Making sense of qualitative answers
Listen to this lesson
Nigerian voice narration
Fictional training case. Adeyemi's survey drew 200 open-text responses plus 30 pages of interview notes. This lesson covers building a coding frame with counts, validating codes on a sample, and reporting contradictory evidence honestly rather than forcing false agreement. Work through the example, identify the evidence needed and practise the method with cleared or synthetic inputs. Any additional figures in the explanation are separate practice examples, not claimed rows in a downloadable file.
Worked example
Adeyemi pastes anonymised open-text responses and asks AI to develop a coding frame: themes, definitions, and counts, with two anonymised quotes per theme. AI proposes themes like 'Time pressure relief' (62 mentions) and 'Trust concerns about accuracy' (38 mentions). Before accepting the counts, Adeyemi manually codes a random sample of 20 responses herself and compares her codes to AI's: she finds 3 responses AI coded as 'Time pressure relief' that she judges actually belong under a different theme, 'Workload redistribution,' revealing AI's coding frame needs refinement rather than blind acceptance. She also finds one subgroup of respondents reporting the opposite of the dominant theme (AI increasing their workload) and instructs AI to report this as a separate, contradictory finding rather than merging it into the majority theme.
The method
Validate AI's coding frame by manually coding a random sample yourself and comparing results before trusting the full-dataset counts, and explicitly instruct AI to report minority/contradictory findings as their own theme rather than folding them into the majority for a tidier narrative.
Worked practice and common mistakes
Sixty-two of two hundred is 31%, not a majority. The largest theme can still represent a minority. Define whether counts refer to comments, people or groups and whether multiple codes are allowed. Refine definitions when manual coding disagrees with AI. Preserve contradictory evidence and trace each quoted passage to the original fictional response rather than a generated paraphrase.
Review before using the result
Qualitative analysis looks for themes and meanings backed by evidence. Anonymise before sharing. Validate AI's coding on a manual sample. Report contradictory evidence honestly instead of forcing agreement.
Your workplace practice
Use the recommended free assistant with this lesson's fictional example. Here are anonymised open-text responses. Develop a coding frame: themes, definitions, counts, with 2 anonymised quotes per theme. Flag contradictory evidence rather than forcing agreement. Keep a copy of the input and draft. Compare the result with the verification steps above, record any correction, and retain placeholders where evidence is missing. The expected result is a checked draft for human review, not an automatically approved decision. Course examples and practice responses are fictional; amounts, dates, rules and thresholds are not official guidance or supplier quotations.
Open Claude (free tier) and paste this prompt there. Suggests themes from cleared fictional excerpts; verify each quote, code and count against the source.
Here are anonymised open-text responses. Develop a coding frame: themes, definitions, counts, with 2 anonymised quotes per theme. Flag contradictory evidence rather than forcing agreement.
Why this tool: Suggests themes from cleared fictional excerpts; verify each quote, code and count against the source.
Hit a free limit? Switch to ChatGPT or Gemini and continue — the prompt works in any assistant.