Immigration and Crime: A Mini Data Dive


This post is longish and doesn’t really have much payoff at the end. It’s just something that turned into a bit of snark hunt, so I figured I’d document it. You have been warned.

It starts with a column by Mona Chalabi, the Guardian’s “data editor,” which claims to outline her research on the question of whether illegal immigrants commit more crimes than native-born Americans. It’s faintly ridiculous and I’m a little annoyed by it, but then I come to this:

I find a study by Bianca E Bersani. I look her up — she’s a associate professor of sociology at the University of Massachusetts, Boston. Using numbers from the Bureau of Labor Statistics, her study finds that about 17% of all first-generation immigrants who are age 16 have committed a crime in the past 12 months….But wait. Is that number high or low? I decide to find out how often native-born people in the US commit crimes. Luckily, her study has that too. It’s higher: about 25% of all native-born people in the US who are 16 have committed a crime in the past 12 months.

That seems kind of high, doesn’t it? Then again, “committed a crime” could encompass things like smoking a joint or stealing a box of paper clips from school, so who knows? The data comes from a paper called “A Game of Catch-Up? The Offending Experience of Second-Generation Immigrants,” so I check it out. But there’s nothing there. The paper has nothing whatsoever to say about either 16-year-olds or first-generation immigrants. What’s going on? Here’s the chart Chalabi presents:

This is a little odd. It suggests that 25 percent of 16-year-olds have committed a crime in the past year, but only 20% of 17-year-olds. That doesn’t jibe with what I know about crime rates. And the source is Pew Research. So let’s go look at the Pew article. It’s a lengthy description of Bersani’s article, and it includes this chart:

This is odd again. It’s the same chart, all right, and the author spends a lot of time describing “A Game of Catch-Up?” But as I mentioned above, that article contains nothing like this at all. What’s more, it appeared in Crime and Delinquency, but the chart is sourced to Justice Quarterly.

So now it’s off to Justice Quarterly. It turns out that everyone is describing the wrong article. I wonder if any of them actually read it? The correct article is “An Examination of First and Second Generation Immigrant Offending Trajectories,” also by Bianca Bersani. Fine. What does that article say? Here is Bersani’s chart, colorized for your viewing enjoyment:

It appears that everyone has been copying the chart properly. For what it’s worth, though, I’d make a few comments:

  • This data is for all immigrants. Donald Trump’s focus is solely on illegal immigrant crime.
     
  • Bersani’s data is from 1997-2005. That’s pretty old. Crime and arrest rates of juveniles have gone down more than 50 percent since then, and the population of illegal immigrants has gone up more than 50 percent since then. I don’t know if that changes the relative values in this chart, but it would certainly change the absolute values.
     
  • The data comes from the 1997 National Longitudinal Survey of Youth, which uses a very large oversample of Hispanic and black youth. Bersani appears to be using the full sample, and since Hispanic and black adolescents commit crimes at higher rates than whites, it means the numbers for native-born Americans are exaggerated. At a guess, the real figures are 2-3 percentage points lower.
     
  • The NLSY97 data includes six types of crime that were included in Bersani’s study: (1) damaging property, (2 and 3) stealing less or more than $50, (4) other property crimes, (5) assault/serious fighting, and (6) selling drugs. By far the biggest contributors were property damage and petty theft, with fighting in third place and the others far behind. Auto theft and using a gun to steal (not included in Bersani’s study) were minuscule:
  • Since the vast majority of the crimes in this study are minor—and we can assume that serious violent crime is even less prevalent—it’s not clear how much this tells us. I don’t think anyone cares much whether immigrant teenagers steal six packs of beer at a greater rate than native-born Americans. We mainly care about more serious violent crimes: robbery, rape, murder, and aggravated assault. Those aren’t addressed at all.

I’d add that Bersani didn’t just add up all the crimes committed by various groups. Her methodology is pretty impenetrable to anyone who’s not an expert:

I use group-based trajectory modeling…identifies clusters of individuals who display similar behavioral trajectories over a period of time…Nagin and Land’s (1993) semiparametric group-based modeling approach…estimated using a zero-inflated Poisson form of a group-based trajectory model:

where ln(kjit) is the natural logarithm of the number of total crimes for persons i in group j at each age t. The equation specified above follows a quadratic function of age (age and age2)….

I have no idea what this means or whether it’s appropriate, but I’m a little skeptical about a model that suggests that 17- and 18-year-olds commit crimes at lower rates than 16-year-olds. Most crime data I’ve seen shows the opposite. Then again, most crime data doesn’t include extremely minor crimes like shoplifting and property destruction. It’s possible that adolescents age out of that stuff pretty early.

Long story short, I wouldn’t draw too many conclusions from this study. The data is old; it’s not limited to illegal immigrants; it looks only at adolescents; the crimes under consideration are pretty minor; and the methodology is probably OK, but who knows? Put it all together, and I’d say it doesn’t tell us too much one way or the other about the serious crime rate of illegal immigrants as a whole.

I have yet to see a study that persuasively suggests a higher crime rate for immigrants than for anyone else. Let’s face it: if there’s anything we native-born Americans excel at, it’s crime. That said, the Guardian’s data editor should have known better. There are tons of studies out there that try to estimate the relative crime rates of native-born Americans compared to undocumented immigrants, and cherry picking this particular one makes no sense. It does provide a rough data point suggesting that crime rates of immigrants aren’t any different from the rest of the population, but it’s nowhere near the best study out there. Citing this one and calling it a day is a real disservice.