What the study did
A demographer and sociologist at UFMG evaluated AI for qualitative coding in migration studies, presenting the results at the Brazilian Association for Population Studies (ABEP) national meeting on migration in November 2023, published in the proceedings. The same interview transcript was coded three ways: manually, with Atlas.ti’s AI coding, and with Evidano.
The transcript came from a study of 14 semi-structured interviews with Indigenous Pataxó migrants who moved from Bahia to Belo Horizonte, where they founded the Naô Xohã village.
How Evidano fit the workflow
- 1Interview transcript
Raw qualitative data
- 2Evidano analysis
Thematic analysis in under a minute
- 37 themes
A synthesis ready for the researcher to review
“The Evidano system works at a much more satisfactory level of synthesis.”
“Although it does not generate codes as in the manual operation or Atlas.ti, the system provides a good synthesis of the results based on a thematic analysis.”
“Its use as a research aid is already possible and can contribute significantly to the analysis of qualitative data in migration research.”
How the three approaches compared on one interview
| Dimension | Manual coding | Evidano |
|---|---|---|
| Time | ~40 min | < 1 min |
| Output | 12 codes | 7 themes; a clearer, more usable synthesis |
Atlas.ti: < 1 min, 120 codes in 5 groups; the author found this too many to categorize. The author reports Evidano’s results were more usable than the other AI tool tested.
In the study, 40 minutes of manual coding became under a minute.
7 clear themes, rather than 120 fragmented codes to untangle.
The tool was named and evaluated openly in published conference proceedings.
The paper frames AI as augmenting, not replacing, human judgment, and its results as a preliminary comparison.