What did independent analyses (Vera, Deportation Data Project, Marshall Project) find about trends in family detention and children’s length of stay from 2016–2025?
Executive summary
Independent analyses by the Vera Institute, the Deportation Data Project, and reporting by The Marshall Project find that family detention largely declined under the Biden administration but then surged again in mid‑2025 when ICE resumed holding families at Texas facilities, while dataset releases and FOIA delays complicate clear, continuous measurement of children’s length of stay from 2016–2025 [1] [2] [3]. The sources supply day‑by‑day facility populations and examples of very long detentions, but none of the provided documents offers a single, definitive national statistic for children’s average length of stay over the full 2016–2025 window—analysts instead reconstruct trends from facility‑level and booking datasets [4] [5] [6].
1. Vera’s facility‑level reconstruction: granular daily populations reveal family detention’s return
The Vera Institute’s ICE Detention Trends dashboard builds on Deportation Data Project and FOIA‑obtained ICE spreadsheets to assemble midnight and 24‑hour population counts for every facility on each day from October 1, 2008 through June 10, 2025, and its July 2025 update documents a marked increase in detention in 2025 tied to the new administration’s enforcement changes [4] [5]. Vera’s analysis specifically notes that the Biden administration had largely phased out family detention because of the harm to minors, but that ICE resumed holding families on June 10, 2025, with Karnes County and the South Texas Family Residential Center in Dilley holding 1,187 and 575 people respectively on that date—data points Vera highlights to show the operational return of family units to detention facilities [1].
2. Deportation Data Project: raw records enable tracking but require reconstruction and linking
The Deportation Data Project collects and posts anonymized government enforcement datasets and serves as the primary raw source that Vera and others use to reconstruct detention histories; the Project’s data releases include ICE “book‑ins,” facility identifiers, and other transactional records that make day‑by‑day population and cohort reconstructions possible but do not by themselves publish a single, national metric of children’s average length of stay across years [2] [6]. The Project’s materials and repository also document that government datasets are siloed, released irregularly, and sometimes lack documentation or consistent linking across court, CBP, and ICE files—limits that constrain how cleanly researchers can produce longitudinal length‑of‑stay statistics for minors [6] [7].
3. Marshall Project reporting: examples of long detentions and transparency gaps
The Marshall Project’s reporting draws on Deportation Data Project material and FOIA litigation to show that detention numbers and enforcement actions spiked in 2024–2025 and that FOIA slowdowns during agency disruptions have obscured timely data access, a practical impediment to precise trend measurement over 2016–2025 [3]. Its journalism provides human cases and data‑driven findings—such as reporting of people facing year‑long detentions and expanded deportations of people with minimal criminal histories—illustrating that long stays are real and consequential even if a consolidated national child‑length‑of‑stay time series is not presented in the cited excerpts [8] [3].
4. What the analyses do—and do not—conclusively show about children’s length of stay
Collectively, Vera and the Deportation Data Project give the empirical machinery to measure length of stay by tracing individual bookings and facility occupancy over time, and Vera’s June 2025 snapshot captures the operational resumption of family detention [4] [5] [1]. However, the documents provided here do not include a published, peer‑reviewed national time series explicitly reporting average or median lengths of stay for children from 2016–2025; instead, researchers must assemble that statistic from facility‑level and booking records, a task complicated by dataset gaps, format changes, and FOIA delays that the Marshall Project and Deportation Data Project explicitly highlight [6] [3].
5. Alternative perspectives, caveats, and institutional agendas
Vera frames its work as transparency and harm‑mitigation research and has explicitly criticized proposals like Project 2025 as pathways to mass deportations and family separation, an advocacy posture readers should weigh when interpreting its findings [9]. ICE’s own public counts—cited by Vera reporting when available—offer contemporaneous tallies (for example, a public ICE report showing ~61,226 detained on a single August 2025 date cited by Vera), but independent compilers warn that differences in dataset compilation can produce discontinuities, so apparent jumps may reflect both policy change and reporting artifacts [1] [4]. The Marshall Project underscores that delayed or stalled FOIA releases can impede independent verification of recent trends [3].
Conclusion: independent analyses document a clear operational decline in family detention earlier in the period and a pronounced resumption and expansion in mid‑2025, and they surface cases and facility counts showing very long detentions; nonetheless, a single, definitive national statistic for children’s average length of stay across 2016–2025 is not published in the provided sources and must be derived from the underlying Deportation Data Project and Vera facility records while accounting for dataset discontinuities and FOIA constraints [4] [5] [2] [3] [1].