How do definitions of “sanctuary” (detainer refusal vs. broader policies) change study outcomes on crime?
Executive summary
How “sanctuary” is defined drives whether researchers detect a crime effect: studies that use a tight, operational definition—counties that refuse ICE detainer requests—tend to find no increase in crime and often report null effects or modest declines [1] [2], whereas work that uses broader or more varied definitions of sanctuary (limits on inquiries, notification policies, ID programs, informal non‑cooperation) produces mixed results and more divergent theoretical claims [3] [4].
1. Why definition choice matters: a measurement problem dressed as policy debate
Research hinges on what researchers call “sanctuary”: a narrow detainer‑refusal measure captures an explicit change in local–federal jail cooperation (refusing ICE detainers), which is straightforward to code and test against monthly crime series [1] [2], while broader definitions sweep in diverse ordinances—training, limits on questioning, identity programs or mere non‑binding statements—that vary in legal force and local practice and therefore introduce heterogeneity that can obscure or produce divergent results [5] [3].
2. What the narrow-detainer literature finds: consistent nulls or small declines
When scholars define sanctuary as refusal to honor ICE detainers and compare counties before and after policy adoption using granular time series, the headline finding is that deportations fall but crime does not rise; some papers report no detectable effect on crime and others even find post‑policy declines in violent crime [1] [2] [6], with robustness checks that exploit administrative records on detainer compliance to isolate the causal channel [7].
3. What broader or looser definitions reveal: mixed evidence and competing mechanisms
Studies that code sanctuary more broadly or include informal practices face conflicting mechanisms: advocates argue sanctuary builds “trust” with immigrant communities and increases reporting and social integration, lowering crime [8] [4], while critics argue sanctuary could reduce the expected cost of offending or create enforcement gaps that attract criminal behavior—an empirical claim that some theoretical papers and select analyses still advance [3] [9].
4. Units of analysis and time resolution change the story
County‑level versus city‑level analyses and the temporal resolution of crime data matter: research using monthly county crime counts and a detainer definition can trace immediate pre/post changes and show no crime surge [1], whereas cross‑sectional city comparisons or studies mixing multiple policy types can conflate long‑run structural differences—like migration patterns or income—that drive crime differences independent of sanctuary rules [10] [11].
5. Confounders, selection and the appearance of effects
Jurisdictions that adopt sanctuary measures are not random: they differ on demographics, economic conditions, policing philosophy and baseline immigrant population, and those differences—median income or rates of foreign‑born residents—are plausibly correlated with crime trends and policy adoption, so broader definitions amplify selection bias unless carefully controlled [10] [8].
6. Politics, incentives and hidden agendas in measurement choices
Definition choices carry political stakes: federal enforcement actors and the DOJ use broad statutory criteria to label “sanctuary jurisdictions” in ways that facilitate enforcement or funding decisions [12], while advocacy groups emphasize detainer refusal as the clearest signal of non‑cooperation to defend public safety arguments; those institutional motives shape which measures get highlighted and which datasets get compiled [13] [14].
7. Bottom line for interpreting studies
The most defensible empirical claim across the literature is conditional: when “sanctuary” is narrowly defined as refusal to honor ICE detainers and measured with detailed temporal data at the county level, there is little evidence of increased crime and some evidence of stable or reduced crime [1] [2] [6]; when researchers or policymakers expand the definition to include varied practices, results become heterogeneous and sensitive to selection, measurement, and theoretical framing [4] [3].