How do ICE and CBP define and record 'criminal history' in detention datasets?
Executive summary
ICE and CBP treat “criminal history” differently across operational and public datasets: ICE’s detention files use a simple convicted-yes/no flag and administrative categories that mix immigration-related violations and prior convictions, while CBP’s “criminal alien” counts rely on post-apprehension record checks for prior convictions whether in the U.S. or abroad [1] [2]. Public researchers warn these variables are incomplete, inconsistently populated, and can conflate pending charges, deportation reentry, and immigration civil violations with criminal convictions, producing results that are often misread in media and policy debates [3] [4].
1. How ICE operationally defines “criminal history” in its detention schema
ICE’s detention reporting includes at least one explicit binary variable that indicates whether an individual has a prior criminal conviction—a “Yes/No” field derived from Enforcement Integrated Database (EID) extracts [1]. ICE also publishes prose categories that distinguish people with convictions, people with pending charges, and “other immigration violators” such as re‑entrants or fugitives, meaning the agency’s operational classification blends criminal-conviction status with immigration administrative categories [5] [6].
2. How CBP frames and records convictions at the border
CBP’s public “criminal alien” counts are constructed from post‑apprehension records checks and are defined to include individuals with one or more convictions, domestic or foreign, prior to interdiction; CBP explicitly excludes conduct that is not criminal under U.S. law [2]. CBP collects individual‑level data on every person it arrests or expels at the border, which feeds into separate CBP datasets that researchers reuse alongside ICE data [7].
3. What the public datasets actually record — variables and provenance
Researchers working with ICE and CBP extracts note that the largest ICE files are encounters, detentions, arrests, detainers and removals, and that conviction fields are drawn from EID and then uploaded into reporting systems such as the CBP Unified Immigration Portal (UIP) [4] [1]. The Deportation Data Project and other third‑party repositories republish ICE’s released tables, but they flag frequent missingness in fields like detainer outcomes and conviction indicators and caution that many records reflect transfers from CBP rather than ICE-initiated criminal arrests [8] [4].
4. Key limitations: missing data, mixed categories, and operational scope
Multiple analyst accounts and fact checks emphasize that ICE datasets omit many kinds of arrests (for example HSI criminal arrests) and that CBP and ICE responsibilities differ by geography, so the composition of detained people depends on arresting agency and data lineage; missing fields may indicate the person remained in criminal custody, a detainer was not honored, or ICE had not updated the record [9] [8] [4]. Public summaries that report percentages of detainees “with no criminal convictions” can obscure that nearly a quarter may have pending charges and that many convictions recorded are for minor offenses such as traffic violations, per independent trackers and news analyses [10] [6] [3].
5. How definitions and recording choices shape narratives and policy
Because ICE uses administrative arrest authority for civil immigration violations and only sometimes makes criminal arrests, the presence or absence of a recorded conviction in ICE’s datasets does not alone indicate whether enforcement targets “criminals” or immigration violators—an interpretation that has driven contested headlines and fact checks [9] [3]. CBP’s broader definition of “criminal alien” (convictions in any jurisdiction) and ICE’s mixing of immigration‑specific categories mean cross‑agency aggregates can amplify or downplay the share of detainees with serious violent convictions unless researchers disaggregate by conviction type and arresting agency [2] [6].
6. Where reporting still falls short and what researchers should do
Primary documentation and FOIA‑based reconstructions remain the best way to interpret conviction flags: analysts should join ICE detentions to arrest and detainer tables, check for missing detainer lift reasons, and separate CBP‑apprehended transfers from interior ICE arrests to avoid conflating datasets; several projects and guides illustrate these methods and warn that single‑figure headlines mislead without this work [4] [8] [7]. Public reporting should therefore present the provenance of “criminal history” flags, the proportion of missing values, and whether recorded convictions are domestic, foreign, minor, violent, or pending—details that current agency releases and third‑party repositories document but do not always standardize [1] [4].