What methodological changes in ICE reporting since 2016 have most affected historical deportation counts?
This fact-check may be outdated. Consider refreshing it to get the most current information.
Executive summary
Since 2016, several discrete methodological changes — and ongoing inconsistencies in how ICE and related DHS offices define, count, and publish enforcement events — have materially altered historical tallies of deportations, arrests, and detentions; the most consequential are changes to the unit-of-count for administrative arrests, selective inclusion/exclusion of certain “book-ins” and transfers in detention and removal tallies, and uneven reporting schedules and revisions that create apparent jumps or drops when datasets are “locked” or reissued (OHSS; ICE; GAO; Deportation Data Project) [1] [2] [3] [4].
1. Counting rules changed: administrative arrests redefined, inflating comparable year‑over‑year figures
A clear inflection comes from a 2016 revision in how ICE ERO counts administrative arrests — prior guidance allowed only one administrative arrest per person per day, but the methodology was revised to align ICE reporting so that multiple administrative actions involving the same individual could be recorded, producing higher event counts for later years when applied to historical data; the OHSS yearbook explicitly flags that the counting methodology for administrative arrests was revised to align with ICE ERO reporting for 2016 [1].
2. Detention tallies understate activity because of exclusions of initial temporary book‑ins
The Government Accountability Office found ICE’s public detention figures exclude people who were first booked into certain temporary facilities and later transferred into ICE custody, a methodological choice that results in tens of thousands of omitted detentions and therefore distorts time‑series comparisons unless researchers reconstruct detention stints from FOIA datasets (GAO) [3].
3. Removals data reliability and opaque corrections force researchers to recompute deportation counts
Independent data aggregators — notably the Deportation Data Project — have refused to publish ICE’s recent removals table because of doubts about its reliability and instead recommend assembling removals from departures recorded in arrests and detention datasets; this implies that official “removal” counts have been unstable and sometimes internally inconsistent, pushing journalists and scholars to recompute removals from lower‑level records rather than accept headline totals [4].
4. Reporting cadence, revisions and “locking” create artifactual spikes and troughs
ICE itself cautions that its statistics “fluctuate until ‘locked’ at the conclusion of the fiscal year” and that quarterly reports may be superseded by corrections, which means short‑term comparisons can reflect paperwork timing and post‑hoc editing rather than enforcement reality; OHSS and DHS statisticians also note that records may not match across agency reports because of differing methodologies and as‑of dates [2] [5].
5. Transparency gaps, FOIA reconstructions and advocacy datasets have shaped the public record
Because ICE’s published counts leave methodological gaps, researchers and advocacy groups have reconstructed detention‑level and arrest‑level datasets via FOIA; dashboards from Vera and analyses cited by journalists rely on these reconstructed, stint‑level records, which often show different trends than ICE’s headline numbers — an implicit consequence is that public debate about “how many” deportations occurred depends heavily on which dataset and counting rules are used [6] [7].
6. Competing narratives, political incentives, and what this means for historical comparisons
Different actors have incentives: administrations emphasize higher removals to tout enforcement, while watchdogs emphasize undercounts or opacity to argue for reform; Wikipedia and press reporting note that counting changes earlier than 2016 (during prior administrations) also affected historical tallies, underscoring that the problem is cumulative and political as much as technical [8]. The GAO urged ICE to publicly report all detentions and explain its methodology; DHS declined, reflecting institutional resistance that itself shapes what counts as an official deportation [3].
Conclusion: which methodological changes mattered most
The methodological shifts that most affected historical deportation counts are (a) the 2016 change in counting administrative arrests that allowed multiple same‑day events to be recorded [1], (b) ICE’s exclusion of certain initial book‑ins in detention reporting that understates detention and thus removal exposure [3], and (c) unstable removals tables and variable reporting/locking practices that force third parties to rebuild counts from raw arrest and detentions stints [4] [2]. These changes — compounded by limited transparency and political pressure — mean headline deportation totals since 2016 are not strictly comparable unless researchers explicitly account for the underlying counting rules and reconstruct historical series from consistent unit‑level records [7] [6].