Latino Immigration, Interaction, and Homicide Victimization

As Latinos spread across the United States, many Americans have begun to fear that their arrival will spark an increase in crime. Unfortunately, early explanations of the immigration-crime link, which found that immigrants disorganized communities, focused on the experience of Eastern European immigrants. This article updates previous literature by focusing on the experience of Latino immigrants. I find that (1) Latino immigration is linked to crime only in new Latino destinations, (2) this link is mitigated by increased interaction among Latinos, and (3) Latino interaction lowers victimization regardless of destination. Notes 1IRCA addressed immigration in three major ways: (1) legalize all illegal immigrants residing the United States, (2) penalize employers for hiring workers without proof of citizenship, and (3) increase federal funding for guarding the border. 2Proposition 187 did the following: (1) stopped undocumented migrants from using social services, (2) required state and local authorities to report suspected illegal aliens, and (3) made the sale or use of false citizenship documents a felony offense (Calavita Citation1996). 3Although this article focuses on Social Disorganization as the theoretical framework for explaining the immigration-crime relationship, there are several other explanations. See Ousey and Kubrin (Citation2009) for a review. 4Shihadeh and Barranco (2010c) do not mention whether this holds for both new and traditional Latino destinations. In general, research on Latino segregation by destination type has been conflicting. For instance, Lichter and colleagues (Citation2010) find that segregation is higher in new immigrant destinations, while Park and Iceland (Citation2011) find the opposite. Therefore, it is difficult to make any predictions regarding segregation differences by destination type. 5The data were obtained directly from the Centers for Disease Control (CDC) and include all geographic areas, regardless of population size. Note. Latino economic deprivation measures (high school, unemployment, single-headed households and poverty) are listed separately for descriptive purposes, but in the multivariate analysis they are combined into a single factor. See Methods section for more detail. *Homicide is expressed as a rate in the descriptive table, but is entered as counts in the negative binomial regression. 6See pages 240–245 in Shryock, Siegel, and Stockwell (Citation1976) for a detailed explanation of calculating age-standardized death rates. 7Where xi, yi, and ti are the number of immigrants, natives, and total population of tract i, respectively, and X is the total number of immigrants in the city. 8I define low skill industries as those where 50% or more of the workers in that sector have no high school diploma. Using the U.S. Bureau of Labor Statistics' Career Guide to Industries (Citation2006), I find the following industries to be low-skilled: [Agriculture, Forestry, and Fishing], [Mining], [Construction], [Manufacturing], [Transportation], [Retail Trade], [Accommodation and Food Services], and [Other Services]. 9Cronbach's alpha for this factor is .55. 10Of the 3,141 counties examined, only 31 fell into both categories. Note. All models adjust for the nesting of counties within states using the cluster option in Stata 11.0. Latino Population is used as an exposure variable. Standard errors are shown in parentheses. Variable coefficients were multiplied by one million to avoid exponential notation. *p ≤ .05; **p ≤ .01; ***p ≤ .001. Note. All models adjust for the nesting of counties within states using the cluster option in Stata 11.0. Latino Population is used as an exposure variable. Standard errors are shown in parentheses. Variable coefficients were multiplied by one million to avoid exponential notation. *p ≤ .05; **p ≤ .01; ***p ≤ .001. 11All models for Tables 2, 3, and 4 were tested for multicollinearity by re-estimating the models in ordinary least squares (OLS) and derived the variance inflation factors. No factors exceeded 3.0. 12This is done by using the formula (eb*s − 1)*100, where b is the coefficient and s is the standard deviation of the variable in question. In this case, Change in Proportion of Latinos Foreign-born. 13Some research suggests that Latinos, and other ethnic minorities, may not benefit from residing in ethnic enclaves (see Sanders and Nee Citation1987; Nee, Sanders, and Sernau Citation1994; Logan, Alba, and Stults Citation2003). However, the enclave-crime literature consistently predicts lower levels of victimization. Note. All models control for total population, moved, vacant houses, housing density, Latino median age, proportion jobs low-skilled, Latino economic deprivation, rural, and change in proportion of Latinos foreign-born. ***p ≤ .001. 14Substantive findings for control variables remained the same. Segregation from whites was not significant in any of the eight models.