How have researchers estimated immigrant involvement in homicide using state-level data like Texas?
Executive summary
Researchers studying immigrant involvement in homicide in Texas rely on a blend of state criminal records, federal immigration status responses, and independent population estimates to produce per‑capita arrest and conviction rates, and they test the robustness of those rates with sensitivity analyses; most recent peer‑reviewed and government‑funded work finds immigrants—documented and undocumented—have lower homicide rates than native‑born Texans [1] [2]. Disputes center not on the basic technique but on how to count the undocumented population (the denominator) and which subset of crimes or identifications to include, and those methodological choices materially change headline rates [3] [4].
1. The data backbone: Texas DPS criminal records and DHS status replies
Scholars begin with the Texas Department of Public Safety’s arrest and conviction records, which list charges—including more than 1,122 homicide charges among identified illegal noncitizens in DPS releases spanning 2011–2026—and then link those records to Department of Homeland Security (DHS) responses to determine whether an arrestee is lawfully present, a legal immigrant, or an undocumented immigrant [5]. Those linked datasets are unique and powerful because they allow researchers to count arrests and convictions by immigration status rather than relying on imperfect proxies [1] [2].
2. Constructing the denominator: population estimates and model choices
To convert counts into rates, researchers must estimate the number of people in each immigration-status group living in Texas—the denominators—and here methods diverge: some use DHS or Center for Migration Studies (CMS) state‑level estimates, others model undocumented population size directly; for example one recent Cato analysis estimated about 2,018,723 undocumented immigrants in Texas in 2022 and noted its figure was within 2% of DHS’s estimate, a small enough difference to leave per‑capita rates largely unchanged [6] [3]. Critics argue that different population series (CMS, DHS, Pew) or incomplete DHS identifications at arrest versus later in prison can materially raise undocumented homicide rates, showing how sensitive results are to denominator choices [4].
3. Choice of numerator: arrests versus convictions and timing of identification
Another pivotal decision is whether to use arrests, convictions, or initial DHS identifications at booking; conviction‑based measures reduce measurement error from wrongful arrests but are fewer in number, while arrest counts are larger but can be biased by enforcement patterns [3] [7]. The Center for Immigration Studies has highlighted that counting only undocumented persons initially flagged at arrest yields different homicide rates than adding those later identified during incarceration, which can increase the undocumented homicide rate in some years [4].
4. Robustness checks, comparative baselines, and peer‑reviewed findings
High‑quality studies—NIJ‑funded analyses and peer‑reviewed work in PNAS—combine Texas DPS data with DHS-derived population series and run alternative specifications to demonstrate consistency: they report that undocumented immigrants had lower felony arrest rates for violent crimes and homicide than native‑born citizens across multiple years and when using alternate population estimates [1] [7] [2]. Cato’s decade‑long conviction‑based brief similarly reports lower homicide conviction rates for illegal immigrants (2.2 per 100,000) and legal immigrants (1.2 per 100,000) versus native‑born Americans (3.0 per 100,000) across 2013–2022, while acknowledging data quality caveats [3] [8].
5. Debates, hidden agendas, and what remains uncertain
Methodological debates reflect different policy and institutional incentives: advocacy and think‑tank reports can emphasize either higher or lower immigrant crime rates depending on choices about timing of identification, inclusion of prison‑identified noncitizens, and which population series to trust, as illustrated by CIS’s critique that certain counting methods understate undocumented criminality versus Cato and academic studies that present lower rates after broader identification and modeling choices [4] [3] [1]. Where reporting is thin, researchers honestly note limits—DPS counts omit noncitizens who were never identified by DHS, and population estimates are model‑dependent—so absolute precision is unattainable even as comparative patterns appear robust [5] [6].