The Center for Immigration Studies (CIS) recently released a short report by Sean Kennedy, Jason Richwine, and Steven Camarota on illegal immigrant crime rates in Texas. The CIS report was previewed in a segment on Tucker Carlson Tonight on Tuesday, October 11^th, and claims Cato’s original research on illegal immigrant criminal conviction and arrest rates in Texas is flawed. CIS produced their results by using updated and higher Texas data on criminal convictions of illegal immigrants that don’t appear in the most recent crime data. CIS also reduced the size of the illegal immigrant population in Texas that mechanically resulted in a higher crime rate in their analysis, contradicting their own research elsewhere that finds a higher population of illegal immigrants.

In 2018, Carlson let Peter Kirsanow give a monologue criticizing my work on his show. Carlson didn’t invite me to appear on the segment, so I responded in a blog post. Carlson has invited me on his show several times before, but he has hosted two guests to criticize my work without inviting me to respond. I’ll respond here.

Cato’s Research on Illegal Immigrant Crime in Texas

My research on immigration and crime in Texas uses Texas Department of Public Safety (DPS) data on arrests and convictions by immigration status and my own estimates of the state-level illegal immigrant and legal immigrant populations. My most recent research on immigrant crime in Texas used population numbers estimated using methods developed by Christian Gunadi, an economist and scholar at the University of California San Diego. I calculated the crime rate per 100,000 for each subpopulation by dividing the number of criminal convictions by the size of the subpopulation multiplied by 100,000.

Cato’s research on the illegal immigrant criminal conviction rate in Texas broadly lines up with research by other scholars and our own work estimating the number of incarcerated illegal immigrants nationwide.

CIS’s Research on Illegal Immigrant Crime in Texas

CIS found different results by changing both the numerator and the denominator in its equation. Their researchers increased the number of illegal immigrant convictions in ways that don’t match recent Texas DPS data and they chose low estimates of the number of illegal immigrants living in Texas. Their reported increase in the number of illegal immigrant convictions, the numerator, is in the realm of possibility given how Texas updates its crime data over time, as I wrote in my previous research on this topic, but more recent data I received from Texas DPS does not match their numbers. However, CIS’s choice of a lower denominator is dubious and contradicts their own estimates of the illegal immigrant population in other CIS papers.

Different Criminal Conviction Counts – CIS is Contradicted by Texas DPS

CIS reported that there were 56 convictions of illegal immigrants for homicide in 2018, compared to my data from the Texas DPS that say 46 illegal immigrants were convicted for the same crime in the same year – a difference of 10. Homicide is the most serious crime, but a reallocation of 10 homicides is just above one percent of the 867 total homicide convictions in Texas in 2018. It’s a small revision. In my 2021 Cato Brief on the illegal immigrant crime rate in Texas, I wrote about how the state updates its crime data over time to correct for errors. I ran a robustness check that still produced lower illegal immigrant rates of conviction relative to native-born Americans. CIS extended my robustness check with slightly different data. Relative to Cato’s work, CIS increased the number of illegal immigrant criminal convictions for homicide by 21.7 percent – 10 more individual convictions out of 867 in Texas in 2018. That puts in perspective how relatively few illegal immigrants are convicted of homicide in Texas.

CIS’s new data on illegal immigrant criminal convictions don’t match more recent Texas DPS data. I received updated data on convictions by immigration status from Texas DPS in April 2022 (the data sets were created in March 2022 by Texas DPS). Every year I request updated Texas DPS data on criminal convictions, individuals convicted, arrests, and individuals arrested by crime, county, month, year, and other variables, so I can check my earlier work and prepare for future projects. I analyzed those more recently compiled numbers for the year 2018 and found 46 homicide convictions of illegal immigrants – the exact same number I had already reported and 10 fewer than what the CIS authors say they found.

Curious, I checked the same most recent data from Texas DPS for 2015, the first year examined in my series of papers on this topic. The most recent Texas DPS data I received just six months ago show that illegal immigrants were convicted of just 42 homicides in 2015, four fewer convictions than I reported when I published my first paper on the topic. CIS reported a 17.4 percent increase in homicide convictions in 2015 compared to the numbers I reported for the same year, but Texas DPS’s most recent 2015 data show an 8.7 percent decline in the number of homicide convictions for illegal immigrants. Thus, following CIS’s advice and looking at the most updated Texas DPS data for a few earlier years, the largest revision I found was fewer homicide convictions of illegal immigrants in 2015. In both 2015 and 2018, Texas DPS reported one and three more homicide convictions of legal immigrants, respectively. CIS is reporting much larger revisions in the data that they acquired in 2021 that do not show up in the data I received in April 2022.

CIS’s authors argue that there is a systematic undercount of illegal immigrant prisoners in the Texas criminal justice system. That could be, but it hasn’t shown up in my most recent data request from the Texas DPS. However, I did unexpectedly find a revised decline in illegal immigrant homicide convictions in 2015. That shows that immigration-related errors in Texas crime data don’t always undercount illegal immigrant convictions, sometimes they overcount. As my colleague David Bier has noted, DHS misidentified at least 228 U.S. citizens as illegal immigrants in just one Texas county from 2005 to 2017. Texas DPS may have serious problem misidentifying legal immigrants if the revisions I identified indicate a general trend. I can’t trust CIS’s analysis of their most recent data because it contradicts what I received from Texas DPS just six months ago. CIS wrote that they received their data in 2021. The most recent data disagree with CIS.

Different Illegal Immigrant Population Estimates – CIS Contradicts Itself

The CIS crime report uses the second lowest available estimate of the illegal immigrant population in Texas provided by the Center for Migration Studies (CMS), which also produces the lowest nationwide estimate. A lower illegal immigrant population mechanically results in a higher illegal immigrant crime rate by reducing the denominator (assuming the numerator stays the same or increases). Oddly, CIS did not use their own estimates of the illegal immigrant population that they produced elsewhere and instead chose to rely on the far lower CMS population estimates. What’s even more odd about CIS’s choice to ignore their own pure population research on the number of illegal immigrants in their crime paper is that both pieces of research are coauthored by Steven Camarota.

CIS’s own research on the size of the illegal immigrant population in their pure population research paper estimates a nationwide illegal immigrant population of 11,390,000 in 2018 and 11,480,000 in 2019, compared to CMS’s estimate of 10,565,000 in 2018 and 10,348,884 in 2019 – a difference of 825,000 in 2018 and over 1.1 million in 2019. In other words, in its paper focused on illegal immigrant population estimates, CIS estimates a nationwide illegal immigrant population that is eight percent higher than CMS in 2018 and 10.9 percent higher in 2019. Yet, the CIS authors used CMS’ lower illegal immigrant numbers for its paper on illegal immigrant crime rates in Texas. CIS’s pure population estimates imply a Texas illegal immigrant population of 1,940,000 (what DHS found using the same methods) but CMS found 1,781,752. CIS thus used a population estimate for the number of illegal immigrants in Texas that is 7.5 percent below their pure population estimates in 2018 and 8.9 percent lower in 2019.

Cato’s own estimate of the illegal immigrant population is based on a method developed by Christian Gunadi that found 10,995,696 illegal immigrants in the United States and 1,856,421 illegal immigrants in Texas in 2018. Compared to the CMS population estimates that CIS used for their crime paper, Cato’s estimates are about 3.4 percent higher in Texas and 3.7 percent higher nationwide in 2018. For 2019, our nationwide estimates are 5.6 percent higher and our Texas estimates are five percent higher than CMS’s estimates. Cato did use the CMS data in an earlier brief on illegal immigrant crime in Texas, as well as another estimate by the Department of Homeland Security (DHS), but that was before we discovered and used the superior Gunadi method.

In their pure population research, CIS bragged that their estimates of the illegal immigrant population were in line with DHS’s own population estimates. CIS’s pure population research didn’t break out their estimates by state, but DHS did. Since the DHS and CIS methods are nearly identical (they use different data sources), I can use DHS’ Texas-level estimates in the following example. CIS’s slightly higher counts of illegal immigrant crime, coupled with with CIS’s pure population research that implies a Texas illegal immigrant population of 1,940,000 in 2018, reveals that illegal immigrant homicide rates are below what Cato’s illegal immigrant population estimation methods would have been if compared to CIS’s crime data (Figure 1). CIS’s pure population research, combined with their illegal immigrant homicide convictions data, produces a homicide rate of 2.9 per 100,000 illegal immigrants compared with Cato’s three per 100,000. Both rates are below Cato’s native-born American homicide rate in Texas in 2018.

When CIS wants to show the illegal immigrant population is high, they use the highest nationwide and state-level estimates that they and others have produced. When CIS wants to show a high illegal immigrant crime rate in Texas, they choose the second lowest estimates of the illegal immigrant population for Texas, which also implies the lowest nationwide population number for illegal immigrants. This inconsistency in CIS’s own research is noteworthy.

Tucker Carlson was quick to support CIS’s research on illegal immigrant crime in Texas that was calculated with some of the lowest illegal immigrant population estimates available. But Carlson has also stated there are likely many more illegal immigrants than the common estimate of roughly 11 million. This matters because a larger illegal immigrant population mathematically results in a lower illegal immigrant crime rate. So, which is it? Are there many more illegal immigrants than reported, or do they have a crime rate similar to native-born Americans? You can only believe one of those given the data on illegal immigrant crime.

It would be great to hear why CIS’s authors change their estimates of the illegal immigrant population based on the type of study they’re publishing.

Another Wrinkle – CIS vs. PNAS

CIS also criticized a paper published in the Proceedings of the National Academy of Sciences (PNAS) that finds an even lower illegal immigrant crime rate than my own research. The authors of the PNAS paper conclude that “many legal noncitizen and foreign-born arrestees are not recorded in” government databases. When they corrected for that, the legal immigrant criminal conviction rate was frequently higher than that of illegal immigrants. Fortunately, the authors of the PNAS paper shared their methods and data in detail, so curious readers can evaluate them.

Conclusion

Relative to my Cato research, CIS made two data changes to estimate higher illegal immigrant criminal conviction rates in Texas. The first change was to use updated and higher Texas DPS data on criminal convictions of illegal immigrants – changes that I can’t replicate in more recent and up to date Texas DPS data. The second change was to reduce the size of the illegal immigrant population. They chose the second lowest Texas illegal immigrant population estimate published, which contradicts CIS’s own pure population research elsewhere that finds a much higher illegal immigrant population in Texas and nationwide.

In other words, CIS found their new criminal conviction estimates by increasing the number of convictions of illegal immigrants in a way that doesn’t match the most recent Texas DPS data but that could be true and implausibly reducing the population of illegal immigrants below what their own research suggests. They increased the numerator and decreased the denominator to produce a higher crime rate. Increasing the numerator doesn’t appear to be correct but it’s not an implausible adjustment. Reducing the denominator below CIS’s own estimates of the illegal immigrant population is not plausible. Using CIS’s homicide conviction data and its own pure population estimates narrows the conviction rate gap between illegal immigrants and native-born Americans in Texas, but it does not vanish and it certainly doesn’t show that illegal immigrants have a higher homicide rate relative to native-born Americans in Texas.