In designing this project, we explored multiple sources of deportation data—each with its own strengths and limitations. After careful review, we selected DeportationData.org as our primary dataset, with the UWCHR's narrative analysis and TRAC's historical archives serving as two other key sources for methodology and context.
Our Sources
Primary Dataset: DeportationData.org
A public-interest database offering downloadable, structured case-level records. This allows us to analyze quantitative patterns in deportations across jurisdictions and time.
Methodological Model: UWCHR's I-213 Narrative Analysis
The University of Washington Center for Human Rights produced one of the only public analyses of how ICE agents describe and justify enforcement actions. While their data is limited to a single state, their ethical framework and their deep dive into the language of enforcement provide our methodological blueprint.
Historical Context: TRAC Archives
The Transactional Records Access Clearinghouse is an invaluable reference for understanding the historical context of the data we analyze.
The Breakthrough: Narrative Analysis at a National Scale
The groundbreaking work of the UWCHR proved that analyzing the narratives on official forms reveals how power is exercised. However, their state-level focus highlighted a critical challenge: scaling this kind of deep, qualitative work nationwide has been nearly impossible.
This is precisely the challenge that AI For The People was created to address.
We believe that by pairing human researchers with powerful AI language models, we can apply the principles of narrative analysis to a national dataset for the first time. This allows us to move beyond just raw numbers and begin to understand how the system justifies its actions, in its own words, across the entire country.