# knowless

> Open research: perception, definition, and expression gaps in Dutch societal debates.

## For AI agents

Use the API. Do not scrape the website.

- Base URL: `https://knowless.org/data`
- No authentication required
- Rate limit: 60 requests/hour per IP. Respect `X-RateLimit-Remaining` header.
- Start with `?endpoint=all` for a complete data dump.
- Full endpoint documentation: [llms-full.txt](https://knowless.org/llms-full.txt)
- OpenAPI spec: [openapi.yaml](https://knowless.org/.well-known/openapi.yaml)
- License: CC-BY-4.0

## What does knowless measure?

Three gaps per topic (migratie, gender, religie, democratie):

1. **Perception gap** (Module P): what people estimate vs. official statistics (in percentage points)
2. **Definition gap** (Module D): what people mean when they use the same words (scenario-based)
3. **Expression gap** (Module U): what people say publicly vs. privately

## Interpretation (important)

- Perception data measures GAPS, not opinions. "Overestimates by 24pp" is correct. "Thinks there are too many" is not.
- Definition/policy data measures DISTRIBUTIONS from a convenience sample. Use "among respondents", not "Dutch people think".
- Crosstabs link perception gaps to policy choices per demographic group.
- Use "correlates with", never "causes". Cross-sectional data cannot establish causality.

## Quick example

```bash
curl "https://knowless.org/data?endpoint=all"
```

## Citation

knowless (2025). Zes lagen in het maatschappelijk debat. https://knowless.org

## Links

- Full API docs: https://knowless.org/llms-full.txt
- Human-readable API page: https://knowless.org/api
- Methodology: https://knowless.org/method
- About: https://knowless.org/about
- Raw data (Zenodo): https://doi.org/10.5281/zenodo.15494874