The right’s too-good-to-be-true proof that immigration drives crime just fell apart

The fraud behind the persistent myth of "more guns, less crime" strikes again.

Attorney General Jeff Sessions CREDIT: Mark Wilson/Getty Images
Attorney General Jeff Sessions CREDIT: Mark Wilson/Getty Images

New research that Attorney General Jeff Sessions used to prop up the Trump administration’s conviction that immigration feeds criminality is based on a basic data error.

The paper Sessions cited is based on the kind of mistake most real scholars take great pains to avoid, as Cato Institute immigration expert Alex Nowrasteh showed Monday. Sessions should have known better than to latch onto the findings, coming as they do from veteran data trickster John Lott — a man notorious for fudging numbers to generate ideologically convenient conclusions. Correcting for the mistake in Lott’s newest work reverses the conclusion that Sessions touted in a speech last Friday.

Lott claimed in a January paper that Arizona’s undocumented immigrant population has been far more likely to get imprisoned than the United States citizenry as a whole. “Undocumented immigrants are at least 142 percent more likely to be convicted of a crime than other Arizonans,” Lott wrote.

Sessions cited Lott’s work in a speech in Virginia just days later, saying undocumented immigrants are dramatically more prone to committing violent crimes. Where Lott had offered a somewhat restrained extrapolation from the Arizona figures to the national population, Sessions proclaimed that “tens of thousands of crimes have been committed in this country that would never have happened if our immigration laws were enforced and respected like they ought to be.”


Those are grand claims, breaking from a longstanding criminologist consensus. High-profile crimes linked by police to an undocumented person may make major anecdotal waves that shape public perception. But in study after study, the actual quantitative reality is that immigrants in general and undocumented ones in particular are less prone to criminality than the average U.S. citizen. Here was the top law enforcement officer of the United States government, telling the public he had proof that his boss was right all along when he portrayed undocumented immigrants as fiendish criminals in the speech that launched his successful campaign for the presidency.

But Lott — and by extension Sessions — are wrong. The share of undocumented immigrants in Arizona prisons is not more then double what overall population demographics there would suggest. It’s actually very slightly lower, Nowrasteh found, “turn[ing] Lott’s finding on its head.”

Here’s where it gets mathy: Lott claimed his incarceration data showed that 11.1 percent of those who went into prison in Arizona in 2014 were undocumented immigrants, while the group accounts for just 4.9 percent of the state’s population. But he abused the data to get to the 11.1 percent figure (and the equivalent annual figure for undocumented people as a share of the newly incarcerated for each year going back to 1985 from his dataset). The variable he used to calculate which convicts were undocumented immigrants and which were citizens or documented legal residents or visitors cannot actually clarify the distinction he said it clarifies.

Disabusing the data, Nowrasteh found that undocumented immigrants were between 3.7 and 4.3 percent of all those sent to prison in Arizona in 2017, the last year of Lott’s dataset. The Cato researcher didn’t change any of the math processes here — he just employed a more correct understanding of what the labels on the data do and do not mean. The data Arizona keeps suggest that at most 4.3 percent of all those admitted to prisons there last year were undocumented.

“Lott thought that ‘non-U.S. citizens and deportable’ describes only illegal immigrants [sic] but it does not,” Nowrasteh wrote, using the nomenclature for undocumented immigrants preferred among conservative commentators. In fact, the data include immigrants of all legal statuses.


“Using another variable… that Lott did not analyze reveals that, at worst, illegal immigrants [sic] in Arizona likely have an incarceration rate lower than their percentage of that state’s population,” Nowrasteh added.

Sessions really should know better by now. Lott’s been torturing numbers until they give up confirmation of his prior ideological convictions since at least the late 1990s. His work on gun safety raised methodological alarms with research peers as far back as 2002. In 2003, he quietly changed a data table in the middle of an argument with his debunkers to make his results appear credible again. That same year, Lott was caught blogging under a fake name to defend his own work and accused of making up an entire research survey that a colleague found no evidence he ever conducted. A lengthy 2014 summary of Lott’s career by Evan DeFilippis and Devin Hughes proclaimed that “[o]n closer inspection his impressive credentials reveal an academic nomad, never able to secure a place in academia.”

Before his grifter tendencies derailed it, Lott’s first career helped build the tentpole arguments that the National Rifle Association and other opponents of gun control legislation still use today. He is almost singlehandedly responsible for the “more guns less crime” claim, which has stuck around long after his work was shown to be bogus nearly two decades ago.

Today, it seems, Lott’s picked up on a new political wind. The nativist stance on immigration Trump espouses needs some kind of rhetorical vector if it is to jump from whatever minority of the population is already eager to hate foreign newcomers and capture enough of the political center to support an election-winning coalition of voters.

As rhetorical vectors go, the fear of violence is about as powerful a tool as exists. But credible academics who have studied crime and immigration for decades have found again and again and again that immigrants — documented or undocumented — are no more likely to commit crime than anyone else, with some analyses finding that first-generation immigrants are in fact less likely to commit crime than the existing “native” population of Americans to whom the Trump movement pedals its frightful yarns.

Lott is flat wrong about the numbers and the data today, just as he was on guns under the last Republican president. But if the trajectory of that story is any guide — and powerful government officials’ willingness to tout Lott’s claims again today is any indication — then disproving his figures will be far too little to curb the influence of his swindles.