What methodologies do watchdogs (TRAC, American Immigration Council, Cato) use to classify detainees by criminal history and citizenship, and how do they differ?
Executive summary
TRAC builds its detainee classifications on ICE transactional records obtained via FOIA and presents categories by conviction status, pending charges, and immigration-only violations using database tools it maintains [1] [2]. The American Immigration Council analyzes ICE/TRAC data to add legal-context classifications—such as “aggravated felon” and offense-level groupings—while Cato ingests ICE’s public statistics to produce high‑level share calculations and trend claims; the three differ in source access, granularity, and how they treat expedited‑removal or non‑criminal immigration cases [3] [4] [5] [6] [7].
1. How TRAC constructs its detainee categories and what that means
TRAC’s methodology relies on FOIA‑obtained ICE detention and transactional records, which it ingests into searchable tools and dashboards that report detainees by country of citizenship, conviction status, pending charges, and other attributes; TRAC explicitly notes its data come from ICE records and provides user tools to slice results temporally and by offense level [1] [2]. TRAC’s public “quick facts” and reports have repeatedly shown a rising share of detainees with no criminal convictions and its datasets are the basis for many secondary analyses [8] [9]. Because TRAC mirrors ICE’s operational fields, its classifications depend on how ICE labels convictions, charges, and immigration status — meaning TRAC’s “no criminal conviction” bucket reflects ICE records rather than independent court‑level adjudication [1] [7].
2. American Immigration Council: adding legal context and offense levels
The American Immigration Council draws on ICE/ TRAC records but layers legal analysis—examining gender, region, entry status, and whether ICE classified someone as an aggravated felon—and reports offense‑level breakdowns (for example, “Level 1” serious offenses like assault, burglary, drug trafficking) to show the composition of detention populations [3] [4]. The Council emphasizes legal definitions and civil-vs-criminal distinctions—pointing out that immigration detention is civil even when it occurs in jail‑like settings—and uses TRAC’s underlying transactional data to map who ICE detains under different statutory categories [3] [4]. That gives the Council more interpretive framing than TRAC’s raw tools, but it remains dependent on ICE/TRAC field codings for convictions and classifications [3] [4].
3. Cato’s approach: public ICE data, percentage claims, and trend emphasis
Cato’s analyses typically use ICE’s public statistics and related datasets to compute headline percentages—such as the share of detainees with no criminal convictions or the fraction with violent offenses—and to highlight shifts over short time periods, producing claims like “65% had no convictions” or that the share of violent convictions is small [5] [6] [10]. Cato focuses on high‑level policy implications (costs, public‑safety rationale) and often aggregates across ICE reporting windows to make comparative claims about pre‑ and post‑policy periods [5] [6]. Because Cato commonly uses ICE’s public dashboards and selected time slices, its findings can differ from TRAC’s more granular, transaction‑level reconstructions.
4. Key methodological differences and consequential ambiguities
The major differences arise from data access (TRAC’s FOIA transactional records vs Cato’s use of ICE public dashboards), analytic layering (American Immigration Council’s legal framing vs TRAC’s raw tools), and time‑window selection; these create diverging headline numbers even when all rely on ICE inputs [2] [3] [5]. A persistent ambiguity is how expedited‑removal cases, credible‑fear detentions, and temporary family‑unit detentions are classified: critics note TRAC’s “no criminal history” group may include people in expedited removal or credible‑fear processing, which affects interpretation of “non‑criminal” claims [11]. Likewise, ICE’s own reporting categories (conviction vs pending charge vs immigration violation) shape every downstream analysis [7].
5. What to take away for comparability and policy debates
When comparing claims, readers should check: the underlying data source (FOIA transactional dump versus ICE dashboard), whether the analyst filters expedited‑removal or parole cases, the offense‑level schema (CIC’s Level 1 vs simple “violent/non‑violent” tallies), and the time frame used for trend claims; differences in any of these account for most apparent contradictions between TRAC, American Immigration Council, and Cato reporting [1] [3] [5]. Reporters and policymakers must therefore treat percentages as methodology‑dependent and consult original ICE fields or TRAC’s tools to reconcile conflicting summaries [2] [7].