Compliance by Design: Evidence from Horizon Europe Research on Artificial Intelligence

Per-proposal coded pattern matrix anchoring a panel-side observational study of ethics governance patterns in a single funding call of Horizon Europe research on artificial intelligence. The matrix records the presence (✓), partial presence (~), or absence (✗) of five declarative ethics-governance elements (D1–D5) and five operational ethics-governance elements (O1–O5) across seven proposals: five involving sensitive personal data (A1–A5) and two low-sensitivity controls (C1–C2). In order to ensure the analysis stays at institutional level (avoiding attribution to individual proposals or applicants), proposals are referenced only by anonymous labels. The per-proposal source files from which the codes are extracted are confidential under the European Research Executive Agency's confidentiality obligations on independent expert evaluators and are not deposited. Contents: methodology.md (sample structure, anonymisation, ex-ante coding scheme, confidentiality regime, re-identification risk assessment); pattern-tables.md (per-proposal × per-element coded matrix: declarative elements, operational elements, qualitative pattern, per-proposal signatures). A subsequent version (2.0.0) will add the aggregate analysis — sample-level counts, rates, hypothesis-test outcomes, and interpretive narrative — that consumes this pattern matrix to support the empirical reporting of the corresponding manuscript.