Social Media Election Oracle: Results
Finally, the chaos and uncertainty of the general election campaign is over and we are into a period of chaotic and uncertain horse-trading between the political parties. Fear not though as this too will soon be over and we will be back to the everyday chaos and uncertainty of stable government.
But who could have predicted that this was going to happen? Well, just about everybody actually, but then they still managed to be surprised and concerned when it came to pass. Last week, I was one of those trying to forecast the make up of the new parliament, using four different ways of quantifying party support – so how did I do? In case you have been living in a cave, the actual election result was a hung parliament with the Conservatives as the largest party, winning 306 seats.
Model 1: Facebook – Facebook’s Democracy UK poll had 43% backing the LibDems, 24% the Conservatives and 23% Labour. If this is an accurate reflection of the general population this would have given the LibDems an outright majority of 182 seats.
Model 2: Homepage Traffic – In the week leading up to May 1st, the Conservatives had 36% of traffic, Labour 33% and the LibDems had 31%. This model gave a hung parliament, with Labour as the largest party, winning 296 seats.
Model 3: Social Media Chatter – I used Cision Social Media to measure the level of chatter around each party in the week leading up to the election, this predicted a hung parliament, with the Conservatives as the largest party, winning 317 seats.
Model 4: Social Media Sentiment – Finally, I looked at the same social media chatter, but only considered the positive statements. This model also gave a hung parliament, with Labour as the largest party, winning 287 seats.
So, three out of four of these methods successfully predicted a hung parliament. Model 3 had the Tories as the largest party and got to within eleven seats of the actual result. I’m pretty impressed with this – but can it really be that simple – chatter = votes? More research is clearly needed, but we will have to wait for another election somewhere. Who knows, if Lord Ashcroft finds some loose change down the back of his sofa, we might have another one here sooner rather than later.
Leave a Comment
Trackbacks & Pingbacks
[…] to use quantitative data from social media to predict the result of the election was problematic (I know, I tried). One of the problems in using this sort of analysis for predicting UK elections are the […]
[…] our regular political commentator away, I was inspired by Hitwise’s look at the search patterns associated with the Labour […]
[…] At Cision, we’re always trying to highlight how social media can be used to provide snapshots of public sentiment on the issues of the day – and some of our efforts using Cision Social Media have surprised even us with their accuracy and predictive power. […]
[…] This post was mentioned on Twitter by Cision UK. Cision UK said: How accurate were our social media #ukelection predictions? Surprisingly so in some cases – check them out on our blog http://bit.ly/9f8Iaw […]