No abstract

No abstract

No abstract

One of the many contributions of Land, McCall, and Cohen’s landmark study was the confirmation of a long-held view in criminology—that deprivation raises homicide. Yet recent literature finds that although Latino immigrant communities are often poor, paradoxically they have low levels of crime. Unfortunately, this seemingly contradictory evidence is based on studies of long-established, well-organized, traditional immigrant communities where Spanish is a modal form of communication. However, recent Latino migrants opted for new destinations that are unprotected by a shell of common culture and language, making Latinos in these areas more vulnerable to serious violence. In acknowledging these critical differences between old and new Latino communities, we observe four interrelated findings: (a) The widely held view that Latinos generally live in safe places is true only for those in traditional destinations; (b) Latinos in new destinations are murdered at an exceedingly high rate; (c) This elevated risk is linked to English nonfluency among Latinos in new destinations only; and (d) In these areas, linguistic isolation increases homicide not just directly but indirectly as well by first increasing Latino economic deprivation. Thus, once differences in place are considered, there is no “paradox” about Latino immigration and crime. Our results uphold the benchmark assessment of Land, McCall and Cohen, that deprivation is linked to homicide—even in Latino communities.

We give a unified treatment of statistical methods for assessing collapsibility in regression problems, including some possible extensions to the class of generalized linear models. Terminology is borrowed from the contingency table area where various methods for assessing collapsibility have been proposed. Our procedures, however, can be motivated by considering extensions, and alternative derivations, of common procedures for omitted-variable bias in linear regression. Exact tests and interval estimates with optimal properties are available for linear regression with normal errors, and asymptotic procedures follow for models with estimated weights. The methods given here can be used to compareβ1 and β2 in the common setting where the response function is first modeled asXβ1(reduced model) and then asXβ2+Zγ(full model), withZ a vector of covariates omitted from the reduced model. These procedures can be used in experimental settings (X= randomly asigned treatments,Z= covariates) or in nonexperimental settings where two models viewed as alternative behavioral or structural explanations are compared (one model withX only, another model withX andZ).