How do researchers estimate crime rates for undocumented immigrants when national arrest records lack immigration status?
Executive summary
Researchers compensate for the absence of immigration-status fields in national arrest records by combining jurisdictional datasets that do record status, population-estimation techniques for unauthorized residents, and careful robustness checks — an approach that can yield credible estimates but carries important measurement and selection caveats (Texas DPS data is the canonical example) [1] [2] [3].
1. Texas as a laboratory: where arrest records include immigration status
The clearest pathway to estimating undocumented-offending rates has been state-level criminal databases that actually record immigration status at booking — most notably the Texas Department of Public Safety, which checks and logs arrestee status and has allowed researchers to compute arrest and conviction rates by nativity and legal status from 2012–2018 [1] [4] [2]. Studies using those data calculate per‑100,000 arrest and conviction rates for undocumented immigrants, legal immigrants, and U.S.-born people, and repeatedly find lower felony arrest and conviction rates among undocumented immigrants in Texas, a finding replicated across arrest- and conviction-based measures in the study appendices [1] [4] [3].
2. Population denominators: estimating how many undocumented people there are
Deriving a crime rate requires reliable denominators, so researchers pair arrest counts with independent population estimates of the undocumented — often constructed via residual methods and specialized sources such as the Center for Migration Studies (CMS) or Census-based techniques that estimate unauthorized shares by national origin and adjust for demographic structure [5]. These independent population controls are critical because the undocumented fraction of the foreign-born population varies by origin and over time; inadequate denominators would distort per‑capita rates [5].
3. Administrative linkages and DHS “legal/illegal” flags
Where possible, scholars supplement local arrest files with federal records: Department of Homeland Security queries can produce binary indicators (“legal” vs. “illegal”) for arrestees, and combining that with state booking records or conviction data lets analysts separate legal and undocumented noncitizens from citizens — though researchers note the DHS variable is incomplete and must be used with caution [3] [6]. The NIJ summary of Texas work explains this blend of administrative sources as what makes the Texas analysis uniquely informative [2].
4. Robustness checks: convictions, offense types, and time trends
To address biases from policing or booking practices, researchers rerun analyses using convictions (not just arrests), examine specific offense categories (violent, property, drug), and test stability over time; the Texas work reports that core findings — lower rates for undocumented individuals — hold across arrests and convictions and across multiple offense types [4] [3]. Peer reviewers and agencies like the NIJ highlight these sensitivity analyses as central to credible inference [2].
5. Limits, biases, and political incentives in the data
All methods have limits: single-state evidence may not generalize, DHS records can be incomplete, population estimates depend on imperfect residual techniques, and enforcement patterns (local policing priorities, sanctuary policies, or targeted operations) can bias arrest counts upward or downward for certain groups [5] [7]. Federal summaries of “criminal aliens” from CBP reflect convictions discovered at interdiction but do not substitute for population-based rates and can be shaped by enforcement priorities [8]. Several organizations and reviews caution against overgeneralizing Texas-based results to the national level [9] [7].
6. What the body of evidence says and what it does not prove
Combining jurisdictional arrest files that record status, defensible population estimates, administrative linkages, and robustness tests produces the best available estimates — and a growing body of research using those tools finds that immigrants, including unauthorized immigrants in studied contexts, tend to have lower criminality than the U.S.-born [1] [10] [9] [7]. This does not resolve every question: national arrest records still lack a standardized immigration-status field, so broader national inferences rely on extrapolation and careful attention to the measurement problems described above [2] [3].