Dynamic Population Flows and Crime in Baltimore: Exploration of Seasonal and Intraweek Variations

• In Event: Macrosocial Factors and Crime

Abstract

Purpose: This paper investigates the longitudinal relationship between dynamic population flows and crime rates within Baltimore City during 2019 and whether any seasonal or intra-week variations exist within such a relationship.

Design: This study combines geo-referenced cellphone signal data as the primary source with Part I crime data, ACS data, and land use information from Baltimore. Homicide, robbery, and burglary are the three major types of crime examined, with inward and outward flow and the ratio of inward-outward flows as the primary predictors. I use the longitudinal Poisson regression with neighborhood-level fixed effect and Moran Eigenvector Spatial Filtering as the primary analytic method.

Findings: Both inward and outward flows show substantial connections with most crime types examined, with the ratio of inward-outward flow also matter for crime rate prediction. Besides, seasonal variations exist within the relationship, while intra-week variations exist for the inward population flow.

Summary: Dynamic population flows should be at the center of our efforts to understand crime victimization. Mobility-crime connections vary both seasonally and throughout the week. The above evidence points to dynamic population flows as critical in crime prevention and public safety.

Author

• Xinyi Situ, University of Nevada, Las Vegas