COVID-19 and income profile: How communities in the United States responded to mobility restrictions in the pandemic's early stages

Abstract

Mobility interventions in communities play a critical role in containing a pandemic at an early stage. The real-world practice of social distancing can enlighten policymakers and help them implement more efficient and effective control measures. A lack of such research using real-world observations initiates this article. We analyzed the social distancing performance of 66,149 census tracts from 3,142 counties in the United States with a specific focus on income profile. Six daily mobility metrics, including a social distancing index, stay-at-home percentage, miles traveled per person, trip rate, work trip rate, and non-work trip rate, were produced for each census tract using the location data from over 100 million anonymous devices on a monthly basis. Each mobility metric was further tabulated by three perspectives of social distancing performance: "best performance," "effort," and "consistency." We found that for all 18 indicators, high-income communities demonstrated better social distancing performance. Such disparities between communities of different income levels are presented in detail in this article. The comparisons across scenarios also raise other concerns for low-income communities, such as employment status, working conditions, and accessibility to basic needs. This article lays out a series of facts extracted from real-world data and offers compelling perspectives for future discussions.

Keywords: COVID‐19 pandemic; Mobile location data; causal impact analysis of income; community differences; social distancing behavior.

FIGURE 1

SDI changes across different periods

FIGURE 2

Staying‐home percentage by census tracts on March 13, 2020, and April 13, 2020

FIGURE 3

Temporal patterns of six mobility metrics from March 2–April 17, 2020, by income groups

FIGURE 4

The disparities in course progression over time of three mobility metrics

FIGURE 5

Distribution of the social distancing performance by income groups

FIGURE 6

Distribution of propensity scores by groups

FIGURE 7

Comparison of absolute standardized mean difference