*3.5. Content Analysis*

NodeXL was used to identify the most frequently occurring co-words, as shown in Table 4. These co-related keywords provided insight into the types of discussions that may have been taking place on Twitter. Word pairs containing mentions of Twitter user handles were removed. This occurred on three occasions in group 3.


**Table 4.** Identifying content across clusters.

\* asterisk placed by authors in expletives.

The most frequent co-related keywords were "wear" and "mask" to form the sentence "wear a mask" and, in some cases, also involved an expletive. There were also tweets related to humor. There were also co-occurring words such as "stay" and "home". Other co-occurring words which appeared across the clusters included "breathing" and "problem", which revolved around the debate of whether face masks should be mandatory as they may lead to breathing problems. In regard to this debate, Twitter users provided evidence to highlight that face masks did not cause breathing problems, whereas other users highlighted the potential for masks to cause breathing issues.

#### *3.6. Regional Analysis of the USA*

In this part of the analysis, tweets were extracted to only focus on the USA. Data were filtered by users who noted in their bios that they were from the USA.
