Evidence & Research Standards
Credibility requires visible boundaries.
These standards govern how ALPRP distinguishes official records, lived experience, analysis, proposals, and prototypes—and how it handles uncertainty, privacy, correction, and conflicting evidence.
Evidence classifications
| Label | Meaning | Publication rule |
|---|---|---|
| OFFICIAL DATA | Government, court, agency, or other authoritative public material. | Preserve value, date, denominator or definition, and source. |
| RESIDENT-REPORTED | A firsthand or relayed account from a resident or family member. | Attribute as reported; do not present as an adjudicated fact. |
| ADVOCACY CLAIM | A conclusion or allegation presented by an advocacy source. | Name the source type and seek independent support. |
| ALPRP ANALYSIS | ALPRP synthesis, interpretation, estimate, or calculation. | Disclose method, assumptions, limitations, and underlying evidence. |
| ALPRP PROPOSAL | A reform, safeguard, program, or operating model ALPRP recommends. | Never describe it as adopted or currently operating in Alabama. |
| ILLUSTRATIVE / PROTOTYPE | A visualization or demonstration of a proposed concept. | Never present it as an existing ADOC facility, service, or program. |
Corroboration status
Classification identifies the kind of evidence. Corroboration records how strongly a specific claim is supported.
Supported by a relevant official record.
Supported by credible evidence independent of the original source.
Similar reports from distinct sources or facilities warrant investigation.
Insufficient independent evidence; not treated as false or established.
Credible sources disagree or use incompatible definitions.
Stronger evidence materially conflicts with the claim.
Research workflow
- Collect: preserve the source, publication date, reported event date, facility, and issue category.
- Classify: label the evidence before interpretation.
- Corroborate: compare distinct accounts, official records, court materials, and credible independent sources.
- Compare: test policy promises against access, delivery, interruptions, and outcomes.
- Respond: translate credible findings into a research question, safeguard, metric, correction, or proposed reform.
Privacy and governance
- Minimize data and remove names, AIS numbers, medical details, and identifying circumstances from public summaries.
- Require informed consent before publishing an identifiable account.
- Use role-based, least-privilege access; segregate medical and mental-health information.
- Maintain retention, disclosure, security, audit, evidence-preservation, correction, and appeal rules.
- Label AI-assisted analysis, test for bias, and keep consequential decisions under authorized human control.
- Avoid disclosures that could increase retaliation or safety risk.
Correction and uncertainty
ALPRP will not silently merge conflicting figures, convert an estimate into an official statistic, or imply that absence of public confirmation disproves a resident account. Material corrections should identify what changed and why. When evidence remains incomplete, the uncertainty stays visible.
Research into requirements
Recurring reports now test seven safeguards
Resident and family reporting informs ALPRP design without turning unverified allegations into facts. The Systemwide Safeguards Standard converts credible recurring patterns into measurable protection, continuity, integrity, and accountability requirements.
Review the safeguards →