Research case
Chain-of-thought qualitative coding matches humans on some tasks (Dunivin)
A sociologist redesigned a codebook for LLM use and compared GPT-3.5 and GPT-4 with human coders on interpretive coding tasks.
What is reported
GPT-4 reached human-equivalent interpretations on some codes while GPT-3.5 did not, and asking the model to justify each code raised fidelity considerably.
Before you use it
This is a specific socio-historical coding case with older models. Its agreement results do not establish validity for another codebook or corpus.
Sources and review
Source checked · Not tested by this site
Checked the source record and abstract. Results describe the reported study, not an independent replication by this site.
License and reuse
No license has been recorded. Check the source’s license before reusing code, models or data.
A license note is not a license audit. Open source does not guarantee free hosting, model access or commercial reuse.