# Machine-readable data

Stable JSON mirror of the data behind this site, compiled from New Jersey
Attorney General public records. All records carry a `sourceRef` (file,
sheet, row) or an equivalent provenance note. See https://policetransparencynj.com/methodology/.

| File | Contents |
| --- | --- |
| statewide.json | Statewide toplines by year. |
| entities.json | Canonical counties and agencies (slug, name, type, county). |
| counties.json | Per-county profiles and agency lists. |
| agency-stats.json | Per-agency yearly IA totals, discipline, and termination counts. |
| report-cards.json | Peer-grouped report-card grades, components, and confidence. Grades are this site's analysis, not a source fact and not an official rating. |
| non-reporting.json | Agencies the AG lists as reporting no IA investigations, by year. |
| encryption.json | Radio-encryption flags joined from PoliceRadioEncryption.com. |
| major-discipline.json | Every named major-discipline record with sourceRef. |
| ia-agency-totals.json | Agency-year internal affairs totals with sourceRef. |
| departments.json | Full per-department bundle (totals, records, benchmarks, IA detail). |
| officers.json | Per-officer bundle of major-discipline records. |
| pension.json | High-confidence PFRS/SPRS salary/pension matches to named officers, as reported by NJ Treasury YourMoney, with source-row provenance. |
| lee.json | High-confidence FBI Law Enforcement Employees (LEE) staffing matches, keyed by canonical agency slug: sworn/civilian headcounts (with a female-officer breakdown), by reported year, with source-row provenance. Agencies with no confident match are simply absent, never a fabricated zero. |
| summaries-index.json | Plain-language summaries for every department and officer page (`departments`, `officers`, both keyed by slug): the same generated, source-gated text rendered on each page, and the exact text used in that page's meta description and Dataset JSON-LD. Only fresh (fingerprint-verified) entries are included; a stale or not-yet-generated summary is simply absent, never a guess. |
| fe.json | New Jersey subset of the Fatal Encounters project (a third-party, independent, crowdsourced dataset), joined to canonical agencies at high confidence only: every recorded death during a police encounter of any kind, 2000-2021 (pursuits, suicides, and medical events included; a record is not a finding of fault). One record page per entry. FE's disposition and intended-use-of-force fields are carried under `caveatedFields` and displayed only with caveats; FE's race imputations are mirrored here for completeness but never rendered on pages. |
| militarization.json | DoD 1033/LESO surplus-equipment transfers by canonical agency (Defense Logistics Agency data, as of the file's `asOf` date). Line items record equipment an agency received over time, not current inventory; values are original DoD acquisition cost, not present worth. Items are classed tactical or mundane, and quantities aggregate only within a unit of issue (`quantitiesByUi`). Receiving surplus equipment is not misconduct. |
| opra.json | Public records requests filed to each department through OPRAmachine (an independent, third-party OPRA platform), joined to canonical agencies at high confidence only. A REQUEST IS NOT A FINDING: every entry records what a member of the public asked an agency for, never anything an agency was found to have done. Statuses (`byState`) are Alaveteli `described_state` values set by the REQUESTER when they classify the response, not by the agency and not by any court. `linkType` says whether the body is the department's own listing (`dedicated-pd`) or the municipal records custodian standing in for it (`municipal-fallback`), in which case the counts cover the whole municipality. Only requests already public on OPRAmachine are included. FROZEN: this is a hand-taken snapshot of the OPRAmachine database dated 2026-07-12 and nothing refreshes it, so every count is a lower bound that goes further out of date over time; see opramachine.com for current activity. |

Officer-to-pension links are high-confidence records-based matches, not certified
identities; grades and benchmarks are analysis, not source facts. Method at
https://policetransparencynj.com/methodology/.