How to Analyze Interview Data: From Transcript to Findings

Before you start: choose your analysis method

Qualitative interview data can be analysed in many ways, and the method you choose shapes your findings. The decision depends on your research question, your epistemological position and what your department expects. The table gives a quick orientation.

Running example: you conducted 12 semi-structured interviews with first-generation students about how they develop a sense of belonging in their first year at university. Each interview lasted 45 to 60 minutes.

Step 1: prepare your data

Analysis starts with transcription. Decide on a transcription style before you begin: verbatim (every word, including fillers and false starts) is standard for most methods; detailed transcription with pauses, overlaps and intonation is needed for narrative or conversation-oriented analysis; clean verbatim (fillers removed) can be acceptable for content-focused analysis. Document your transcription rules in the methods chapter.

Anonymise while transcribing: replace names with pseudonyms or IDs (P01 to P12), and remove or generalise places, course names and other identifying details. Keep the key file that links IDs to real names separate and secure, according to your ethics approval and data protection rules.

• Use consistent participant IDs or pseudonyms throughout
• Add line numbers or timestamps so you can reference quotes precisely
• Check each automated transcript against the audio; speech recognition makes systematic errors with accents and technical terms
• Store a short case summary for each interview (context, mood, key points)

Step 2: familiarise yourself with the data

Read every transcript at least twice, ideally while listening to the recording once. Make notes in the margin: what surprised you, what seems important, what contradicts your expectations? Write a case summary of half a page for each participant. These summaries help you keep individual stories in view later, when you work with codes across interviews.

This is also the moment to start a reflexive journal. If you are a first-generation student yourself, note how this might influence what you notice and what you overlook.

Step 3: code systematically

Work through every transcript and label segments that are relevant to your research question. Depending on your method, you start with predefined codes (deductive), develop them from the data (inductive) or combine both. Use one or two first cycle methods such as descriptive, in vivo or process coding (Saldaña, 2021), and write memos about interesting ideas.

Code all interviews with the same attention, including the less articulate ones. After the first round, revise your code list, merge duplicates and recode. If your method uses a codebook, define each code with inclusion and exclusion criteria and an example.

• Code meaningful segments (usually one to five sentences) with enough context
• Allow multiple codes per segment where appropriate
• Keep a codebook or code list with short definitions
• Mark particularly expressive quotes for later use

Step 4: develop categories or themes

In a second cycle, you organise codes into higher-level units: categories in content analysis and grounded theory, themes in thematic analysis, experiential themes in IPA. Group codes that share meaning, look for relationships between groups, and draw a map or diagram. A code matrix (participants in rows, codes in columns) shows at a glance whether a pattern appears across the sample or only in a few interviews. Our free coding tool at /codieren generates such a matrix and lets you export coded segments for each code.

Check each theme or category against the data: do the coded extracts actually support it? Are there negative cases that contradict it? Contradictions are not problems to hide; they often lead to the most interesting findings.

Step 5: check quality and trustworthiness

Qualitative research is judged by criteria such as credibility, dependability, confirmability and transferability (often summarised as trustworthiness). For interview analysis, typical strategies include an audit trail of analytic decisions, memos, peer debriefing with fellow students or your supervisor, thick description of context, reflexivity and, for codebook-based methods, a second coder. Member checking (asking participants for feedback) is possible but debated, because participants may not recognise themselves in an interpretive analysis.

Choose strategies that fit your method. A kappa value suits qualitative content analysis but not reflexive thematic analysis. Your supervisor or department may require a particular approach: follow your university’s guidelines (you can upload them to Professor Digital).

Step 6: write up your findings

Structure the findings chapter around your themes or categories, not around your interview questions or participants (unless your method is case-based, as in narrative analysis). For each theme, state the analytic claim, illustrate it with carefully selected quotes, and interpret what the quotes show. Use participant IDs or pseudonyms after each quote.

Keep findings and discussion apart unless your department allows a combined chapter: in the findings, you present what you found; in the discussion, you relate it to the literature and your research question. A realistic timeline for 12 interviews is shown below.