Summary
The goal was to analyze tweets from January to April 2020 that mentioned the World Health Organization and COVID, with the aim of detecting main topics, quantify the toxicity of the conversation, and detect triggering events for most popular conversations.
Data
Data for this study is obtained from a bigger dataset including 327.5 million tweets and collected calling the Twitter Streaming API from 20 January to 23 April, from a collaborator (Dr. Manlio de Domenico, Bruno Kessler Foundation, Trento, Italy). The resulting sample of filtering WHO mentions includes 222,774 tweets (70,966 unique tweets).
Results
Around 21% of the overall conversation touching on the Covid-19 pandemic and the role of WHO on the crisis are toxic tweets. The volume of tweets and percentage of toxic tweets increase after March 26th.
There are three main types of events that fuel the conversation: WHO press conferences, just around updates of the pandemics evolution; media news around political biases of WHO; and, in particular, elite criticism versus WHO, towards the end of the period. The latter triggered an increase in toxicity of the tweets.
Lastly, some toxic conversations are attached to certain hashtags that help to identify ‘campaigns’. Some of these campaigns show significantly higher percentage of users that are not active any more (suspended or missing) compared to the overall conversation.
This resulted in a Reuters Institute paper that can be found here.
