Advances in Modeling Unauthorized Status in a National Sample Using a Bayesian Approach

• In Event: NIJ-Day: Exploring the Relationship between Immigration Status and Crime: New Approaches with Large Datasets

Abstract

Scholarly consensus on the relationship between immigration and crime suggests a negative or weak association. However, studies often omit potentially relevant measures such as documentation status, examine the relationship at broad geographic levels, examine crime without assessing the role of crime reporting, or fail to examine differences between traditional and emerging immigrant destinations. Furthermore, current methods to estimate the unauthorized population are limited in their ability to produce estimates at fine geographic levels. This study uses data from the 2008, 2014, and 2018 waves of the Survey of Income and Program Participation (SIPP) along with several secondary datasets to produce geography-specific predictive models of unauthorized immigrant status. Employing a Bayesian approach to modeling in the SIPP allows for the use of informative priors obtained from other data sources. These models are used to produce census tract-level estimates of the unauthorized population a number of cities. This analytical process compares differences in estimates gained from a national model to those built both with local information and informative priors obtained from other datasets. This study builds on previous techniques aimed at estimating the unauthorized population. These updated jurisdiction-specific estimates will be used to assess the relationship between unauthorized immigration, crime, and crime reporting at the census tract level.

Authors

• Christopher Inkpen, RTI International

• Jonathon Holt, RTI International

• Kimberly Janda, RTI International

• Kasey Jones, RTI International