What methodology did peer-reviewed studies use to compare crime rates of undocumented immigrants vs. U.S.-born residents?
Executive summary
Peer‑reviewed studies comparing crime rates of undocumented immigrants and U.S.‑born residents typically combine administrative crime records (arrests, incarcerations, convictions) with separate population estimates for undocumented people to build numerators and denominators, then run robustness checks and alternative specifications to test sensitivity; the best‑known examples use Texas arrest data paired with Center for Migration Studies (CMS) residual population estimates [1] [2] [3]. These studies repeatedly report lower rates for immigrants—including undocumented people—than for U.S.‑born residents, but they explicitly note measurement limits such as misclassification at arrest, underreporting, and uncertainty in population denominators [4] [5] [6].
1. What data constitute the “numerator”? — administrative arrest and incarceration records
Leading peer‑reviewed work uses official criminal justice records as the numerator: comprehensive arrest records from state law‑enforcement agencies (the Texas Department of Public Safety in the PNAS study), state prison and local jail records, or national incarceration counts, depending on the study, to count arrests, convictions, or incarcerations by immigration-related status where available [1] [2] [4] [7]. Researchers choose arrests, convictions, or incarceration as alternative crime measures and sometimes substitute misdemeanors or felony distinctions to check if findings hold across outcomes [4] [5].
2. What data constitute the “denominator”? — estimating the undocumented population with residual methods
Because no universal registry of undocumented residents exists, peer‑reviewed studies rely on external population estimates to calculate rates per 100,000: many use the Center for Migration Studies (CMS) residual estimates and sometimes replicate results with Pew or Census‑based residual counts, which subtract legally present foreign‑born from total foreign‑born to infer the undocumented share [6] [8] [1]. The CMS approach includes additional controls (country‑of‑origin adjustments, emigration and mortality corrections) to reduce sampling error and is described in the literature as a vetted, standard method for producing unauthorized population estimates [6] [8].
3. How do studies assign immigration status at the time of arrest? — classification and verification steps
Researchers typically classify individuals recorded in arrest databases as U.S.‑born, naturalized, legal foreign‑born, or “undocumented” using cross‑referencing fields in police or detainer records and/or by excluding those with indicators of lawful status; the Texas study used “uniquely comprehensive” DPS records that allowed separation of undocumented from legal immigrants in many arrests, and researchers ran alternative classifications to test sensitivity [1] [2] [4]. Studies acknowledge possible misclassification at arrest and therefore perform sensitivity analyses using alternative definitions and outcome measures [5].
4. Statistical approaches — rates, comparisons, and robustness checks
The core method is straightforward rate comparison: compute arrest/incarceration rates per population group (e.g., per 100,000 undocumented vs U.S.‑born) and compare ratios across crime categories (violent, property, drug, homicide). Peer‑reviewed articles augment raw comparisons with robustness checks—substituting different population estimates (CMS vs Pew), alternative crime measures (convictions, misdemeanors), temporal trends, and sensitivity analyses to show results are not driven by a single assumption [4] [5] [8]. Some national papers use longer historical incarceration series or panel methods to examine trends across places and time [7] [9].
5. Known limitations and steps to mitigate bias
Scholars explicitly note several caveats: population estimates for undocumented people carry sampling uncertainty; undocumented people may underreport victimization and avoid contact with police; and not all arrests lead to convictions, while some crimes go unsolved, especially homicides—factors that can bias numerators and denominators [6] [5]. To mitigate these, peer‑reviewed studies run triangulation checks (different population series), alternative crime measures, and sensitivity analyses; these batteries of tests are reported to bolster confidence that undocumented criminality is not higher than U.S.‑born rates in the examined datasets [5] [4].
6. What the methods do and do not prove
Methodologically careful, peer‑reviewed work demonstrates that, within the limits of administrative records and residual population estimation, undocumented immigrants tend to have lower arrest and incarceration rates than U.S.‑born residents across multiple crimes and locales tested—findings reproduced in Texas‑level research and broader national and historical studies [1] [4] [7]. These methods cannot eliminate all uncertainty: they cannot directly observe every undocumented resident or every criminal act, and researchers openly report those limitations rather than claiming absolute certainty [6] [5].