Which independent databases or investigations have attempted to count shootings by ICE/CBP across administrations, and what methodologies did they use?
Executive summary
Independent efforts to tally shootings by U.S. immigration agents have been carried out by news organizations, a crowd-sourced encyclopedia page and investigative outlets, but no single, fully independent national database produced by an academic or nonpartisan research institute is visible in the reporting provided; each effort uses different inclusion rules and source types — news reports, agency disclosures, internal memos, and open-source evidence such as body‑cam footage — producing counts that are not directly commensurable [1] [2] [3]. These projects expose gaps in official transparency and diverge sharply in methodology: some count only incidents where someone was struck, others include any discharge, and most rely on media and public records rather than systematic agency logs [1] [2] [3].
1. The newsroom roll-ups: NBC, The New York Times and The Washington Post as quasi‑databases
Major news organizations have compiled incident lists and timelines that function as de facto databases by aggregating reporting, agency notices and public records; NBC’s list explicitly limits entries to cases “in which an on‑duty ICE or CBP officer fired at and struck someone,” excluding discharges that missed and non‑firearm uses like pepper balls, a definitional choice that narrows its counted universe but increases confidence that entries represent actual injuries or deaths [1]. The Washington Post reported trends — for example, noting 16 DHS shootings since July that the administration had publicly justified before probes concluded — working from agency announcements and its own reporting rather than an independent, normalized dataset, a methodology that risks double‑counting or omitting incidents without public reporting [4].
2. The Trace and investigative outlets: attempting systematic context and patterns
Investigative outlets such as The Trace have tried to build longer‑term narratives about shootings by immigration agents and to place them in policy context, drawing on public records, interviews and law‑enforcement data; The Trace’s piece mapped incidents to policy changes and cultural explanations but focused more on qualitative patterning than on producing a formal, repeatable counting methodology accessible as a cleaned dataset [3]. Such outlets often describe methodological choices — why certain incidents are included or excluded — but their compilations remain dependent on media and official records rather than independent field verification [3].
3. Crowd‑maintained compilations: Wikipedia’s incident lists and their limits
Wikipedia hosts lists of shootings by immigration agents that aggregate media reports, official statements and other sources; as a crowd‑edited compilation it functions as an accessible chronology but lacks the consistency controls and independent verification processes of an academic database, and its entries carry qualifying notes where shooter identity or agency affiliation is disputed (for example, entries where both ICE and CBP were present and the page assumes CBP until proven otherwise) [2]. Wikipedia’s transparency about sourcing is useful, but its volunteer maintenance and variable sourcing standards make it an imperfect tool for rigorous cross‑administration counting [2].
4. What these efforts rely on: public notices, body‑cam, agency investigative units and local reporting
Across the projects, the primary raw materials are media accounts, federal notifications to Congress, internal agency reports and where available, body‑worn camera footage; for example, CBP’s Office of Professional Responsibility produced a minute‑by‑minute preliminary report on the Alex Pretti shooting that lawmakers received — a type of primary document newsroom compilers and investigators used to construct incident narratives [5] [6]. The reliance on such fragmentary sources means counts are only as complete as what agencies disclose and what local journalists or whistleblowers surface, leaving unknowns where agencies withhold footage or when state investigators are blocked [7] [8].
5. Divergent inclusion rules and the problem of comparability
Methodological differences are the clearest obstacle to crosswalks: NBC’s decision to count only shootings where someone was struck produces a smaller, more certain set than a count that includes all rounds fired; crowd lists and investigative stories may include preliminary agency claims about armament or threat that later evidence contradicts, as seen in internal reports that challenged DHS narratives in specific cases [1] [5]. Because projects use different start dates, differing definitions of “immigration agent” (ICE vs. CBP vs. USBP), and varying reliance on official vs. third‑party sources, their totals cannot be naively summed or treated as interchangeable [2] [1].
6. What is missing and why it matters
None of the reporting provided documents a single, independent, machine‑readable national database maintained by an academic institution or oversight body that standardizes definitions across administrations; instead, transparency depends on patchwork journalism, agency disclosures and volunteer compilations, a structural weakness flagged repeatedly in coverage of recent high‑profile incidents and the government’s inconsistent investigative handoffs [4] [7] [8]. That absence matters because policy debates and oversight hinge on comparable counts — how many people were shot, under what rules of engagement, and whether administrative conclusions were premature — and current efforts, while valuable, leave important evidentiary gaps [4] [7].