How do ICE’s biweekly point-in-time detention counts differ from year‑end totals and ‘people processed’ metrics?
Executive summary
ICE’s biweekly detention spreadsheets publish point‑in‑time snapshots — typically “midnight population” counts — that show how many people are in custody at a particular moment, while year‑end totals are final, corrected summary figures that may be revised after the fact; metrics described as “people processed” or event counts (book‑ins/book‑outs) record multiple encounters and thus can be much larger than momentary population totals [1] [2] [3].
1. What the biweekly point‑in‑time numbers actually measure
The figures released every two weeks by ICE are essentially point‑in‑time population snapshots — often the midnight population — and are meant to provide a rolling picture of the detained population at the moment the count is taken rather than an aggregate of all people who passed through facilities over a period [1] [4]. Journalistic and researcher scrutiny of these biweekly spreadsheets has produced facility‑level time series and helped NGOs and newsrooms track surge locations, but the underlying measure is still a population‑at‑a‑moment metric, not a cumulative flow measure [5] [4].
2. Why year‑end totals can differ from biweekly snapshots
Year‑end totals released by ICE and summarized in annual reports are “locked” only after fiscal‑year reconciliation and can include corrections, removals of duplicate administrative records, and late data submissions — meaning those end‑of‑year counts can supersede earlier biweekly releases and typically present a different, more finalized view of detention levels [2]. Researchers and groups such as TRAC and the Vera Institute have used both biweekly feeds and locked year‑end data to construct trends, noting that end‑of‑year population figures can be higher or lower depending on late adjustments and how transient stays are recorded [6] [1].
3. How ‘people processed’ or event counts inflate comparisons
Operational reporting like the Detentions KHSM and ICE’s own event databases count detention events — every initial book‑in and final release — so an individual who is booked in, released, re‑booked and transferred can be counted multiple times; the effect is that “people processed” or event tallies measure throughput rather than unique persons in custody and therefore will exceed point‑in‑time populations sometimes dramatically [3]. Multiple datasets that researchers use — including ICE’s Enforcement Integrated Database and third‑party processed files — explicitly note that repeat book‑ins inflate totals unless de‑duplicated for unique individuals [3] [7].
4. Alternative metrics and their implications for reporting
Because point‑in‑time counts, year‑end snaps and event tallies answer different questions, analysts have proposed alternative constructs — for example “Interval Average Daily Population (Interval ADP)” — which aim to better estimate the day‑to‑day average held at facilities and to reconcile differences between instantaneous reported counts and cumulative flows [5]. Advocates of Interval ADP argue it’s more timely and reflective of facility load than ICE’s reported ADP for some centers, and journalists and researchers use such measures to highlight discrepancies between what biweekly snapshots show and how many people facilities actually handle over a period [5].
5. What the differences mean in plain terms for interpretation
Comparing the three metrics without care is misleading: a biweekly midnight count tells how many beds are occupied at the count moment, year‑end totals are the reconciled system‑wide snapshot after administrative corrections, and “people processed” or event counts show throughput and can double‑ or triple‑count individuals with multiple encounters [1] [2] [3]. The consequence is practical: policy debates anchored on “how many people ICE detained” need to specify whether they mean instantaneous population, cumulative unique individuals, or total detention events, because each paints a different picture of capacity, enforcement intensity and human turnover [8] [6].
6. Caveats, hidden biases and where the data still leave questions
ICE’s publications note data fluctuate until locked and are published with a lag, and federal dashboards and independent FOIA‑based projects (Vera, Deportation Data Project, TRAC) surface differences in facility reporting practices that can skew interpretation if not adjusted; independent methods and explicit de‑duplication or interval averaging are necessary to answer questions about individuals rather than beds [2] [1] [7]. This reporting draws attention to institutional incentives — facilities are funded on occupied beds, and event counts are operationally useful — but the public discourse often blurs the three measures, which benefits competing narratives unless reporters and analysts make the distinction explicit [8] [5].