The numbers the government is citing don’t match the numbers researchers are actually finding. That gap — wide, significant, and growing — is now at the center of a fierce debate over how the Trump administration is characterizing its immigration enforcement surge.
A close examination of federal detention data, independent analyses, and court records reveals a portrait of ICE enforcement in 2025 and early 2026 that looks strikingly different from the picture painted in recent Department of Homeland Security press materials. While DHS has promoted dramatic statistics about threats against its agents, the underlying data on who is actually being arrested and detained tells a more complicated — and, to immigration researchers, more alarming — story.
The Conviction Gap Nobody’s Talking About
Here’s the number that keeps surfacing in independent reviews: as of February 7, 2026, 73.6% of the 68,289 people held in ICE detention had no criminal conviction of any kind. Not a violent offense. Not a drug charge. Nothing. That figure, drawn from federal custody data compiled by the Transactional Records Access Clearinghouse at Syracuse University, cuts directly against the administration’s repeated framing of its enforcement targets as dangerous criminals.
It’s not that there are no individuals with serious records in the system — there are. But the share of people with criminal convictions has been quietly shrinking, not growing. Fewer than one in three people in ICE custody hold a criminal conviction under any standard definition. And within that group, arrests of people convicted specifically of violent crimes have actually declined: in early October 2025, ICE arrested 409 people convicted of violent offenses, compared to 502 in early June — a drop of roughly 20%.
That’s the catch. The administration has been loudest about enforcement during a period when the violent-crime-conviction subset of its arrests has been quietly contracting.
Community Arrests Are Surging — And That’s the Point
So who is being picked up? The answer, increasingly, is people encountered not inside jails or prisons but out in communities — at workplaces, near schools, outside churches. The shift is documented in state-level data as well as federal records. In Connecticut alone, ICE apprehended 319 people in 2025 who carried only pending charges — not convictions — compared to just 45 in 2024. That’s a 600% increase in arrests of people who, legally speaking, remain innocent until proven guilty.
Nationally, the pattern holds. The volume of so-called “at-large” arrests — people detained away from any correctional facility — rose substantially from 2024 to 2025, a trend confirmed by both government releases and independent tracking organizations. Critics argue this represents a fundamental reorientation of enforcement priorities. Supporters say it reflects a broader mandate to restore interior enforcement that had been scaled back under previous administrations. Both things can be true at once, and that tension is exactly what makes this debate so difficult to resolve with a single statistic.
The Statistics DHS Is Promoting
The DHS press materials in circulation have cited extraordinary figures — including claims of a more than 1,300% increase in assaults on ICE officers and an 8,000% increase in death threats. Those are numbers that, if accurate, would represent a genuine public safety crisis for federal law enforcement personnel. But here’s the problem: the sourcing behind those figures is opaque, the baseline years aren’t consistently specified, and no independent dataset reviewed for this article corroborates the specific percentages cited.
That doesn’t mean assaults on agents haven’t increased — hostility toward federal enforcement operations has, by multiple accounts, intensified in certain jurisdictions. Still, extraordinary percentage claims require extraordinary documentation. When the denominator is small enough, almost any uptick produces a dramatic-looking percentage. A rise from two incidents to 28 is, technically, a 1,300% increase. Without the raw numbers, the figure is essentially unverifiable — and that’s precisely the kind of statistical presentation that tends to collapse under scrutiny.
A Credibility Problem With Consequences
Does any of this mean the administration’s broader enforcement push lacks public support? Not necessarily. Polls consistently show that a significant portion of Americans support stricter immigration enforcement, and the political mandate behind the current surge is real. The question isn’t whether enforcement is happening — it plainly is, at a scale not seen in decades. The question is whether the statistical case being made to justify and describe that enforcement holds up.
When governments overstate or misframe data — even in service of policies that might otherwise be defensible — they create a credibility deficit that’s hard to claw back. Journalists, lawyers, and legislators who probe the numbers and find them wanting don’t just question the statistics. They start questioning everything.
And in an enforcement environment this aggressive, with this many lives directly affected, that’s not a small problem. It’s the whole ballgame.
