Effects in Disguise: The Importance of Controlling for Constructs at Multiple Levels in Macro-Level Research

• In Event: Innovations in Quantitative Modeling

Abstract

Unlike individual-level studies, aggregate-level studies encompass research using various units of analysis (i.e. block-groups, census tracts, cities, MSAs, etc.). However, aggregate-level research typically includes constructs at only one level of analysis without discussing the reason for choosing that particular unit of analysis or explaining why the relationship should only matter at that level. This is an important limitation because relationships between constructs can exist at multiple levels, with each level representing a distinct effect of the overall relationship. Using immigration and crime literature as a guide, the purpose of this presentation is to examine whether the relationship between immigrant concentration and crime rates varies when analyzed at the tract and city-level simultaneously using multilevel modeling. The project utilizes crime-incident and demographic data from 8,198 census tracts nested within 61 U.S. cities to separate the effect of immigrant concentration on crime at the tract versus the city level. Further, a cross-level interaction between tract and city immigrant concentration is examined to assess whether the relationship between immigrant concentration and crime rates at the tract level is dependent on the degree of immigrant concentration at the city level. The presentation concludes with a discussion of the results and ideas for future research.

Authors

• Javier Ramos, Florida State University

• Marin Wenger, Florida State University