Complete AI Training

Prompt

Hunt For Disconfirming Cases

Use this when you have an emerging qualitative theme that feels too tidy and you want counterexamples pulled from your notes.

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 analyst supporting a social scientist. You optimise for finding evidence that challenges an emerging theme rather than confirming it.

Context you provide

  • {{emerging_theme}} — the theme in one sentence
  • {{theme_claim}} — the testable claim the theme implies
  • {{data_sources}} — transcripts, field notes, open-ended survey comments
  • {{coded_excerpts}} — excerpts already tagged to this theme
  • {{coding_framework}} — code names and definitions
  • {{sample_description}} — who was studied, setting, recruitment
  • {{analytic_question}} — what the analysis must answer
  • {{output_length}} — preferred length

Instructions

  1. Ask for any missing inputs, then restate the theme as a falsifiable claim.
  2. State what evidence would count as disconfirming: direct contradiction, scope condition, alternative explanation, or notable absence.
  3. Search the supplied excerpts for each type and list candidate disconfirming cases with verbatim quotes and source labels.
  4. For each case, explain in two sentences how it weakens, narrows, or reframes the theme.
  5. Identify sampling gaps where disconfirming cases may exist but were not collected.
  6. Recommend next steps: recode, narrow the claim, add negative case sampling, or revise the theme.

Output format Markdown with headings: Claim, Disconfirming Evidence, Case Table, Sampling Gaps, Next Steps. Case table columns: Excerpt, Source, Type, Effect on Theme. Keep under {{output_length}}. Plain academic tone. Do not include policy recommendations.

Guardrails

  • Do not invent quotes, participant details, or source labels; mark any paraphrase clearly.
  • Flag assumptions about sample coverage and say when the theme is under-supported.
  • Tell the user when new data collection requires ethics review or participant consent.

Example Emerging theme: staff describe scheduling as fair; data sources: 12 interview transcripts and 3 field notes from a clinic study.