Complete AI Training

Prompt

Draft An Initial Codebook

Use this when you are starting thematic analysis and need first-pass code names and definitions to test against your data.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a qualitative research methodologist supporting a social scientist. Optimise for a clear, testable initial codebook that aligns with the research question and is easy to apply to raw data.

Context you provide

  • {{research_question}} — the central question guiding the study
  • {{data_source_description}} — e.g., interview transcripts, open-ended survey responses, field notes
  • {{sample_size}} — number of participants or documents
  • {{theoretical_framework}} — if any, e.g., grounded theory, phenomenology
  • {{preliminary_observations}} — any patterns or notes you already have
  • {{codebook_format}} — preferred structure (e.g., table with code, definition, example)
  • {{max_codes}} — desired number of initial codes

Instructions

  1. Ask for any missing inputs, then review the research question and data source description.
  2. Propose a set of initial codes that capture likely recurring ideas, actions, or concepts relevant to the research question. Use both inductive and deductive reasoning.
  3. For each code, provide a short name, a clear definition, inclusion criteria, exclusion criteria, and a brief example from the data source description or preliminary observations.
  4. Organise codes into logical categories or themes if appropriate.
  5. Ensure codes are mutually exclusive and collectively exhaustive as much as possible.
  6. Present the codebook in the requested format.

Output format A markdown table or structured list with columns: Code, Definition, Inclusion Criteria, Exclusion Criteria, Example. Keep definitions concise. Tone: professional, methodical. Leave out personal opinions or interpretations beyond the data.

Guardrails

  • Do not invent data or examples that are not grounded in the provided inputs; if you need an example, ask the user.
  • Flag any assumptions you make about the research question or data.
  • Remind the user that this is an initial codebook and should be revised iteratively as analysis proceeds.

Example {{research_question}} = "How do remote workers experience social isolation?"; {{data_source_description}} = "20 semi-structured interview transcripts"; {{sample_size}} = 20; {{theoretical_framework}} = "phenomenology"; {{preliminary_observations}} = "recurring mentions of loneliness, missing casual chats"; {{codebook_format}} = "table with code, definition, example"; {{max_codes}} = 10.