Trump Says the National Guard Cut Crime in D.C. The Study He's Citing Says Otherwise.

The Niskanen Center, a nonpartisan D.C. think tank, published a study on June 4, 2026 treating the August 2025 Guard deployment to Washington as a natural experiment — comparing crime trends before and after troops arrived ^[2]. Its finding: the National Guard had "no effect on violent crimes, including robberies," in the city ^[2]. What it did do was cut opportunistic property crime — theft and vehicle break-ins — by 24%, because visible troops deterred crimes of convenience but weren't stationed where interpersonal violence actually happens ^[2][4].

Researcher Richard Hahn, who co-authored the study, said the deployment was "a blunt and expensive instrument": guard members were concentrated in "highly visible commercial, transit, and tourist areas rather than high-crime neighborhoods," which explains why they moved property-crime numbers but not violent-crime numbers ^[2][4]. He told NPR: "You could get the same or better outcomes, possibly much better outcomes, for much cheaper, if you just were very thoughtful about policing" ^[2].

The price tag makes the mismatch sharper. The five-month D.C. deployment (August–December 2025) cost roughly $185 million, or about $607 per Guard member per day — nearly double the $384 a day the Metropolitan Police Department spends per officer ^[4]. A cost-benefit analysis found the Guard deployment returned "less than $1 for every dollar spent," while ordinary MPD policing during the same window generated an estimated $2.2 billion in societal benefit ^[4]. Crime analyst Jeff Asher noted the D.C. drop wasn't even unusual: "This is happening everywhere" — a 9.3% national violent-crime decline was already underway before the Guard arrived, in "dozens of places seeing similar drops without any sort of heavy-handed intervention" ^[4].

Criminologists are also pushing back on the idea that any single policy — federal or otherwise — explains the national numbers. Adam Gelb of the Council on Criminal Justice told CBS News: "The homicide drop is historic and crystal clear. But right now, what's driving it is about as clear as a glass of milk" ^[3]. He pointed out that crime is falling "in cities with very different public safety strategies, economic conditions, and political leadership," which cuts against crediting one administration's tactics over the dozens of factors — from technology to demographic shifts — that researchers say are plausibly at play ^[3].

The claim isn't a fringe rumor — it's coming from the White House itself and is being amplified through official channels, campaign events, and cable coverage of the Nassau County speech, where Trump also used the crime numbers to rally support for a New York gubernatorial candidate and to attack Gov. Kathy Hochul's sanctuary-city and cashless-bail policies ^[1]. The Associated Press described the event, held at a police academy, as "overwhelmingly political" despite the law-enforcement setting ^[1].

That's what makes this different from an anonymous hoax: it's a real, high-profile statistic (the FBI drop is genuine) stretched into an unsupported causal story (federal deployment caused it), then repeated in political messaging ahead of the November midterms without the caveats researchers who actually study the deployments have attached to it.

MISLEADING. The underlying FBI statistic is real and historic — violent crime and murder did fall sharply nationwide in 2025 ^[1][3]. But the specific claim being pushed alongside it, that National Guard deployments to cities like D.C. drove that drop, is contradicted by the only rigorous independent study of the D.C. deployment, which found no measurable effect on violent crime and attributed the modest gains that did occur to ordinary policing changes and a property-crime deterrent effect, not troop presence ^[2][4]. Experts who track national crime trends say the true drivers are unclear and almost certainly multiple, making any single-cause credit claim — Trump's included — an overstatement of what the data can show ^[3][4].