Effects in Disguise: The Importance of Controlling for Constructs at Multiple Levels in Macro–Level Immigration and Crime Research

Contemporary research suggests that immigrant communities often have lower rates of crime despite their disadvantaged status. Yet prior work often examines the immigration and crime association using only one level of analysis without regard for how this relationship might vary when analyzed across multiple levels of analysis simultaneously. Research also suggests that the immigration–crime link varies across spatial contexts. Using hierarchical Poisson Regression among a sample of 6,660 tracts nested within 55 cities, we examine whether the relationship between immigration and crime varies when examined at the tract and city levels simultaneously. We also include a cross–level interaction in our model to test whether the tract–level association between immigration and crime varies by the size of the foreign–born population at the city level. Results show that the immigration–crime link depends on the level of analysis, such that the relationship is positive at the tract level but negative at the city level. However, we find no support for our cross–level interaction.