A Computational Analysis of the Coronavirus Pandemic Response of Tri-State Area Politicians on Twitter

EasyChair Preprint 5984

Abstract

The 2020 novel coronavirus (COVID-19) pandemic has undoubtedly altered the normal way of life for people throughout the world. In the United States, nearly 10 percent of all cases reside in the states New York, New Jersey, and Connecticut, commonly referred to in the region as the “Tri-state area.” We present our findings on a computational analysis of coronavirus response-related tweets by prominent politicians that represent the Tri-state area: Governors Andrew Cuomo, Phil Murphy, and Ned Lamont, Mayor Bill de Blasio, and President Donald Trump. We extract tweets over the span of 20 months, 14 months prior to the first confirmed case in the Tri-state area as a baseline, and six months during the pandemic. With this data, we observe a drastic increase in Twitter activity and apply latent dirichlet allocation (LDA) and latent semantic analysis (LSA) models to discern topic changes of these politicians during the pandemic period. Moreover, we analyze sentiment and lexical changes of these politicians resulting from the pandemic. Ultimately, we illustrate the increase in Twitter usage by these politicians to engage with their constituents during the pandemic and create a narrative of COVID-19 mobility and awareness, resulting in a comparatively more neutral and even tweet sentiment.

Keyphrases: COVID-19, LDA, LIWC, LSA, NLP, Sentiment Analysis, Twitter, politics, tri state area

@booklet{EasyChair:5984,
  author    = {Zachary Dau},
  title     = {A Computational Analysis of the Coronavirus Pandemic Response of Tri-State Area Politicians on Twitter},
  howpublished = {EasyChair Preprint 5984},
  year      = {EasyChair, 2021}}