How can researchers accurately measure homicides by perpetrator immigration status in the U.S.?
Executive summary
Accurately measuring homicides by perpetrator immigration status in the U.S. requires combining centralized homicide databases with immigration-control records, careful case-by-case linkage, and transparent handling of missing data and biases; no single national dataset currently contains reliable, comprehensive perpetrator immigration status for all homicides [1] [2]. States such as Texas have uniquely detailed arrestee-immigration tracking that researchers can emulate, but reliance on law-enforcement or enforcement-agency data introduces reporting gaps and political incentives that must be declared and mitigated [3] [4].
1. Define the target and accept current limits
Researchers must start by defining whether the question is legal status at time of offending, country of birth, or citizenship because federal and academic sources record these differently and incompletely; many national surveys and administrative systems do not systematically record perpetrator immigration status, which constrains analysis [5] [6]. The National Violent Death Reporting System is the most comprehensive public homicide source for victims and some incident context across participating states, but it does not uniformly capture perpetrator immigration status for all cases, so its use is a partial solution rather than a complete one [2] [1].
2. Use multi-source linkage as the core method
The most accurate approach is deterministic and probabilistic linkage of homicide incident records (NVDRS, local medical examiner and police records) to immigration-control and biometric databases maintained by DHS or state systems, combined with criminal records and incarceration databases; aggregated multi-source datasets avoid reliance on voluntary state reporting alone and improve coverage of perpetrator attributes [1] [3]. Texas provides a real-world model because it links arrestee biometrics to DHS systems to estimate arrests and convictions by immigration status; researchers should study and, where possible, replicate that technical linkage and governance model [3].
3. Standardize variables and document uncertainty
Standardize definitions (foreign-born, lawful permanent resident, unauthorized, citizen) and report uncertainty intervals; when immigration status is not observable, use transparent imputation and sensitivity analyses rather than omitting cases, and always report how many homicides lacked determinable status [5] [1]. Where possible, prioritize homicide as an outcome because it is more reliably recorded than other violent crimes and less subject to underreporting in immigrant communities, a point reinforced by public-health literature using NVDRS data [1] [2].
4. Control for denominator and compositional effects
Any rate comparison must use appropriate denominators: population counts by immigration status (from Census estimates or administrative records) and time- and place-matched population-at-risk measures; raw counts of perpetrators without denominator adjustment are misleading, and prior work shows that immigration is often associated with stable or reduced violent-crime rates at city and state levels [4] [7].
5. Biases, nonreporting, and enforcement artifacts
Measurement is distorted by differential reporting, selective enforcement, and agencies' incentives: enforcement agencies publish “criminal alien” tallies that reflect records checks and arrests, not endemic homicide rates, and those statistics can be shaped by agency priorities [8] [9]. Victim and community underreporting is also higher in immigrant populations for nonlethal crimes, which is why homicide is often used as a more reliable proxy—yet even homicidal incidents can suffer classification and record-linkage failures [1] [2].
6. Legal, privacy, and ethical guardrails
Linking law-enforcement, immigration, and biometric records raises legal and privacy constraints; automated case tools like ACIS provide limited public information and are not substitutes for formal data-sharing agreements, so rigorous governance, minimal disclosure, and institutional review are required before constructing linked datasets [6]. Researchers must balance transparency with protection against misuse of findings by actors seeking to stigmatize communities or to advocate for enforcement-only policies such as those advanced by VOICE and similar offices that emphasize crimes by noncitizens [10].
7. Transparent reporting, replication, and political context
Any study must disclose data sources, linkage methods, missing-data rates, and sensitivity checks because the literature contains competing claims—some scholarship and enforcement reports emphasize reduced criminality among immigrants while other advocacy and older studies point to data gaps and local hotspots—so clarity about methods and limitations is the only way to produce credible, reproducible estimates [4] [5] [11]. Ultimately, accurate measurement is feasible but requires investment in cross-agency data architecture, ethical approvals, and careful communication to prevent politicized misinterpretation.