Video summary

Qualitative Data Analysis - Coding & Developing Themes

Main summary

Key takeaways

Educational

Main Ideas / Lessons Conveyed

  1. Coding should follow familiarity with the data

    • Coding is best done only once the researcher is fully familiar with the dataset already gathered.
    • Coding itself is described as straightforward: it involves labeling sections/passages of text with one or more code words.

  2. Two levels of coding

    • Manifest-level coding: focuses on the words explicitly present in the transcript (what is directly said).
    • Latent-level coding: involves the researcher’s judgment—reading between the lines and interpreting deeper meaning beyond the literal wording.

  3. Practical workflow for coding (hands-on and iterative)

    • Coding often uses highlighter pens, colored pens, and Post-it notes.
    • The goal is to become progressively more familiar with the data while:
      • going through the entire dataset
      • applying codes to relevant text segments
      • allowing multiple codes to be assigned to the same segment

  4. How to perform the coding process (step-by-step)

    • Step 1: Read the transcript and highlight parts that are interesting/salient to the research questions or objectives.
    • Step 2: Assign a code representing the meaning/features of that highlighted segment.
    • Step 3: Maintain a long list of codes aligned with different transcript segments.
    • Step 4: Keep codes brief and succinct (example: “Low sickness rate”).
    • Step 5: Refocus and refine analysis by organizing/sorting codes into groupings that can form broader thematic categories.

  5. From codes to themes

    • Codes that are similar or related should be merged/grouped to create an overarching theme.
    • Example:
      • Initial codes related to prisoners being locked in their cells (e.g., being locked up, time slowing down, worrying, boredom)
      • These were grouped into the theme “bang up”, representing prisoners’ loss of control in their prison cell.

  6. Theme interrelationships and hierarchies via thematic networks

    • The method is attributed to Jennifer Atri Sterling, who distinguishes:
      • Basic themes
      • Organizing themes
      • Global themes
    • Example using “bang up”:
      • Another related theme is “feeling infantilized” (disempowering treatment like being treated as children).
      • The commonality linking basic themes is losing control.
      • “Losing control” is described as an organizing theme.
      • Multiple organizing themes can combine to form a global theme (example: “control”).
    • A thematic network shows how basic, organizing, and global themes interlink and may form hierarchies.

  7. Common pitfalls in qualitative thematic analysis

    • Pitfall 1: Endless quotations used to support thematic claims
      • Listing quotes under a heading is not enough.
      • Quotes should be used sparingly, with more analytical commentary explaining the interpretation.
    • Pitfall 2: Confusing interview/focus group questions with themes
      • Data collection questions are not themes; they’re just questions.
      • A poor shortcut is using the questions as headings and merely answering them.
      • Proper analysis requires real work to develop thematic categories from the raw data.
    • Pitfall 3: Lack of grounding in the original data
      • Themes must be clearly linkable back to the original raw dataset.
      • This grounding supports trustworthiness in the analysis.

Methodology / Instructions (Detailed)

Coding Approach

  • Ensure familiarity first
    • Code only after the researcher is fully familiar with the collected data.
  • Decide coding level
    • Manifest coding: label text using the literal words present.
    • Latent coding: label text using interpretation/judgment (reading between the lines).
  • Use practical coding tools
    • Highlighting and annotation tools such as highlighters, colored pens, and Post-it notes.
  • Process the entire dataset
    • Go through the whole dataset and identify segments relevant to the research questions/objectives.
  • Assign codes to segments
    • Highlight salient parts and attach one or more code words to each highlighted segment.
    • Allow multiple codes per segment where appropriate.
  • Keep codes concise
    • Use brief, succinct code labels (example: “Low sickness rate”).
  • Maintain a code list
    • Build a long list of codes that correspond to different transcript passages.

Moving From Codes to Themes

  • Group and merge codes
    • Sort codes into groupings based on similarity/relationship.
  • Form overarching thematic categories
    • Combine related codes into a broader theme capturing a shared meaning.
  • Use the theme to express an interpretive point
    • Example: codes about being locked up → theme “bang up” → meaning loss of control.

Thematic Networks (Hierarchy / Interlinking)

  • Identify basic themes
    • Themes sharing a common underlying process (example: losing control).
  • Identify organizing themes
    • Organizing themes link multiple basic themes (example: “losing control” linking “bang up” and others).
  • Identify global themes
    • Multiple organizing themes interlink to form one central global theme (example: “control”).
  • Map relationships
    • Create a thematic network showing interrelationships between basic, organizing, and global themes (including hierarchy/order).

Avoid Common Pitfalls

  • Don’t rely on quotes alone
    • Use quotations sparingly and pair them with analytical commentary.
  • Don’t treat questions as themes
    • Transform raw data into themes; don’t merely restate answers to interview questions under headings.
  • Ground themes in original data
    • Ensure every theme/idea can be traced back to raw transcripts to maintain analytical trustworthiness.

Speakers / Sources Featured

  • Speaker (unnamed): the presenter who explains the coding process and pitfalls.
  • Jennifer Atri Sterling: cited as the source for the thematic network framework (basic themes, organizing themes, global themes).

Original video