What policy, enforcement, and selection biases could affect comparative crime‑rate estimates between undocumented immigrants and U.S.-born citizens?
Executive summary
Comparisons of crime rates between undocumented immigrants and U.S.-born citizens are shaped less by a single truth than by a tangle of measurement choices, enforcement practices, and who is included or visible in the data—factors that can push observed arrest and incarceration rates in different directions depending on methodology and locale [1] [2]. Scholars who find lower arrest or incarceration rates for immigrants point to selection, deterrence, and robustness checks, but they also acknowledge that arrest-based measures reflect law-enforcement activity as much as underlying offending [1] [2].
1. Measurement choice: arrests, convictions, or incarceration change the story
Studies using arrest records—like the Texas DPS dataset used in PNAS—report substantially lower felony arrest rates for undocumented immigrants than for U.S.-born citizens, and authors stress robustness to using convictions instead of arrests [1] [3]. Government summaries of that work highlight numeric differences—213 per 100,000 violent felony arrests for U.S.-born citizens versus about 96.2 per 100,000 for undocumented immigrants in Texas—yet also warn that arrest counts are an imperfect proxy because they partly reflect policing patterns [2]. Academic work on long-run incarceration finds immigrants have lower incarceration rates historically, but incarceration itself can under- or over-represent true offending depending on charging, plea bargaining, and sentencing disparities [4] [5].
2. Enforcement intensity and policing priorities bias comparisons
Local variation in policing—how aggressively police enforce drug, property, or immigration-adjacent offenses—shapes who appears in arrest statistics; researchers note immigrants may have less law‑enforcement contact in some settings and that sanctuary policies can alter cooperation with federal authorities without increasing crime [6] [7]. The Texas study benefits from "complete information for every jailable arrest" in that state, improving comparability there, but authors and NIJ cautioned that findings might not generalize to places with different enforcement regimes [1] [2].
3. Immigration enforcement creates deterrence and detection effects
Researchers propose a deterrence mechanism: undocumented immigrants face harsher non-criminal consequences—chiefly deportation—so they may be less likely to offend or to be visible to police, which would lower arrest rates relative to citizens [1]. At the same time, immigration enforcement programs like Secure Communities can increase detection of noncitizens when they are arrested, potentially inflating relative arrest counts where those programs are active; Texas researchers attempted to address such biases through alternative classifications and population estimates [8] [3].
4. Selection bias among immigrants: self‑selection and survivorship
Multiple studies argue immigrants are a positively selected group—healthier, more risk-averse, and often motivated by economic opportunity—traits associated with lower criminality, and long-run data show immigrants have not historically had higher incarceration than the U.S.-born [1] [5] [4]. However, selection can cut both ways: those remaining undocumented may differ from those who naturalize or leave, and researchers acknowledge that varying immigrant composition across places and time complicates simple comparisons [1] [9].
5. Population denominator uncertainty and sensitivity tests
Estimating the size of the undocumented population is inherently uncertain; the Texas study conducted sensitivity analyses showing that undocumented numbers would have to be far smaller—less than half current estimates—to erase reported differences in violent crime rates, but that conclusion depends on which population estimates are used [1]. Government summaries reiterate that the result is robust to alternative population estimates, yet broader national-level work must grapple with differing methodologies for counting unauthorized residents [3] [2].
6. Legal status classification and miscoding risks
Classifying immigration status at arrest or conviction is fraught: administrative records may misclassify lawful noncitizens, temporary visa holders, or those with pending relief, and different studies use distinct rules for who counts as “undocumented,” which affects rates [3]. The PNAS authors explicitly tried alternate classifications and still found lower offending rates for undocumented people in Texas, but they caution about generalizability beyond that data-rich setting [1] [3].
7. Policy implications and what the evidence cannot settle alone
Empirical patterns—lower arrests and incarcerations for many immigrant groups in multiple studies—do not automatically translate into universal policy prescriptions because enforcement choices, population estimates, and local contexts shape observed rates [7] [6]. Authors and policy analysts alike call for replication across states and attention to measurement limits: arrest data are informative but not definitive about true offending, and different biases can mask or magnify real differences depending on the question and dataset [2] [6].