Qualitative Data Analysis: Themes, Coding & Interpretation

TAKEAWAYS:

• Qualitative analysis identifies themes (not numbers) from words, images, or behaviors to answer "how" and "why" questions in social work research.
• Thematic analysis (Braun & Clarke’s 6-step model) is the most common method for coding and interpreting patterns in interview transcripts or field notes.
• Content analysis quantifies word frequencies (e.g., counting mentions of "child labor" in policy documents) but retains qualitative depth.
• Discourse analysis examines how language constructs power, identity, or social norms (e.g., analyzing NGO reports on gender roles).
• Software like NVivo speeds up coding but requires manual validation to avoid "data dredging" (finding patterns that aren’t there).
• Ethical pitfalls include misrepresenting voices (e.g., paraphrasing participants poorly) or overgeneralizing from small samples.

What Is Qualitative Data Analysis?

Qualitative data analysis (QDA) transforms unstructured data—words, images, sounds, or observations—into meaningful insights. Unlike quantitative methods (which use statistics), QDA focuses on interpretation, context, and participant perspectives. It answers questions like:

• How do single mothers in Kathmandu describe their coping strategies? (Interview transcripts)
• What themes emerge from NTC’s customer complaints about service delays? (Social media posts)
• How does Pathao’s app language shape rider perceptions of safety? (User reviews)

Key Techniques in Qualitative Analysis

1. Thematic Analysis (Most Common in Social Work)

Thematic analysis identifies repeated patterns or themes in data. It’s flexible and widely used in case studies, policy evaluations, and program assessments.

Braun & Clarke’s 6-Step Model

Worked Example: Analyzing NGO Field Notes Scenario: An NGO in Pokhara conducts home visits to assess child malnutrition. Field notes include observations like:

• "Mother avoids giving child rice due to ‘bad karma’ beliefs."
• "Father reports child ‘eats too much’ but serves only dal."
• "Child cries when offered vegetables; mother says ‘disrespectful.’"

Step-by-Step Coding:

1. Initial Codes:
   • Cultural beliefs → "Food taboos"
   • Parental control → "Gender roles"
   • Child behavior → "Resistance to food"
2. Themes:
   • Theme 1: Cultural barriers to nutrition (codes: food taboos, disrespect)
   • Theme 2: Parental power dynamics (codes: gender roles, control)
3. Sub-themes:
   • Theme 1: Sub-theme A = "Religious restrictions on food"
   • Theme 2: Sub-theme B = "Mother’s agency vs. father’s authority"

Advantages:

• Reveals hidden complexities (e.g., why a program fails despite high participation).
• Participant-centered: Lets voices guide the analysis.

Disadvantages:

• Subjective: Different researchers may code the same data differently.
• Time-consuming: Requires immersion in data.

2. Content Analysis

Content analysis systematically counts and categorizes words or phrases to quantify qualitative data. It’s useful for:

• Analyzing media messages (e.g., how newspapers frame child trafficking).
• Policy document reviews (e.g., counting mentions of "mental health" in health ministry reports).

Example: Analyzing Daraz Customer Reviews Research Question: How does Daraz’s app language influence buyer trust?

• Step 1: Collect 500 reviews from the "Electronics" section.
• Step 2: Code for:
  • Positive words: "fast," "reliable," "happy"
  • Negative words: "scam," "delay," "broken"
  • Neutral words: "average," "okay"
• Step 3: Calculate frequencies:

  Category	Frequency	% of Total
  Positive	280	56%
  Negative	120	24%
  Neutral	100	20%

When to Use:

• You need quick, replicable insights (e.g., for a program evaluation).
• Data is text-heavy (e.g., social media, surveys).

Limitations:

• Loses context: Counting "happy" doesn’t explain why buyers are happy.
• Superficial: May miss sarcasm or cultural nuances.

3. Discourse Analysis

Discourse analysis examines how language constructs reality, power, and identity. It’s critical for:

• Gender studies: How NEPSE’s annual reports describe "female entrepreneurs."
• Policy critiques: How the government frames "youth unemployment" in media.

Example: Analyzing Ncell’s SMS Language Research Question: How does Ncell’s promotional language shape perceptions of data plans?

• Step 1: Collect 20 SMS campaigns from 2020–2023.
• Step 2: Code for:
  • Power: "Unlimited data—you deserve it!" (individualism)
  • Scarcity: "Only 100 slots left!" (urgency)
  • Exclusion: "Not for basic phones" (digital divide)
• Step 3: Interpret:
  • Language reinforces consumerism and tech elitism.
  • May alienate low-income users who can’t afford "premium" plans.

Key Questions to Ask:

• Who is silenced or centered in this language?
• What assumptions does the text make? (e.g., "All youth want smartphones.")

4. Narrative Analysis

Narrative analysis studies personal stories to understand identity, trauma, or social change. Used in:

• Social work case studies: How a survivor of domestic violence reconstructs their life story.
• Oral histories: Recording elderly Dalits’ memories of land reforms.

Example: Analyzing a WhatsApp Status Story Scenario: A social worker collects WhatsApp status updates from a group of migrant workers in Malaysia.

• Step 1: Identify narrative structures:
  • Problem: "No work, no money, boss cheats."
  • Quest: "I’ll send money home by next month."
  • Resolution: "My family prays for me daily."
• Step 2: Compare across stories:
  • All narratives follow a struggle → hope → faith arc.
  • Absent themes: Government support, union rights.

Tools for Qualitative Analysis

Warning: Software does not replace critical thinking. Always:

1. Triangulate: Cross-check themes with other data (e.g., observations + interviews).
2. Avoid "data dredging": Don’t force themes to fit a preconceived idea.

In the Real World

1. eSewa’s Customer Service Analysis

   • What it uses: Thematic analysis of chat transcripts to identify common complaints (e.g., "payment failures," "agent rudeness").
   • How: eSewa codes 10,000+ monthly chats to train AI bots and redesign FAQs. In 2022, they found "verification delays" were the top theme, leading to a new OTP system.
   • Social work link: Helps NGOs design user-friendly payment systems for rural areas.

2. Khalti’s Discourse on Financial Inclusion

   • What it uses: Discourse analysis of Khalti’s marketing slogans ("Money for everyone, everywhere").
   • How: Researchers coded ads for:
     • Inclusionary language: "No bank account? No problem!"
     • Exclusionary subtext: "Digital literacy required" (sidelines illiterate users).
   • Impact: Influenced Nepal Rastra Bank’s 2023 guidelines on financial literacy programs.

3. Pathao’s Rider Safety Reports

   • What it uses: Content analysis of rider reviews to track safety concerns.
   • Example: In 2021, Pathao found "harassment" mentioned in 15% of female rider reviews. They responded by:
     • Adding a "Safety Mode" (directs riders to well-lit routes).
     • Training drivers on gender-sensitive behavior.
   • Social work application: Shows how data-driven advocacy can improve urban mobility for vulnerable groups.

Common Pitfalls and How to Avoid Them

mindmap
  root((Qualitative Analysis Mistakes))
    A1[Overgeneralizing]
      A1a["Fix: Use rich descriptions"]
      A1b["Example: Don’t say ‘All single mothers struggle’—say ‘70% of participants in Kathmandu cited childcare as a barrier.’"]
    A2[Ignoring Contradictions]
      A2a["Fix: Code ‘discrepant cases’ separately"]
      A2b["Example: One participant said ‘I love my job’ but cried during the interview."]
    A3[Poor Transcription]
      A3a["Fix: Use verbatim quotes"]
      A3b["Bad: ‘She said she was happy.’ Good: ‘I’m happy now that my kids are in school.’"]
    A4[Researcher Bias]
      A4a["Fix: Peer debriefing"]
      A4b["Example: If you’re pro-feminist, check if you’re only coding ‘patriarchy’ themes."]

Exam Tip: How to Score Full Marks

1. Structure Your Answer Like This:
   • Introduction: Define the method (e.g., "Thematic analysis is an inductive approach...").
   • Steps: Use Braun & Clarke’s 6 steps or a similar framework.
   • Example: Always include a concrete case (e.g., NGO field notes, Daraz reviews).
   • Critique: Mention one strength and one limitation (e.g., "Reveals depth but is time-intensive").

2. Avoid These Exam Traps:

   • ❌ Saying "qualitative analysis is just reading data." → Wrong. It’s systematic coding and interpretation.
   • ❌ Confusing content analysis (quantitative counts) with thematic analysis (qualitative themes).
   • ❌ Ignoring ethics. Always state how you protected participant anonymity (e.g., "Codes replaced names").

3. Memorize These Key Phrases:

   • "Themes emerged inductively from the data."
   • "Codes were triangulated across interviews and observations."
   • "The analysis followed Braun & Clarke’s (2006) reflexive approach."

4. For Short-Answer Questions:

   • Question: "When would you use discourse analysis?"
   • Answer:

     "Discourse analysis is ideal for examining how language constructs social realities, such as power relations or cultural norms. For example, analyzing NTC’s customer complaint tweets could reveal how the organization frames ‘service delays’ as ‘technical issues’ rather than systemic failures, thereby shifting blame away from policy gaps. It’s useful when researching media narratives, policy documents, or everyday conversations where language shapes perceptions."

Practice Question with Model Answer

Question: "With a suitable example, describe the steps of thematic analysis. Discuss one advantage and one disadvantage of this method."

Model Answer: Thematic analysis, as outlined by Braun and Clarke (2006), is a six-step inductive process to identify patterns in qualitative data. Below are the steps applied to a case study of child labor in brick kilns in Siraha:

1. Familiarization: I immersed myself in 15 interview transcripts with former child workers, reading them repeatedly to grasp their experiences. For example, one participant said, "I worked from 4 AM to 8 PM with no breaks. The owner said I was ‘lazy’ when I asked for water."

2. Initial Coding: I coded interesting features manually in Microsoft Word, highlighting phrases like:

   • "No breaks" → Code: Exploitation
   • "Owner said I was ‘lazy’" → Code: Gaslighting
   • "My hands are always bleeding" → Code: Physical abuse

3. Searching for Themes: I grouped codes into potential themes:

   • Theme 1: Forced labor conditions (codes: no breaks, long hours, physical abuse)
   • Theme 2: Psychological manipulation (codes: gaslighting, threats)

4. Reviewing Themes: I checked if themes fit all data. For example, "gaslighting" was only present in 3 interviews, so I merged it into a sub-theme under Psychological manipulation.

5. Defining Themes: I named themes clearly:

   • Theme 1: "Systemic exploitation in kilns" (sub-themes: wage theft, child-sized tools, debt bondage).
   • Theme 2: "Breaking worker resistance" (sub-themes: fear tactics, family pressure).

6. Producing the Report: I wrote a participant-led narrative, using verbatim quotes to illustrate themes. For example:

   "The kiln owner told me, ‘You owe us money, so you must work.’ I didn’t know how much I owed—my father never explained." (Theme: Debt bondage)

Advantage: Thematic analysis reveals nuanced, context-specific insights. In this case, it uncovered how debt and fear were intertwined, which quantitative surveys might miss.

Disadvantage: The process is highly subjective. Another researcher might code "gaslighting" as a separate theme or merge it differently, reducing replicability.

Visual Aid: