Qualitative Data Analysis
Session at a Glance
Thematic analysis (Braun & Clarke's 6 phases); inductive vs. deductive coding; content analysis; CAQDAS tools; writing qualitative findings with evidence; trustworthiness in analysis
Hands-on coding of interview transcripts; developing themes from codes; using NVivo/ATLAS.ti or manual coding; writing a thematic findings section with participant quotes
2 hrs Lecture + 12 hrs Lab/Project
Progress Report 1 — methodology execution status, preliminary findings, revised timeline
Learning Objectives
By the end of this session, you will be able to:
- Execute Braun and Clarke's (2006) six phases of reflexive thematic analysis — from data familiarisation through theme development to writing the analytic narrative — on your own qualitative data
- Apply systematic coding procedures — including inductive (data-driven) and deductive (theory-driven) approaches, open coding, and the organisation of codes into a coherent coding framework
- Distinguish between thematic analysis and content analysis, selecting the appropriate approach based on whether your RQs require interpretive depth (thematic) or systematic quantification of manifest content (content analysis)
- Use CAQDAS software (NVivo, ATLAS.ti, or free alternatives) to manage, code, and retrieve qualitative data — while maintaining a critical understanding of what the software can and cannot do
- Write a qualitative findings section that presents themes in a logically structured narrative, supports each theme with rich participant evidence, and connects findings back to research questions and the existing literature
Session Planner
Suggested breakdown of the 4-hour contact session.
| Time | Segment | Activity | Mode |
|---|---|---|---|
| 0:00–0:08 | Opening | Recap Weeks 13–14; transition: "Quantitative analysis asks 'how much?' Qualitative analysis asks 'what does it mean?' Both require systematic procedures. This week: the systematic procedures for making meaning from words." | Whole class |
| 0:08–0:30 | Lecture 1 | Thematic analysis — Braun & Clarke's 6 phases in depth; inductive vs. deductive coding; from codes to themes; the coding process demonstrated on sample transcript data | Lecture + Demo |
| 0:30–0:50 | Lecture 2 | Content analysis; qualitative vs. quantitative content analysis; developing a coding frame; inter-coder reliability. CAQDAS — what it does (store, code, retrieve, visualise) and what it doesn't do (think for you). | Lecture |
| 0:50–1:05 | Activity | Code a sample transcript excerpt individually, then compare codes with a partner. Discuss: where did your codes converge? Where did they differ? What does this reveal about the interpretive nature of coding? | Pairs |
| 1:05–1:15 | Discussion | Debrief the coding exercise. Discuss inter-coder agreement — when is divergence a problem (content analysis) vs. a productive interpretive resource (thematic analysis)? | Whole class |
| 1:15–1:30 | Break | — | — |
| 1:30–1:55 | Lecture 3 | Writing qualitative findings — the thematic narrative; selecting and integrating participant quotes; connecting themes to RQs; establishing trustworthiness in the analysis process; the thematic map as a writing tool | Lecture |
| 1:55–2:05 | Progress Report Briefing | Progress Report 1 requirements: methodology execution status, data collected to date, preliminary findings (even if partial), challenges encountered, revised timeline. Template walkthrough. | Whole class |
| 2:05–3:40 | Lab Work | Part A: Code one full transcript and develop initial themes; Part B: Construct a thematic map; Part C: Write one thematic subsection with participant quotes | Individual |
| 3:40–3:55 | Peer Review | Exchange coded transcript or thematic map; peer checks for: are codes grounded in data? Do themes logically group codes? Are quotes compelling? | Pairs |
| 3:55–4:00 | Exit Ticket | Submit coded transcript excerpt and preliminary thematic map | Individual |
1. The Qualitative Analysis Landscape — From Raw Data to Findings
Qualitative data analysis is the process of transforming raw data — interview transcripts, field notes, documents, images, audio recordings — into coherent, evidence-based findings that answer your research questions. Unlike quantitative analysis, which follows a relatively linear path (screen data → run test → interpret output), qualitative analysis is iterative and recursive: you move back and forth between data, codes, themes, and interpretations, refining your understanding with each pass. The goal is not to "extract" meaning that exists objectively in the data but to construct a rigorous, defensible, and insightful interpretation grounded in systematic procedures.
Qualitative data analysis is the systematic process of organising, categorising, and interpreting non-numerical data to identify patterns, themes, and relationships that address the research questions. It involves: (a) data management — organising and preparing data for analysis; (b) coding — labelling segments of data with interpretive tags; (c) theme development — clustering codes into broader patterns that capture something meaningful about the data in relation to the RQs; and (d) interpretation — constructing a coherent, evidence-based narrative that goes beyond description to offer insight. The researcher is the primary analytical instrument — software supports analysis but does not perform it.
1.1 Approaches to Qualitative Analysis — Choosing Your Method
| Approach | What It Does | Best For | Output |
|---|---|---|---|
| Reflexive Thematic Analysis (Braun & Clarke) | Identifies, analyses, and reports patterns (themes) across a qualitative dataset. Themes are actively constructed by the researcher through engagement with the data — not "discovered" or "emerging." | Understanding participants' experiences, perspectives, and meanings; identifying shared and divergent patterns across cases; RQs about "how" and "what" rather than theory-building. The most widely used approach in capstone research. | A set of themes, each with a clear central organising concept, supported by data extracts, woven into an analytic narrative that addresses the RQs |
| Grounded Theory Coding (Charmaz; Strauss & Corbin) | Builds theory from data through systematic coding: open (initial), axial (relating categories), and selective (integrating around a core category). Constant comparison throughout. | When your RQs ask about processes and you aim to generate theory; when existing theories are inadequate for your context (Week 10). Distinct from thematic analysis in its commitment to theory generation. | A substantive theory — a set of interrelated concepts with specified relationships, grounded in the data |
| Qualitative Content Analysis | Systematically categorises textual data using a coding frame, which may be developed inductively (from data) or deductively (from theory). More structured and less interpretive than thematic analysis. | When you need to systematically describe the manifest and/or latent content of textual data; when a structured coding frame is appropriate; when inter-coder reliability is important (multiple coders). | A structured coding frame with category definitions, frequencies, and relationships; may include quantification (e.g., "65% of responses referenced cost as a barrier") |
| Framework Analysis (Ritchie & Spencer) | A matrix-based approach: rows = cases, columns = themes. Systematically summarises data within a thematic framework. Developed for applied policy research. | When you have specific questions, a limited time frame, and a need for transparent, systematic analysis that can be reviewed by others; when comparing across cases systematically. | A thematic framework matrix with summarised data, enabling systematic cross-case comparison |
The most common qualitative analysis error: applying thematic analysis procedures (Braun & Clarke) but claiming grounded theory, or vice versa. If your methodology chapter says "this study uses grounded theory," your analysis must follow grounded theory coding procedures (open → axial → selective, constant comparison, theoretical sampling, memo-writing, theoretical saturation). If your methodology says "phenomenology," your analysis should follow phenomenological analytical procedures (e.g., Moustakas's modified van Kaam method, or Interpretative Phenomenological Analysis). The analysis method is not a free-floating choice — it is determined by your qualitative tradition (Week 10). If you have forgotten which tradition you committed to, revisit your methodology chapter before proceeding with analysis.
2. Reflexive Thematic Analysis — Braun & Clarke's Six Phases
Braun and Clarke's (2006, 2019, 2022) reflexive thematic analysis (TA) is the most widely used qualitative analytic method in student research — and the most widely misunderstood. It is not a mechanical procedure of "finding themes" that exist in the data waiting to be uncovered. It is an active, reflexive process in which the researcher constructs themes through deep engagement with the data, informed by their theoretical assumptions, research questions, and interpretive judgement.
2.1 The Six Phases — Not a Linear Recipe
What you do: Immerse yourself in the data. Read and re-read each transcript at least twice. Listen to audio recordings while reading transcripts. Make notes on initial impressions, interesting observations, potential patterns. Output: Familiarisation notes — informal, exploratory jottings about what strikes you, puzzles you, or seems significant. Common error: Skipping this phase and jumping straight to coding. You cannot code data you don't know intimately. Familiarisation is analysis, not procrastination.
What you do: Systematically work through each transcript, labelling segments of data that are potentially relevant to your RQs. A code is a short label (a word or brief phrase) that captures the essence of a data segment. Code inclusively — when in doubt, code it. You can discard codes later; you cannot recover uncoded data. Code the entire dataset before moving to theme development. Output: A list of codes (30–80 is typical for a capstone with 12–20 interviews), each with associated data extracts. Common error: Coding at too high a level of abstraction (coding for themes rather than for specific meanings), producing thin, generic codes that could apply to any dataset.
What you do: Sort codes into potential themes. A theme is NOT a topic summary ("Barriers to adoption") — it is a pattern of shared meaning, organised around a central organising concept. "Invisible gatekeepers: how informal power structures in family businesses block technology adoption" is a theme. "Technology adoption barriers" is a topic. Output: A collection of candidate themes, each with its constituent codes and associated data extracts. A thematic map (visual diagram of themes and their relationships). Common error: Creating topic-summary themes that describe what participants talked about rather than capturing what the data MEANS.
What you do: Two-level review. Level 1: Re-read all extracts coded for each theme — do they form a coherent pattern? If not, revise the theme boundaries. Level 2: Re-read the entire dataset — do the themes capture the dataset as a whole? Are there data segments that contradict or complicate the themes? Output: A refined thematic map with clear theme definitions. Some candidate themes may be discarded, merged, or split. Common error: The MECE trap — themes don't need to be perfectly mutually exclusive. Some codes legitimately belong to multiple themes. Forcing perfect separation loses nuance.
What you do: For each theme, write a detailed analysis: what is the theme about? What is its central organising concept? What is its scope and boundaries? How does it relate to the RQs? Name each theme — the name should be concise, evocative, and immediately communicate the theme's essence. Output: A written definition for each theme (2–3 sentences), with a clear name. Common error: Theme names that are single words or generic phrases ("Trust," "Barriers," "Facilitators"). These signal topic-summary themes, not analytic themes.
What you do: Write the findings section. For each theme: introduce it (what is this theme about?), present the evidence (participant quotes — the data that grounds the theme), and provide your analytic commentary (what does this evidence mean? how does it connect to the RQs? what insight does it offer?). Output: The findings chapter. Common error: The quote-heavy narrative — a series of participant quotes with minimal analytic commentary. The researcher's voice must be present. Quotes are evidence, not analysis. Your commentary IS the analysis.
2.2 Inductive vs. Deductive Coding
| Dimension | Inductive (Data-Driven) Coding | Deductive (Theory-Driven) Coding |
|---|---|---|
| Starting Point | The data — codes emerge from engagement with the data without a pre-existing coding framework | A pre-specified coding framework derived from theory, prior research, or your conceptual framework |
| What You Do | Read the data line by line. Label segments with codes that capture what participants are saying. Let the data suggest the codes. | Start with a list of codes derived from your theoretical constructs. Search for data that relates to each code. Add new codes for data that doesn't fit. |
| When to Use | Exploratory RQs; when you want to foreground participants' perspectives rather than imposing theoretical categories; the default for reflexive TA | When you have a strong theoretical framework and want to map data onto existing constructs; when you need to test or extend theory; useful for content analysis and framework analysis |
| Typical Capstone Approach | Most capstone TA uses a primarily inductive approach — codes are generated from the data. This is consistent with the exploratory nature of many qualitative RQs. | Hybrid approaches are common: start with a few deductive codes from theory, then inductively code data that doesn't fit. Be transparent about which codes are deductive and which are inductive. |
Braun and Clarke are emphatic: themes do not "emerge from the data" like mushrooms after rain. This language suggests the researcher is passive and the themes are objectively present in the data. In reality, themes are constructed through the researcher's active engagement — their interpretive choices, their theoretical lens, their research questions, and their analytical skill. A different researcher with the same data might construct different (equally valid) themes. This is not a weakness of qualitative analysis — it is its epistemological premise. What matters is that you can show HOW you arrived at your themes through systematic, transparent procedures, and that your themes are grounded in the data (credibility), not that your themes are the "only possible" interpretation.
3. Coding — The Engine of Qualitative Analysis
Coding is the bridge between raw data and findings. A code is a label assigned to a segment of data that captures its essential meaning in relation to your research questions. Codes are the building blocks from which themes are constructed. Good coding is specific, grounded in the data, and analytically productive. Poor coding is vague, overgeneral, or merely describes the topic of the data segment rather than its meaning.
3.1 What Good Coding Looks Like
"When I joined, there was no onboarding. They gave me a laptop and said 'figure it out.' I spent the first two weeks just trying to understand who does what. It was embarrassing to keep asking people basic questions. I almost quit in the first month. The work itself was interesting, but I felt completely lost. One senior developer — not my manager, just someone sitting nearby — started checking in on me. He'd ask how I was doing, explain things I was stuck on. That one person is the reason I stayed."
| Data Segment | Poor Code (Topic Summary) | Good Code (Interpretive) | Why the Good Code is Better |
|---|---|---|---|
| "no onboarding...gave me a laptop and said figure it out" | "Onboarding" | "Sink-or-swim induction" | Captures the experience of abandonment, not just the HR process. Conveys the emotional tone. |
| "embarrassing to keep asking people basic questions" | "Asking for help" | "Shame of visible incompetence" | Captures the emotional experience and the mechanism (public nature of help-seeking). |
| "I almost quit in the first month" | "Turnover intention" | "Precarious early commitment" | Captures the fragility of retention during early days — not just the thought of leaving but its temporal location. |
| "One senior developer...started checking in...That one person is the reason I stayed." | "Mentoring" | "Anchoring through informal mentorship" | Captures both the mechanism (informal, not organisational) and its function (anchoring the new hire to the organisation). |
3.2 Practical Coding Guidelines
- Code inclusively. When deciding whether to code a segment, err on the side of coding. You can always discard codes later. A segment that seems marginal on first reading may become significant as themes develop.
- Code at the right grain size. A code that applies to five entire pages is too broad. A code that applies to three words is too narrow. A good code typically applies to a sentence, a short paragraph, or a brief exchange — a coherent "chunk" of meaning.
- Use gerunds where possible (Charmaz's advice). "Negotiating legitimacy" rather than "Legitimacy." "Calculating risk" rather than "Risk perception." Gerunds capture process and action.
- Keep a codebook. As you code, maintain a list of codes with brief definitions. This ensures consistency: when you encounter a segment similar to one you coded 10 transcripts ago, you code it the same way. A codebook also supports dependability (Week 10 trustworthiness).
- Code for what IS in the data, not what you expected. Resist the urge to code only data that confirms your expectations. The most analytically productive codes often come from data that surprised you, contradicted your assumptions, or didn't fit your initial framework.
- Single vs. multiple coding. The same data segment can receive multiple codes if it contains multiple meanings. This is normal and expected — it is why qualitative analysis cannot be reduced to counting codes.
4. CAQDAS Tools & Content Analysis
4.1 CAQDAS — What Software Can and Cannot Do
CAQDAS (Computer-Assisted Qualitative Data Analysis Software) — NVivo, ATLAS.ti, MAXQDA, Dedoose, and the free alternative Taguette — are tools for managing, organising, coding, and retrieving qualitative data. They do not perform analysis. They are to qualitative research what SPSS is to quantitative research: they handle the mechanical tasks (storing data, applying labels, retrieving coded segments, visualising relationships), freeing the researcher to do the intellectual work (interpreting meaning, constructing themes, building arguments).
| What CAQDAS DOES | What CAQDAS DOES NOT DO |
|---|---|
| Store and organise all your data files (transcripts, audio, images, documents) in one project | Read your data for you. You must still do the deep, repeated engagement that constitutes familiarisation. |
| Enable systematic coding — create codes, apply them to data segments, retrieve all segments coded with a particular code | Generate codes from your data. The researcher decides what to code, what to name the code, and what the code means. |
| Organise codes hierarchically (parent/child codes) and visually (code maps, network views) | Identify themes. The researcher clusters codes, identifies patterns, and constructs themes — software visualises the relationships you define. |
| Run queries — find all segments coded with Code A AND Code B; find co-occurrence patterns; generate code frequency matrices | Interpret what the query results mean. A co-occurrence matrix shows which codes appear together — it does not tell you WHY or what that MEANS. |
| Track your coding process — provide evidence of systematic engagement with the data, supporting dependability and confirmability | Guarantee trustworthiness. Trustworthiness comes from your procedures and transparency, not from the software brand. |
4.2 Content Analysis — Systematic Categorisation of Text
Content analysis (CA) is a more structured approach than thematic analysis, focused on systematically categorising textual data using a coding frame. In its qualitative form, CA focuses on interpreting the meaning of categories while maintaining systematic, transparent procedures. In its quantitative form, CA counts frequencies of categories and enables statistical analysis of coded data.
| Dimension | Thematic Analysis | Qualitative Content Analysis |
|---|---|---|
| Goal | Rich, interpretive understanding of patterns of meaning across a dataset | Systematic description of textual data organised into a structured coding frame |
| Coding Frame | Evolved iteratively; codes are generated from the data and refined throughout analysis | May be developed inductively or deductively; typically more structured, with clear category definitions and decision rules |
| Researcher Role | Actively constructs themes; subjectivity is a resource, not a problem to be eliminated | Systematic application of the coding frame; inter-coder reliability is often used to establish consistency |
| Output | Themes with narrative interpretation; participant quotes woven into an analytic narrative | Coding frame with category definitions; may include frequency counts; typically presented in a more structured format |
| When to Choose | RQs are exploratory; depth of interpretation is prioritised; the researcher is comfortable with the interpretive nature of TA | RQ requires systematic categorisation; multiple coders are involved and inter-coder reliability is desired; the research has a more applied or structured orientation |
5. Writing Qualitative Findings — Evidence, Analysis, Narrative
The qualitative findings chapter is where your analytical work becomes visible to the reader. It is not a data dump of participant quotes organised by theme. It is a curated, argued, and evidenced narrative in which your interpretive voice guides the reader through what you found, supported by the strongest evidence the data provides.
5.1 The Anatomy of a Thematic Subsection
1. Theme Introduction (2–3 sentences): Name the theme. State its central organising concept. Briefly indicate what it captures about the data in relation to the RQs. "Theme 2: Performing Competence — The Emotional Labour of Appearing Capable. This theme captures participants' descriptions of the effort required to manage others' perceptions of their competence, particularly in the early months of employment. Rather than simply learning job skills, participants described actively performing confidence and capability while concealing uncertainty."
2. Evidence Block (1–2 participant quotes): Present the strongest, most compelling quotes that illustrate the theme. Introduce each quote: who is speaking (pseudonym, relevant context)? "Priya, a software developer with three years of experience, described the gap between her internal state and external presentation:" [quote].
3. Analytic Commentary (3–5 sentences): This is YOUR voice. Interpret the evidence. Connect it to the theme's central concept. Note what is significant about this data. Connect to the RQs. Compare across participants (convergence and divergence). "Priya's account foregrounds the performative dimension of newcomer competence — the effort is directed not at learning per se but at managing others' impressions of her learning. This pattern recurred across participants, though the costs of the performance differed: for those with prior experience (like Priya), the performance felt manageable; for those entering a new industry, the performance was described as 'exhausting' (Ravi) and 'unsustainable' (Ananya, Farah). This divergence suggests that the emotional burden of performing competence is moderated by the degree of domain familiarity."
5.2 Selecting and Using Participant Quotes
| Guideline | What This Means |
|---|---|
| Quotes are evidence, not analysis. | A quote illustrates your theme; it does not replace your interpretive commentary. If a findings section has more participant words than researcher words, it is under-analysed. |
| Select the most compelling quotes, not all quotes. | You may have 15 quotes supporting a theme. Present the 2–4 strongest ones. The reader trusts that you have more evidence than you are showing — you can state this: "This pattern was evident across 12 of the 15 participants." |
| Use short quotes woven into your narrative. | Long block quotes (5+ lines) should be rare and reserved for particularly rich data. Short quotes integrated into your sentences keep the reader engaged and foreground YOUR voice. "Participants described the onboarding process as 'chaotic' (Arjun), 'nonexistent' (Meera), and in one case, 'designed by someone who had clearly never been a new employee' (Vikram)." |
| Attribute every quote. | Use pseudonyms. Provide relevant context: role, experience, organisation type. "Rahul, a product manager at a Series A startup, explained..." Consistent attribution enables the reader to track individual voices across themes. |
| Show divergence, not just convergence. | If 12 participants expressed Pattern A and 3 expressed Pattern B, show both. The minority voice is often analytically valuable — it reveals boundary conditions, contextual factors, or alternative experiences. "While most participants described the transition as challenging but growth-promoting, three participants — all from non-engineering backgrounds — described it as alienating and career-damaging." |
5.3 The Thematic Map as a Writing Tool
Before writing, construct a thematic map — a visual diagram showing your themes and their relationships. The map serves as the outline for your findings chapter. Each theme becomes a section. The spatial arrangement (which themes are adjacent, which are overarching, which are subordinate) suggests the logical flow of your narrative.
The worst qualitative findings chapters read: "Theme 1: Barriers to Adoption. Participants identified several barriers to adoption. Ravi said [quote]. Priya said [quote]. Arjun said [quote]. Theme 2: Facilitators of Adoption. Participants also identified facilitators..." This is a catalogue of themes with supporting quotes. A good findings chapter makes an argument: it builds a case that the themes collectively reveal something significant about the phenomenon, answers the RQs, and contributes insight. Each theme section advances the argument. The reader finishes the chapter understanding not just WHAT you found but WHY it matters. This is the difference between descriptive thematic analysis (competent but boring) and interpretive thematic analysis (the standard to which you should aspire).
6. Progress Report 1 — Documenting Your Capstone Journey
Progress Report 1 is a structured checkpoint that documents where you are in your capstone execution. It serves three purposes: (a) accountability — ensuring you are on track, (b) early intervention — identifying problems while there is still time to address them, and (c) reflection — creating space to evaluate what is working and what needs to change.
| Section | Content | Key Question to Answer |
|---|---|---|
| 1. Methodology Execution Status | What have you DONE? Review each element of your methodology: data collection status (how many surveys/interviews/experiments completed vs. planned), instrument deployment, any deviations from the approved proposal with justification | "Is my data collection on track? If not, what specifically is delayed, and what is my plan to recover?" |
| 2. Data Collected to Date | Describe your dataset: sample characteristics (demographics, relevant descriptors), data quality (completeness, any issues encountered), preliminary assessment of whether the data can answer your RQs | "Does the data I have collected look capable of answering my RQs? Are there gaps or quality issues I need to address?" |
| 3. Preliminary Findings | Even if analysis is incomplete, present what you can see so far. For quantitative: descriptive statistics, preliminary correlations, initial model results. For qualitative: initial codes, candidate themes, interesting patterns. Be honest about how preliminary these are. | "What is the data telling me so far? What patterns are emerging? What surprises or puzzles have I encountered?" |
| 4. Challenges and Solutions | Identify the 2–3 most significant challenges you have encountered. For each: what was the challenge, how did it affect your research, what did you do about it, and what was the outcome? | "What has gone wrong, and what have I done about it? Are there unresolved challenges that I need help with?" |
| 5. Revised Timeline | Present your original timeline alongside your actual progress. Identify any milestones that have slipped. Provide a realistic revised timeline for the remaining weeks. | "Can I still complete this capstone to quality within the remaining time? If not, what scope adjustments or support do I need?" |
Think Deeper — Cross Questions
Discuss in pairs before sharing with the class.
You and a peer independently code the same 3-page interview transcript. When you compare, you find that 40% of your codes are different — you coded segments they didn't, and vice versa, and some similar segments received different code labels. Is this a problem? How does the answer differ for thematic analysis (where coding is interpretive and the researcher is the instrument) vs. content analysis (where inter-coder reliability is a quality criterion)? What does this exercise reveal about the nature of qualitative coding?
A student presents their thematic map with six themes: "Communication," "Training," "Support," "Workload," "Technology," and "Culture." Their supervisor comments: "These are topics, not themes." The student replies: "But my participants talked about all of these things. They ARE the themes in my data." Who is right? What is the difference between a topic and a theme? Rewrite two of these topics as genuine themes by giving them a central organising concept.
You have coded 18 interview transcripts and identified 4 main themes. As you write your findings, you notice that one theme is supported almost entirely by quotes from senior-level participants, while frontline employees rarely mentioned it. Another theme is dominated by male participants' accounts. How should you handle this in your writing? Is it legitimate to present a theme that characterises only a subset of your sample? What reporting practices would ensure transparency about whose voices are represented in each theme?
Your Progress Report 1 reveals that you have completed only 6 of your planned 20 interviews by Week 15 — you are significantly behind schedule due to difficulties recruiting participants. Your original timeline allocated 4 weeks for data collection. What do you do? Consider: (a) extending the data collection timeline (what gets compressed downstream?), (b) reducing the sample size (can saturation be achieved with fewer interviews?), (c) changing the sampling strategy (is there a more accessible population?), and (d) adjusting the scope of the RQs. Which option would you recommend, and why?
Quick Check — Coding and Theme Diagnosis
Diagnose the problem with each element of qualitative analysis.
1. A student's code list for 15 interview transcripts contains 12 codes: "Leadership," "Communication," "Motivation," "Culture," "Change," "Technology," "Training," "Feedback," "Teamwork," "Trust," "Stress," and "Growth."
2. A findings chapter section: "Theme 3: Trust. Participants talked about trust in various ways. Ravi said, 'I trust my manager because she supports me.' Priya said, 'Trust is essential in any team.' Arjun said, 'You need trust to collaborate effectively.' Overall, trust was an important theme in the data."
3. A student writes in their methodology: "Themes emerged from the data through a rigorous process of thematic analysis." The findings chapter describes themes as if they objectively exist in the data. No mention is made of the researcher's role, theoretical position, or interpretive choices.
4. A BCA student analysing developer interviews reports: "I used NVivo to analyse my data. The software identified 6 key themes from the 12 interview transcripts."
Knowledge Check — Interactive Quiz
Test your understanding of qualitative data analysis.
Q1. In Braun and Clarke's reflexive thematic analysis, what is a theme?
Q2. Which of the following is true about CAQDAS software (NVivo, ATLAS.ti)?
Q3. In the six phases of thematic analysis, "searching for themes" (Phase 3) involves:
Q4. What is the primary difference between a topic-summary theme and an analytic theme?
Q5. In a qualitative findings chapter, what is the ideal balance between participant quotes and researcher commentary?
Lab Activity — Coding and Theme Development
Part A: Code One Full Transcript (60 min)
- Select one of your interview transcripts (or use the provided practice transcript).
- Read it twice for familiarisation. Make familiarisation notes: initial impressions, interesting passages, potential patterns.
- Code the transcript systematically, line by line or paragraph by paragraph. Aim for 25–50 codes for a typical 8–12 page transcript.
- Maintain a codebook: list each code with a brief definition and an example data segment.
| Code | Definition | Example Data Segment | Notes |
|---|---|---|---|
Part B: From Codes to Themes (45 min)
- Transfer all your codes onto sticky notes, index cards, or a digital canvas.
- Sort codes into clusters based on shared meaning. Ask: "Which codes seem to be about the same underlying thing?"
- For each cluster: articulate the central organising concept — what holds these codes together? Write a one-sentence theme definition.
- Draw a thematic map showing your candidate themes and their relationships.
- Review Phase 4 style: re-read the coded transcript. Do your themes capture the data? Are there data segments that don't fit any theme — and what does that tell you?
Part C: Write One Thematic Subsection (45 min)
Select one theme. Write a 400–600 word thematic subsection following the anatomy in Section 5.1: theme introduction → evidence (2–3 quotes) → analytic commentary. Submit for peer review. Peer checks: (a) is the analyst's voice dominant, not the participants'? (b) are quotes integrated into the narrative or just dropped in? (c) does the commentary interpret rather than paraphrase the quotes?
Exit Ticket
Submit with your coded transcript and draft thematic subsection.
- Submit your preliminary thematic map (even if incomplete). How many candidate themes do you have?
- How many codes did you generate from your first transcript? Do your codes feel overly topic-level, or are they capturing meaning?
- What was the most difficult part of coding — and what does this reveal about your analytical approach?
- For Progress Report 1: What is the single most significant challenge you have encountered in your capstone execution so far? What is your plan to address it?
- On a scale of 1–10: How confident are you that your data (collected or planned) will enable you to produce a compelling findings chapter?
Key Takeaways — Week 15
Themes do not "emerge" from data. They are actively constructed through the researcher's systematic engagement, interpretive choices, and analytical skill. Acknowledge your role in the construction of themes — this is reflexivity, not a weakness.
Coding is not data management — it is analysis. Every code is an interpretive choice. Good codes capture meaning, not topics. They use gerunds. They are specific to your data, not generic labels that could apply to any study. Coding well is the foundation of everything that follows.
A findings chapter with more participant words than researcher words is under-analysed. Quotes are evidence, not analysis. Your commentary — interpreting, connecting, arguing — is the analysis. The reader should hear your voice leading the narrative, supported by participant voices as evidence.
CAQDAS handles the mechanical tasks of qualitative analysis — storage, coding, retrieval, visualisation. It does not read, interpret, or think. Never write "NVivo identified themes." NVivo retrieved the coded segments you told it to retrieve. You identified the themes.
Facilitator Notes
Preparation Checklist
- Prepare 2–3 sample interview transcripts (4–6 pages each) for coding practice — one BBA domain, one BCA domain. Annotate one with "model" coding to show students what good coding looks like after they have attempted their own.
- Prepare a "bad" thematic map (topic-summary themes, vague labels, no central organising concepts) and a "good" thematic map (analytic theme names, clear hierarchy, visible relationships). Show both during the lecture — the contrast is instructive.
- Have NVivo, ATLAS.ti, or Taguette (free) installed and ready for demo. Prepare a project with pre-loaded transcripts so students can see a working project rather than an empty interface.
- Prepare the Progress Report 1 template as a downloadable document. Include section prompts and word count targets. Share exemplar progress reports from previous cohorts (anonymised) — one that identifies problems honestly (good) and one that is overly optimistic (instructive).
Common Student Difficulties
- Topic-summary coding and theming: The single most pervasive problem. Students produce codes that are topic labels and themes that are topic summaries. The coding exercise (Part A) with peer comparison, followed by the contrast between good and bad thematic maps, is designed to make this distinction visible.
- Under-coding: Students generate 15–20 codes for a full transcript because they are coding at too high a level of abstraction. Push them to code more granularly — a 10-page transcript should generate 30–50+ codes on first pass.
- Quoting without commentary: The findings chapter draft reveals this immediately. If a student's thematic subsection has long block quotes followed by "This shows that trust is important," they haven't yet understood the difference between evidence and analysis.
- Progress report denial: Students who are behind schedule may present an overly optimistic progress report rather than honestly identifying problems. Emphasise that the progress report's purpose is early intervention — hiding problems now makes them worse later. Modelling honest reporting (showing an exemplar that says "I am 4 weeks behind on data collection; here is my recovery plan") helps normalise this.
Pacing Tips
- The coding exercise (Activity, 15 min) is the critical pedagogical moment. Students discover through practice that coding is interpretive, not mechanical. The peer comparison — where codes differ despite the same data — is the most powerful demonstration of this. Do not skip or shorten this exercise.
- Students doing purely quantitative capstones still benefit from understanding how qualitative analysis works — it is part of research literacy and essential for reading and citing qualitative literature. They can spend the lab time on Progress Report 1 rather than coding.
- Week 15 marks the approximate halfway point of data collection and analysis (Weeks 13–20). The Progress Report 1 deadline creates a natural checkpoint. Follow up individually with students whose progress reports indicate significant delays — a 15-minute troubleshooting conversation now is worth hours of panic later.