CODEOWNERS & Review Accountability (2024–2025): Open-source pull request latency dataset

This deposit contains the reproducible dataset used to analyze how CODEOWNERS-style ownership governance influences pull request (PR) review and merge behavior across open-source repositories. Scope• Coverage: ~2,000 of the most active GitHub repositories (Jan 2024 – Sep 2025)• Records: ~525,000 governed PRs where ownership existed before review• Focus metrics: hours_to_first_review, hours_to_merge, merge IQR (predictability)• Sources: GitHub Archive (BigQuery) + GitHub REST API metadata (CODEOWNERS presence, creation date, owner count) Contents• msr-final-results.csv.gz — PR-level dataset (GZIP compressed)• owner_band_summary.csv — Aggregated summary table by ownership bands (1, 2, 3–6, 7–10, >10)• README (reference) — Workflow summary and data dictionary (see related GitHub repository for full README, SQL, and scripts) Reproducibility & MethodsQueries were executed in BigQuery (standard SQL) and joined with ownership metadata derived from GitHub’s REST API. The analysis includes only PRs where ownership governance (e.g., CODEOWNERS) was established before the first review and where timing fields are complete. For cost reasons, we recommend limiting BigQuery scans to smaller windows when exploring the logic. Licensing• Data in this deposit: CC BY 4.0• Code/scripts are maintained in the companion GitHub repository under the MIT License (see Related/External identifiers).