In November of last year, my colleague Jaime Brugueras ventured into some treacherous territory – predicting Presidential elections. Using Networked Insights’ analytics engine Kairos to examine millions of online and Twitter-based conversations about Donald Trump and Hillary Clinton through the lens of emotions and intent to vote, he was able to make a prediction was far more accurate than the vast majority of major polling sources.
Three days before the election, our data showed Trump winning. Our ultimate map, which was shared throughout the day by our election coverage partners at Wired Magazine, predicted a potential electoral tie of 269 to 269, which was a very wide departure from the trouncing by Clinton that other pollsters predicted. Huffington Post, for example, gave her a whopping 98% chance of winning.
So how were our results so different than others? At Networked Insights, we were analyzing social data to get a more accurate, real-time picture of voter sentiment than survey-based data alone could ever gather.
From Trump to the U.K.
As a global company expanding into the European market, we thought we would once again test our hypothesis – that social conversations, particularly those expressing strong negativity and intent to vote, could prove more powerfully predictive than traditional polling – on the upcoming U.K. general election.
For this analysis, Networked Insights reviewed close to 5M publicly available conversations for the last month around the U.K. elections, using our analytics platform Kairos to better understand emotions, attitudes and voter intent around the candidates. The top 2 candidates alone are generating more than 100,000 conversations a day in the UK and increasing, so there is plenty of data to draw from.
What we found was that the U.K. elections could be a far closer race than many believe.
The two key metrics that Networked Insights used to predict the 2016 US elections were intent to vote and negativity surrounding each candidate. We used the same metrics to measure all UK candidates. Below is a comparison for the two leading parties between the 2015 and the 2017 election.
Two main takeaways emerge from the data: 1) Tories have an alarming rate of negativity surrounding Theresa May and 2) Jeremy Corbyn has a higher percentage of intent to vote than his predecessor Ed Miliband.
This leads us to the conclusion that the results of the 2017 elections will be closer than the 2015 election, leading to a hung parliament.
What’s driving the conversations? Voters are most concerned about Brexit, healthcare, and national security.
While May is seen as a more stable candidate and stronger with the economy, security failures are hurting her image when fighting terrorism. Voters like Corbyn’s plan on terrorism and his foreign policy, but he faces trust issues and voters aren’t sure where he stands on immigration.
Updated Data (June 6, 2017):
Since Networked Insights did its original analysis of the 2017 U.K. General Election (see below), predicting a Hung Parliament due to increased negativity towards Theresa May since the 2015 elections and higher intent to vote for Corbyn, two major terrorist attacks have rocked London.
We took a look at social conversations to find out if these new events changed voter perception towards the candidates.
The short answer is yes, but we’re still calling a Hung Parliament. Why?
While we expected to find increased negativity to be directed at May as a result of the attacks as the incumbent, the events resulted in increased negativity towards both candidates. Since the attacks, May’s negativity went from 10.4 to 10.5 while Corbyn’s went from 7.5 to 8.1.
Not surprisingly, voters’ conversations around defense have skyrocketed, up 10 points from 12% in our earlier analysis. More updates after the election June 8th.
Want to dig deeper into the data? Click here to receive the full 2017 U.K. Elections Report.
Want to learn more about how we do it? Contact us today.