- Can our finance guy beat the election pollsters?
When Theresa May shocked the UK by calling the snap election, it seemed fairly certain that she’d walk the race leaving a disheveled Jeremy Corbyn in her wake. However, the straight-talking socialist has been building momentum whilst May takes a more reserved approach. The result? Well, if recent YouGov polling is anything to go by, the two diametrically opposed candidates are now closer than ever as they lumber tumultuously towards the finish line… But, is it as close as they are suggesting?
With all the official exit polls, YouGov studies and industry experts giving their opinions on this week’s election, our data team challenged our Finance Manager, Connor, to take on the world of data science to see if he could beat the experts.
Here are his predictions…
Connor’s key findings:
- There was not an inherent bias towards the left, which multiple news outlets and research firms suggest is prevalent on major social media platforms.
- More often than not, individuals Tweeted positively when mentioning a party or party leader.
- At the time this research was being carried out, the results were not incredibly dissimilar from other polls such as YouGov, Comres, ICM and Ashcroft.
Looking at these findings, it appears that the UK’s approach to the political landscape is more positive and balanced than traditional media outlets would have you believe.
And here’s the science bit…
The total UK sample was made up of seven sub-samples of approximately 350,000 Tweets – these samples were randomly pulled from Twitter between 18th May – 30th May during the election period. The data was then filtered and parsed through a sentiment algorithm, where the polarity of each Tweet was used to assess an overall sample sentiment which should indicate voting preference… If the original hypothesis was correct of course.
This method was applied to all parties that currently hold seats. However, exact predictions were made only for Conservatives, Labour, Liberal Democrats, Green, SNP, Plaid Cymru with all other parties are listed under ‘Other’. Finally, seats were then allocated proportionally based on the overall sentiment score of each party.
So, a final thought from us to keep you thinking: should social media platforms offer a service to provide balanced news feeds to prevent the incessantly mentioned echo chamber effect? And, more importantly, are social channels like Twitter a reliable alternative to polls such as YouGov? We’ll soon see…
If you have any questions or would like any further information on our research or methodology, please get in touch with Emma on firstname.lastname@example.org.