Papers and data, published in the open.
Each SaneCite paper is listed here with its preprint, the full text as a PDF, the data and the code. Everything is free to read and reuse.
Retracted, still cited as support: a census of post-retraction citations to 22,483 retracted biomedical papers
Stephan Joseph (ORCID 0009-0001-0250-5446). Posted on MetaArXiv, doi:10.31222/osf.io/uwqxa_v1. License CC BY 4.0.
A pre-registered census of the citations retracted biomedical papers received after their retraction. AI models read each citing passage in open-access papers and labelled it: cited as support, mentioned in passing, or flagged as retracted or unreliable. The paper reports how often retracted work is still cited as support, how that changes with time since retraction, and how often work is still cited as support when the citing paper's own reference list marks it retracted. The figures are in the paper and on Findings.
- Full text: PDF (16 pages), or the MetaArXiv page.
- Protocol, code and data: GitHub, archived at Zenodo (doi:10.5281/zenodo.23358099).
- The citation labels as a dataset: Hugging Face.
- The named Lists built on it: Lists. The case stories: Stories.
Found an error? Write to SaneCite at letters@sanecite.com. Corrections are made in public and logged.