President Donald Trump never misses an opportunity to depict unauthorized immigrants—especially of the Hispanic variety—as “rapists and criminals.” He did it again in his State of the Union address when he drew attention to two Long Island teenage girls killed by the El Salvadorian gang MS13. Those deaths are tragic, but they don’t say much one way or the other about the propensity of these immigrants to commit crimes.

You wouldn’t, however, know that from restrictionist pundits who are working overtime to sell the “illegal immigrants are criminals” narrative. A case in point is former US Civil Rights Commission member Peter Kirsanow’s recent piece in National Review purporting to show that these immigrants are more likely to commit crimes than the native born. But Kirsanow uses incomplete and cherry-picked data—and makes rookie mistakes in interpreting it to boot—that eviscerate the credibility of his case.

Kirsanow is correct that most of the disagreements over the criminality of undocumented immigrants could be resolved by better data. But that doesn’t absolve us from accurately reading the data we do have. Kirsanow, however, does not. His entire case is based on a gross misreading of the 2011 Government Accountability Office (GAO) report on the State Criminal Alien Assistance Program (SCAAP), a federal program that partially reimburses states and localities for the cost of incarcerating certain criminal aliens.

The SCAAP report shows that in 2009, there were 295,959 criminal aliens incarcerated in state and local prisons at any given time that year. From this number, he subtracts those in the country legally and assumes that the balance gives one the total number of illegal immigrants incarcerated that year. He compares that number with the population of illegals in various states to estimate their crime rates. Then he compares that rate with the crime rate of citizens to come up with a massively inflated “incarceration rate” of these aliens.

But here’s the problem with his analysis:

Kirsanow assumed, as some others before him with only a passing familiarity with these databases, that the 295,959 figure refers to the number of individuals incarcerated. In fact, it is the total number of incarcerations. In other words, if a criminal alien was incarcerated for 10 short sentences, released after each one, and then re-incarcerated, then that single alien would account for 10 incarcerations under the SCAAP figure for that year. But Kirsnaow counts that as 10 individuals.

However, when it comes to estimating the incarceration rate of natives, Kirsanow compares the number of individuals incarcerated with their total population. This nonsensical apples-to-oranges comparison yields an exceedingly unfavorable “incarceration rate” for undocumented immigrants.

Kirsanow failed to appreciate that the purpose of the GAO report was to estimate the reimbursement that Uncle Sam owes state and local governments for incarcerating criminal illegal immigrants. Thus, the agency was only interested in the total number of incarcerations over the course of a year. It didn’t care to separate out the number of offenses from the number of offenders. That is why the GAO report is nearly worthless for any scholarly attempt to estimate illegal immigrant crime rates.

A quick look at American Community Survey (ACS) data further confirms just how out-of-line Kirsanow’s estimate is. (The ACS is an annual mini-census that, among other things, gathers information about prisoners in adult correctional facilities. It doesn’t report on the broad legal status of immigrants but does indicate whether they are American citizens and their country of birth, making it possible to separate immigrants from Americans.)

For 2008, the ACS reported that there were 156,329 non-citizens incarcerated in all three—federal, state, and local—adult correctional facilities. This is only half of the 296,959 incarcerations that SCAAP reports in just state and local prisons making it logically impossible for the 296,959 figure to be referring to the total number of criminal aliens incarcerated.

Kirsanow is merely an individual whose analysis can be discounted. But there is no discounting the Alien Incarceration Report jointly released by the Departments of Justice (DOJ) and Homeland Security (DHS) last December. It too misrepresented data when it estimated that “one-in-five of all persons in the [federal] Bureau of Prisons custody were foreign born, and that 94 percent of confirmed aliens in custody were unlawfully present.” That seems shockingly high as illegal immigrants are, at most, about 4 percent of the population. But if this report were right, they would be 19 percent of all prisoners.

But the report had no solid basis for its conclusion because it did not have all the prison data. If you scroll down beyond the report’s press release and Summary of Findings, it admits as much. It notes:

This report does not include data on the foreign-born or alien populations in state prisons and local jails because state and local facilities do not routinely provide DHS or DOJ with comprehensive information about their inmates and detainees. This limitation is noteworthy because state and local facilities account for approximately 90 percent of the total U.S. incarcerated population. DHS and DOJ are working to develop a reliable methodology for estimating the status of state and local incarcerated populations in future reports.

Of course that didn’t stop Fox News and other similar outfits from using it to peddle their “illegal immigrants are hardened criminal” line.

It is really important to bear in mind that the federal prison population is not representative of the incarcerated populations in state and local prisons. That’s because federal prisons house illegal immigrants who commit immigration offenses. The ones who commit more serious crimes tend to be housed in state adult correctional facilities.

Only 85 total people who were convicted of murder were sentenced to federal prison in 2016. But the total number of murder convictions nationwide that year was 17,785. Clearly, only a small fraction of a percent of all murderers are incarcerated in federal prisons so most undocumented immigrants in these facilities are not hardened criminals.

As Kirsanow acknowledged, the government doesn’t keep good data on illegal immigrant incarcerations in state correctional facilities. But the data we do have suggests that they are actually much less crime-prone than native-born Americans.

The Texas Tribune reported, after examining data obtained from the Texas Department of Criminal Justice, that illegal immigrants are underrepresented in local jails. They are only 4.6 percent of Texas inmates while they make up 6.3 percent of that state’s total population.

Some academic researchers have examined quasi-natural policy shifts to see how crime rates change due to more intense immigration enforcement. If illegal immigrants are more crime-prone, then more aggressive immigration enforcement in an area should lower crime rates. But they found no overall reduction. This suggests, at a minimum, that illegal immigrants’ crime rate is no higher than that of the broader population.

More recent research conduced by Michelangelo Landgrave and me finds similar results. We applied a statistical technique that is used to figure out the employment, age, and occupations of immigrants in the census to the incarcerated population data in the American Community Survey. This allowed us to estimate the percentage of illegals among the incarcerated. We found that even if one includes in the mix those in detention facilities—most whom are there for immigration-related offenses—illegal immigrants are 44 percent less likely to be incarcerated than native-born Americans. Excluding those in immigration detention yields an incarceration rate that is almost identical to that of legal immigrants: A dramatic 69 percent lower than that of natives.

Restrictionists want the public to believe that undocumented immigrants are criminals in order to justify harsh enforcement policies and crackdowns. But before America goes down this draconian path, it is vital that it gets the facts straight. Taking amateur analyses or government spin at face value will hurt peaceful and hardworking immigrants without making Americans safer.