Contributor Guide

Help by making the public record cleaner.

Contributions are useful when they improve source coverage, correction quality, duplicate detection, or school metadata without introducing private evidence or unsupported claims.

Ways to Contribute

• Submit a public source for review.
• Request a source-backed correction to an existing record.
• Report a likely duplicate record.
• Suggest a school metadata correction.
• Review methodology gaps, source-type gaps, or coverage gaps.
• Use the Reviewer Queue to find priority review samples.
• Help test downloads, search filters, and generated detail pages.

Accepted Sources

• Government releases, datasets, OCR materials, court records, and public legal filings.
• University statements, annual security reports, public safety notices, and policy notices.
• Campus newspaper reporting, reliable local journalism, and reliable national journalism.
• Nonprofit or civil-rights reports when the methodology and source basis are clear.

Rejected Material

• Private testimony, private screenshots, direct messages, private emails, or anonymous tips.
• Unverified social media-only claims.
• Private contact information or unnecessary names of private individuals.
• Legal conclusions not present in the public source.
• Submissions that ask the project to rank schools by hate, safety, or severity.

Submission Workflow

Use the submit page to prepare a structured packet. The packet can be copied into the appropriate GitHub issue template when the repository is public and ready for outside intake.

• Source submission: `.github/ISSUE_TEMPLATE/source-submission.yml`
• Correction request: `.github/ISSUE_TEMPLATE/correction-request.yml`
• Duplicate report: `.github/ISSUE_TEMPLATE/duplicate-report.yml`
• School metadata correction: `.github/ISSUE_TEMPLATE/school-metadata-correction.yml`

Review Standard

A submission does not become a public record until a reviewer confirms the source URL, school, date, category, affected community, attribution language, verification status, confidence label, and privacy risk.

Correction Standard

Corrections must identify the record ID, the disputed field, the requested change, and a public source supporting the change. Accepted corrections regenerate record hashes, dataset hashes, release notes, source audit metadata, and the public changelog.

Partner and Reviewer Path

Organizations, researchers, journalists, student groups, and civil-rights reviewers can help by reviewing methodology, identifying missing source categories, auditing classification rules, testing the dataset against a narrow research question, or checking whether the Research Guide prevents overclaiming. The Reviewer Brief gives outside reviewers the shortest entry point, and the Trust & Review Packet defines the broader review path. Acknowledgments should appear only after documented review or collaboration.

Reviewer Entry Points

• Methodology review: test inclusion, exclusion, source hierarchy, confidence, and deduplication rules.
• Source audit: check whether source URLs, publishers, dates, and source types remain accurate.
• Classification audit: inspect category and affected-community labels against source text.
• Research audit: use the archive for one narrow question and report where the guide or data model fails.
• Research packet: use the Research Workspace to create a citation packet with source URLs and snapshot hash.
• Reviewer brief: send the Reviewer Brief or Markdown packet when asking for outside critique.
• Reviewer checklist: use the GitHub `reviewer-checklist.yml` issue template to document scope, findings, suggested changes, and acknowledgment permission.
• Outreach and acknowledgment rules: see docs/outreach-email.md and docs/partner-acknowledgment-policy.md.

Local Data Workflow

Code or data contributors should read docs/contributing.md, then run `npm run prepare:data` and `npm run check` before proposing dataset changes.

Contributor Rule

If a contribution would make the dataset less public, less neutral, less attributable, or less auditable, it does not belong in the MVP.