How to Code and Theme Qualitative Interview Data

Thematic analysis is the most common way to make sense of interview data — and the most commonly done badly. Here is the disciplined version that holds up to review.

Codes, themes, and why the difference matters

A code is a short label attached to a segment of data that captures something meaningful about it (‘fear of automation’, ‘peer support’). A theme is a broader pattern that organises a group of codes around a central idea. Beginners often present a list of codes and call them themes; reviewers notice. A theme has a story — it says something about the data, not just names a topic.

The six phases of thematic analysis

Braun and Clarke's six-phase approach is the most widely cited framework, and following it visibly signals rigour:

• 1. Familiarisation. Read every transcript at least twice before coding. Note first impressions; do not code yet.
• 2. Generating initial codes. Work through the data systematically, labelling every segment relevant to your question. Code generously — you can merge later.
• 3. Searching for themes. Cluster related codes into candidate themes. This is interpretive work, not sorting.
• 4. Reviewing themes. Check each theme against its coded extracts and against the whole dataset. Split, merge, or discard themes that don't hold.
• 5. Defining and naming themes. Write a short definition of each theme — if you can't define it in two sentences, it isn't coherent yet.
• 6. Producing the report. Weave themes, your interpretation, and illustrative quotes into an argument.

Inductive or deductive coding?

Decide your stance up front. Inductive coding lets codes emerge from the data — suited to exploratory work. Deductive coding applies a pre-set framework from theory or prior research — suited to confirmatory or comparative work. Many studies are hybrid: a deductive skeleton with room for inductive surprises. State which you used; reviewers will ask.

Choosing quotes that earn their place

Quotes are evidence, not decoration. A good illustrative quote is vivid, clearly tied to the theme, and representative of more than one participant's view. Attribute each to an anonymised participant, and avoid building a theme on a single striking quote from one person — that is an anecdote, not a pattern.

Demonstrating trustworthiness

Qualitative rigour is judged by four criteria (Lincoln & Guba). Build evidence for each:

• Credibility — member checking, triangulation across sources, time in the data.
• Transferability — thick description of context so readers can judge what transfers.
• Dependability — a clear, documented audit trail of analytic decisions.
• Confirmability — reflexivity about how your own position shaped interpretation.

If two coders worked independently, report how you handled disagreement (discussion to consensus is usually more meaningful than a kappa coefficient for interpretive work).

Tools and export

Software does not analyse for you, but it makes coding, retrieval and audit far easier. Interview Insight Studio lets you code and theme in the browser and export to NVivo, ATLAS.ti, MAXQDA and Dedoose, so you can start free and move to institutional software later without re-coding. Pair it with the Transcribe tool to get from recording to coded transcript in one workflow.

Final thought

Good qualitative analysis is transparent about how it got from raw talk to themes. Follow a recognised method, keep an audit trail, choose quotes as evidence, and show your trustworthiness work. Do that and your themes carry the authority of method, not just the appeal of good quotes.