Data was originally extracted from Twitter's Historical Power Track API, via Crimson Hexagon's Forsight platform (Crimson Hexagon, 2019). 

We queried first for keywords “Brexiteer” and “Remainer,” then for “Brextremist” and “Remoaner,” producing two separate datasets in raw javascript object notation (JSON) format, including all tweet object fields. 

The tweet object encompasses fundamental variables obtained for each tweet such as unique id, date, and text, along with information about retweets, favorites, and hashtags. 

The raw JSON files were very large (over 80 GB), so we used a Python script to extract only the four key variables required for analysis: unique tweet id, date and time of tweet, text of tweet, and keyword associated with tweet. 

The original number of users in the data set was n = 844,881. The resulting tweets were stored in a csv file and included a total of n = 9,027,822 tweets posted between June 1, 2016 and February 13, 2019.

For the purposes of this data upload, we retained only date and time of tweet and text of tweet, which is necessary to keep users anonymous.