of the study, what participation involves, how recordings will be used, who may hear or read them, and the right to withdraw. If transcription or analysis support will involve a third party, your consent form should say so in plain language. Read the form aloud in the participant language where literacy is limited.

Anonymisation

Replace names with codes, and consider whether combinations of details could still identify someone. In a small village, a mention of the only secondary school headteacher is as identifying as a name. Keep the key linking codes to identities in a separate, protected location.

Secure handling of recordings and transcripts

  • Store recordings on encrypted or password protected devices, not on shared phones.
  • Share files only through secure channels and only with people who need them.
  • Remove direct identifiers before sending material for transcription where possible.
  • Agree in advance when files will be deleted, in line with your approval.
  • Never post excerpts on social media or in group chats, even without names.

For broader ethical guidance on research with human participants, the World Health Organization and the American Psychological Association both publish widely referenced standards that Kenyan researchers often cite alongside local requirements.

Presenting qualitative findings

Many strong analyses are weakened by thin presentation. Readers need to see the evidence, not only your conclusions. A reliable structure for each theme is: a clear name, a short statement of what the theme means, two or three well chosen quotations, and your interpretation linking the theme to the literature and to your research question.

Choosing quotations

Select quotations that illustrate rather than decorate. Avoid using the same articulate participant for every theme, and show variation, including views that disagree. Identify speakers by code, role or site, for example “Community health volunteer, Site B”, without revealing identity.

Visuals and tables

A theme map, a table of themes with sub themes and a joint display for mixed methods can help the reader follow your reasoning. Keep tables readable, label every figure and describe each one in the text. Your writing should explain what the reader is meant to notice.

Tie it to the whole thesis

Your findings must connect back to the objectives in Chapter One and forward to your discussion and recommendations. If you need structure feedback across the whole document, the thesis and dissertation research support service is the natural companion to analysis support.

How TaskLynk supports qualitative and mixed-methods work

TaskLynk is a managed platform. When you submit a request, a project manager reviews what you need, clarifies the scope and coordinates a vetted specialist with suitable experience. You are not left to negotiate with strangers, and the stages above are kept separate so that each one is scoped honestly.

You can explore the full range under qualitative and mixed-methods research support, covering transcription, coding assistance, codebook development, NVivo and ATLAS.ti help and guidance on presenting findings. Pricing depends on the volume and complexity of the work and is confirmed by quotation, so please see the pricing page for how that works, or contact the team with your questions.

What support can include

Transcription, applying your codebook, project set up, software training, methodology guidance and feedback on your drafts.

What stays with you

Your research questions, your interpretation, your conclusions and the final writing. These must be your own work.

A practical checklist before you begin

Before requesting help, gather a few things. Confirm your ethics approval and consent terms. Decide which languages and transcript style you need. Choose a software tool, or ask your supervisor which one is expected. Draft a preliminary codebook, even if it is rough, because it gives a specialist something concrete to work from. Finally, decide which stage you truly need: transcription alone, coding assistance, training so you can code yourself, or broader project guidance.

Qualitative research rewards patience. When voices are transcribed faithfully, coded transparently and interpreted honestly, the final themes carry real weight. You can read related guidance on the TaskLynk blog.

Frequently asked questions

What is the difference between transcription and coding?

Transcription converts audio or video into written text. Coding is an analytical step in which you label segments of that text with short descriptive tags. A transcript is raw material, while codes are the first layer of analysis.

Can someone else write my interpretation of the findings?

No. Interpretation belongs to the researcher. A specialist can explain methods, check that your themes are supported by the data and give feedback on your drafts, but you remain responsible for the conclusions and the writing.

Do you handle interviews held in Kiswahili or mixed with Sheng?

Many Kenyan interviews switch between English, Kiswahili, Sheng and mother tongues. When you request transcription, state the languages involved and whether you need a verbatim record, a translated version or both, so the right person can be matched to the work.

Which is better, NVivo or ATLAS.ti?

Both support rigorous qualitative work. The better choice is usually the one your supervisor, department or funder already uses, or the one whose interface suits your data. Support is available for either tool.

How is pricing decided?

Pricing depends on the number of recordings, their length, audio quality, languages and the type of support needed. TaskLynk confirms the cost through a quotation before work begins, so there are no surprises.

How are my recordings and transcripts kept safe?

Share only what is necessary, remove names and identifying details where possible, and use secure channels. Your consent forms should also explain who may access recordings. Raise any restrictions from your ethics approval when you submit the request.

Ready to turn your recordings into defensible themes?

Tell us which stage you need and a project manager will confirm the scope and quotation.

Qualitative and Mixed-Methods Research Support in Kenya | TaskLynk