FAQ

1) What is a claim graph?

A claim graph is a paper‑level map of stated relationships between standardized concepts. Nodes are mapped to JEL concepts; edges capture what the authors claim, and edges are annotated with evidentiary basis.

2) What do you mean by “causal” vs “non‑causal”?

“Causal” edges are claims supported by canonical causal identification designs (e.g., DiD/event studies, IV, RDD, RCTs, synthetic control).

“Non‑causal” edges are supported by other forms of evidence (theory, descriptive patterns, correlations, etc.).

3) Does this project tell me which claims are true?

No. We extract what papers claim and how they support those claims. We do not verify correctness.

4) What papers are included?

The core corpus is 44,852 working papers from NBER and CEPR covering 1980–2023.

5) How do I get the data and code?

Everything is on GitHub. Contact us for any advice on using it :) 

6) Can I run this on my own paper?

Yes. The repository includes an extraction workflow (Stage 1–4). You provide your own API key via environment variables and run the scripts on your input files.

7) How should I cite this work?

Please cite the paper (see the Paper page) and include the GitHub repository link when referencing data/code.

Still have questions?

Email: team@causal.claims