*2.1. Tweet Sampling*

Twitter data were collected using the keyword "mask" from 27 June 2020 21:21:21 to 4 July 2020 19:50:40 to provide coverage of about a week of data in a period when this topic was highly present in social media. The total number of tweets which were gathered (worldwide) were *n* = 452,430. By using the keyword "mask", tweets including words such as "masks" or "#mask" were also retrieved and included. A systematic random sample of 1% (*n* = 4525) of the tweets was extracted and analyzed using social network analysis, as described in the next section.

#### *2.2. Social Network Analysis*

The software NodeXL (Social Media Research Foundation, California, CA, USA) was used to conduct a social network analysis of the data [16]. In understanding the network graph, the results of this study build upon previous research [17,18], which has highlighted that Twitter topics may follow six network shapes and structures [19]: broadcast networks, polarized crowds, brand clusters, tight crowds, community clusters, and support networks. In the network graph provided in Figure 1, circles represent individual Twitter users and the lines between them represent connections such as mention and reply. The network graph was laid out using the Harel–Koren Fast Multiscale layout algorithm which is built into NodeXL. NodeXL uses the Search Application Processing Interface (API). Influential users were identified using NodeXL and were anonymized by providing a description of the account in line with previous research [18]. A specific subanalysis was performed for tweets that originated solely from the USA. Regional information was extracted as follows: a total of *n* = 13,265 tweets were extracted where users had included "USA" in their user bios and a 5% sample of tweets was extracted and analyzed in NodeXL. Individual users and/or organisations that were deemed to be not sufficiently in the "public domain" were anonymized.

**Figure 1.** Network graph of masks from 27 June to 4 July.
