> For the complete documentation index, see [llms.txt](https://evidano.gitbook.io/ailyze-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://evidano.gitbook.io/ailyze-docs/methods-and-best-practices/qualitative-research-best-practices.md).

# Qualitative Research Best Practices

Get credible, defensible qualitative findings from Evidano that hold up in a paper, a stakeholder review, or a board meeting.

This page shows you how to use Evidano so your qualitative findings hold up to scrutiny. AI speeds up qualitative analysis, but it doesn't replace your judgment — these practices keep your work rigorous.

## 1. Prepare clean inputs

* Keep **one interview or group in each file**, with **Speaker Labels** intact.
* Fix transcription errors first — garbage in, garbage out. Use the transcript editor.
* Give Evidano context: drop your interview guide or framework into the research context panel at the start of document analysis so [themes](/ailyze-docs/reference/glossary.md#theme) line up with your project.

## 2. Treat AI themes as a first draft

* Generate themes with AI, then *curate* them: rename, merge, split, and define each [code](/ailyze-docs/reference/glossary.md#sub-theme) in the codebook editor.
* Add a description to every code so labeling stays consistent and auditable.
* Don't accept the [codebook](/ailyze-docs/reference/glossary.md#codebook) just because it looks plausible — check it against your data.

## 3. Stay close to the data

* Read the [quotes](/ailyze-docs/reference/glossary.md#quote), not just the counts. A high [frequency](/ailyze-docs/reference/glossary.md#frequency) with weak quotes is a weak theme.
* Use the [co-occurrence network](/ailyze-docs/reference/glossary.md#co-occurrence-network) and [hierarchy graph](/ailyze-docs/reference/glossary.md#hierarchy-graph) to see relationships, not just how often something comes up.

## 4. Look for disconfirming evidence

Confirmation bias is the enemy. Use the [chat](/ailyze-docs/analyze-data/chat-with-your-data.md) to ask questions like *"What evidence contradicts theme X?"* and *"Which participants don't fit this pattern?"* Then report the nuance.

## 5. Check reliability

* Run **Calculate AI Accuracy** *(Pro)* to gauge how consistently the codebook was applied.
* Open the **AI Labels as Checklist (with code justifications)** tab to see *why* the AI assigned each code.
* For high-stakes work, have a second researcher review a sample.

## 6. Be transparent about method

* If you publish, name your [methodology](/ailyze-docs/reference/glossary.md#methodology) and describe how you used AI (it suggested themes; you curated them), your codebook, and your validation steps.
* Keep your codebook and exports as an audit trail. Reuse it across waves with **Use Previous** for consistency.

## 7. Handle quotes and people ethically

* Verify every quote against the source before you publish it.
* Anonymize: turn on personal-information (PII) redaction on transcripts, and be careful with the **Respondent Name/ID** field.
* Get consent for recording and analysis, and mind GDPR (the EU data-privacy law) and IRB (ethics-review board) requirements.

## 8. Mind sample size when comparing

Small [segments](/ailyze-docs/reference/glossary.md#segment) make for noisy comparisons. Don't over-interpret a difference in a group of five. See [Compare groups](/ailyze-docs/analyze-data/compare-groups-segments.md).

## A credibility checklist

* [ ] Inputs cleaned and given context
* [ ] Codebook curated, defined, and saved
* [ ] Quotes read and verified
* [ ] Disconfirming evidence sought
* [ ] Reliability checked
* [ ] Method and AI use documented
* [ ] Consent and anonymization handled

## Next steps

* [Build a codebook](/ailyze-docs/analyze-data/build-a-codebook.md) — curate the themes the AI proposes.
* [Discover themes and patterns](/ailyze-docs/analyze-data/discover-themes-and-patterns.md) — where the first draft of themes comes from.
* [Reporting findings](/ailyze-docs/methods-and-best-practices/reporting-findings.md) — write up findings with evidence.