Social Media Opinions on Working From Home in the United States During the COVID-19 Pandemic: Observational Study

Abstract

Background Since March 2020, companies nationwide have started work from home (WFH) owing to the rapid increase of confirmed COVID-19 cases in an attempt to help prevent the disease from spreading and to rescue the economy from the pandemic. Many organizations have conducted surveys to understand people’s opinions toward WFH. However, the findings are limited owing to a small sample size and the dynamic topics over time. Objective This study aims to understand public opinions regarding WFH in the United States during the COVID-19 pandemic. Methods We conducted a large-scale social media study using Twitter data to portray different groups of individuals who have positive or negative opinions on WFH. We performed an ordinary least squares regression analysis to investigate the relationship between the sentiment about WFH and user characteristics including gender, age, ethnicity, median household income, and population density. To better understand the public opinion, we used latent Dirichlet allocation to extract topics and investigate how tweet contents are related to people’s attitude. Results On performing ordinary least squares regression analysis using a large-scale data set of publicly available Twitter posts (n=28,579) regarding WFH during April 10-22, 2020, we found that the sentiment on WFH varies across user characteristics. In particular, women tend to be more positive about WFH (P<.001). People in their 40s are more positive toward WFH than those in other age groups (P<.001). People from high-income areas are more likely to have positive opinions about WFH (P<.001). These nuanced differences are supported by a more fine-grained topic analysis. At a higher level, we found that the most negative sentiment about WFH roughly corresponds to the discussion on government policy. However, people express a more positive sentiment when discussing topics on “remote work or study” and “encouragement.” Furthermore, topic distributions vary across different user groups. Women pay more attention to family activities than men (P<.05). Older people talk more about work and express a more positive sentiment regarding WFH. Conclusions This paper presents a large-scale social media–based study to understand the public opinion on WFH in the United States during the COVID-19 pandemic. We hope that this study can contribute to policymaking both at the national and institution or company levels to improve the overall population’s experience with WFH.

Caregiver Faculty: Working through a Pandemic

“…While not within the scope of our project, in reviewing the literature, we found a few additional areas of interest. One area that we did not capture in our research: Why do older workers with higher incomes prefer WFH initiatives ( Xiong et al 2021 )? Is this because they have already climbed the career ladder and are comfortable in their positions, or is there another less obvious reason that contributes to this situation?…”

“…They concluded that the women in their sample reported reduced work hours, which may explain why they reported more positive opinions about WFH as it allowed them to spend more time with family, especially children. Interestingly, they also found that older workers with higher incomes tend to hold more positive opinions on WFH ( Xiong et al 2021 ). While analyzing worker productivity of WFH during the pandemic, other researchers found similar findings.…”

Human behavior in the time of COVID-19: Learning from big data

“…Social platforms have abundant multimodal data. The publicly available profile image uploaded by a user may contain the demographic information of this user (Lyu et al, 2020 ; Xiong et al, 2021 ; Zhang X. et al, 2021 ). This side information can potentially improve the performance of measuring human behavior.…”

Human Behavior in the Time of COVID-19: Learning from Big Data

“…Social platforms have abundant multimodal data. The publicly available profile image uploaded by a user may contain the demographic information of this user (Xiong et al 2021;Zhang, Lyu, and Luo 2021;Lyu et al 2020). This side information can potentially improve the performance of measuring human behavior.…”