Prediction of Anxiety Due to COVID-19 Across States from 2020 to 2021

• Activity Number: 75 - SPEED: Data Challenge and SPAAC
• Type: Topic-Contributed
• Date/Time: Monday, August 9, 2021 : 10:00 AM to 11:50 AM
• Sponsor: Section on Statistical Computing
• Author(s): Yifan Lu* and Stephen Frantz and Duwani Katumullage and Hon Keung Tony Ng and Lynne Stokes
• Companies: Southern Methodist University and Southern Methodist University and Southern Methodist University and Southern Methodist University and Southern Methodist University
• Keywords: Survey data; Logistic regression; Visualization; Simulation; Covid-19; Spatial and temporal

Abstract

The COVID-19 pandemic has adversely affected mental health in the US. To consider that effect, this project will identify how anxiety progresses across different geographic areas from 2020 to 2021. The Census Bureau's Household Pulse Survey is used to build a set of logistic regression models to achieve this target. Each model is created for a particular week and provides the probability that a person is anxious based on his/her characteristics, including age, race, income, insurance status, and marital status. Then, these models are applied to the 2019 ACS PUMS dataset to simulate how people's anxiety will change during the pandemic. The expected anxiety ratios in several states at different time points are calculated and mapped onto the Public Use Microdata Areas (PUMA). The preliminary result indicates that cities, compared to rural areas, are more prone to anxiety. With visualization of spatial and temporal change of anxiety, this project provides a good idea of which areas require more attention from the lawmakers. The demographics that have a significant impact on anxiety can be identified as well.