Fairness and Bias in COVID-19 Social Distancing Complaints: Evidence from Large-Scale Mobility Data

• Activity Number: 90 - Novel Statistical Methods for COVID Pandemic and Other Current Health Policy Issues
• Type: Contributed
• Date/Time: Monday, August 9, 2021 : 10:00 AM to 11:50 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Constantine E. Kontokosta* and Boyeong Hong and Bartosz Bonczak
• Companies: New York University and New York University and New York University
• Keywords: COVID-19; Geolocation data; data science; reporting bias; fairness in AI; mobility behavior

Abstract

In the public health context, bias in reporting behavior can lead to suboptimal – and unfair – outcomes in terms of where social distancing rules are enforced, against whom, and how frequently. This paper uses 311 data from New York City, covering over 70,000 social distancing complaints from April to July, and geolocation data derived from more than 800,000 unique smartphone devices to estimate the localized crowding conditions in specific areas proximate – in time and space – to reported complaints. We evaluate how the patterns of social distancing complaint reporting vary with neighborhood characteristics, and estimate the difference in the relative threshold deviation from normal activity that triggers a complaint. By matching neighborhoods based on similar land use and retail establishment characteristics using a k-NN algorithm, we quantify the variation in reporting behavior to identify socio-demographic and socio-cultural influences on the likelihood to complain. Following this, we model the effect of these differences on enforcement and outcomes and discuss the implications for social equity in public health policies predicated on resident reporting.