Aiimi Analyses: Question Time
This week Aiimi are predicting the UK General Election. In any 21st century campaign, social media plays an important role, as such, we are monitoring Twitter for the hashtag #GE2017. Whilst Twitter as a data source is biased towards a younger audience, it proves a useful tool to gauge reactions to events, such as the BBC Question Time Leader’s Specials. In this blog, I will be demonstrating the power of monitoring social media feeds, to gain a better understanding of impact, using sentiment and geospatial analysis.
Two Question Time specials were filmed, Theresa May and Jeremy Corbyn on Friday, and Tim Farron and Nicola Sturgeon on Monday. Looking at the first special, just on candidate and party mentions, we can see the impact of the show, running from 20:30 to 22:00, with May on first and Corbyn second.
Representing the two main parties, it is difficult to identify any lasting impact of the special in terms of mentions as both Labour and the Conservatives typically take up 40 – 50% of the twitter conversation at any given time anyway. However, if we consider the sentiment analysis of tweets around the event (on a -1 to +1 scale), comparing 4 hours before and after, we can see a substantial rise in sentiment for Jeremy Corbyn.
Whereas, Theresa May, although fluctuating during the show, maintains a relatively constant sentiment. Although we know the younger demographic on Twitter leans more towards Jeremy Corbyn and such a positive reaction could be suspected, the lack of impact from Theresa May suggests she simply offered more of the same, whereas Jeremy Corbyn impressed.
Next, we look at the second special, between Tim Farron’s Liberal Democrats and Nicola Sturgeon’s Scottish National Party (SNP). With both parties typically not discussed as much as Labour and The Conservatives, there is something to play for in terms of the impact on mentions.
Here we see normalised mentions for party and leader names, the SNP line is calculated just from users based in Scotland, whereas the Liberal Democrat line is nation-wide, to fairly represent the impact on people who can actually vote for each party. Interestingly, the show (which ran from 21:00 to 22:00, with Farron on first and Sturgeon second), gave Sturgeon a clear impact boost, whereas Faron’s prominence spiked and quickly dropped back down to its earlier level. But, is all publicity good publicity?
In this case – yes. Although nothing to shout home about like Jeremy Corbyn, both Farron and Surgeon see slight rises in sentiment polarity. It’s difficult to comment on whether in-fact these are just as or more significant than Corbyn’s rise general due to the uncertain Twitter demographic.
Much like his mentions, Farron’s sentiment spikes, before dropping down to a less exciting level, although interestingly, this occurs before the show has actually started, suggesting there was some anticipation for his appearance.
It’s also worthy of note that Sturgeon actually received a lot of criticism on the show over the state of the Scottish government (of which she is First Minister). Interestingly, the sentiment line doesn’t seem to reflect that, perhaps Twitter contains a significant number of pro-SNP users, swinging the balance?
Whilst the only unbiased datasource will be the ballot counts on Thursday night, that doesn’t mean that we can’t help justify apparent biases on Twitter by comparing to other feeds. Whilst the media is far from unbiased, we take a newsfeed (using NewsAPI) from 19 different newspapers and websites, averaging over these arguably gives a more representative view of the electorate’s mood than twitter. I have prepared two graphs, one comparing the sentiment of tweets solely about one party and the other, headlines solely about one party.
This view seems to confirm our suspicions about the SNP being misleadingly popular on twitter, although it is worth remembering that a lot of the news articles will be aimed at an English audience, who typically have a negative view of the SNP. Interestingly, the Liberal Democrat support seems to be higher on Twitter too, but the clear disparity between Labour and The Conservatives remains in both sources. Given Labour’s recent surge in the opinion polls, perhaps this is not so surprising.
Assuming this comparison suggests some doubts in Tim Farron’s and Nicola Sturgeon’s popularity, but largely justifies the success of Jeremy Corbyn over Theresa May. I have compared the impact of both mentions and sentiment on Twitter and ranked the four candidates in order of their performances on Question Time.
The Twitter sentiment suggests Jeremy Corbyn secured a victory over his opponent, Theresa May on Monday night. Whereas, both Nicola Sturgeon and Tim Farron made gains, it appears that this was heavily influenced by their existing supporters posting, rather than winning new support, Sturgeon just edging above Farron due to holding onto the limelight for longer. Theresa May appears to have performed the poorest on Question Time, failing to make an impact on either her social media prominence or sentiment. It is fair to comment, perhaps Theresa’s target demographic does not make significant use of social media, but regardless, there is nothing to suggest her performance was particularly strong in the Twitter feed.
Do you agree with my conclusions? Please get in touch with either Aiimi or myself on Twitter and let us know what you think!
Whoever won on Question Time this week, there’s no guarantee that will translate to real votes on Thursday. For us, the challenge is only just beginning, follow #AiimiAnalyses to keep up to date with our exciting journey!
As published on the Aiimi website