TickerTags First to Call Brexit Results Using Twitter Data

TickerTags uses Twitter data to predict Brexit results hours before major news sources

TickerTags
2 min readJun 24, 2016

--

Early in the evening the day of the Brexit vote, Twitter exit polling showed 52/48 in favor of Remain, but the tide turned in favor of Leave in the last hours of voting.

At 3:27 PM CDT (21:27 GMT+1), just minutes before the polls closed, TickerTags using real time organic (non-bot, non-retweet) Twitter mentions of EU Referendum related word groupings to indicate a 52/48 victory for Leave.

Timeline:

3:27 PM CDT — 21:27 GMT+1 TickerTags indicates Leave at 52/48

4:00 PM CDT — 22:00 GMT+1 Polls close

4:00 PM CDT — 22:00 GMT+1 YouGov exit poll predicts Remain

10:00 PM CDT — 04:00 GMT+1 Nigel Farage declares British ‘Independence Day’ in victory speech to supporters

10:58 PM CDT — 04:58 GMT+1 CNN calls Leave

11:01 PM CDT — 05:01 GMT+1 CNBC calls Leave

11:11 PM CDT — 05:11 GMT+1 BBC Breaking News Email reports Leave

12:07 PM CDT — 06:07 GMT+1 Bloomberg reports Leave

12:22 PM CDT — 06:22 GMT+1 Reuters reports Leave

TickerTags aggregated organic mention volume of the following tags. Mentions purporting to have voted for Leave comprised 52% of the total mention count compared to 48% for mentions purporting to have voted for Remain.

Aggregated Tags*:

“voted” +leave

“ivoted” +leave

“ivotedleave”

“voted” +remain

“ivoted” +remain

“votedremain”

*Aggregated tags were chosen based on how well they represented the total volume of mentions for each side. While some tags were able to be formatted consistently, others were slightly different to represent the difference in conversation between the two groups. For example, #iVoted was trending on Twitter for Leave and Remain voters, but Leave voters used #IVotedLeave while Remain voters used #VotedLeave.

Prior to the official results being announced, most published polls predicted Remain would win by a slim margin. The YouGov exit poll predicted Remain would win with 52% of the vote.

--

--