In September Jeremy Corbyn won a surprise election for the leadership of the UK Labour Party. He didn’t just win, he dominated the election winning a higher percentage of votes than any other Labour leader. Could we have seen this coming by analysing social media posts? I have been looking at the data for both the Labour leadership and the current race for the Republican presidential nomination. You may be surprised at the level of insights we can gain from even a cursory analysis of social media posts and engagement.
The UK Labour Leadership Election 2015
I analyzed all of the public Facebook posts that mentioned Jeremy Corbyn and his opponents during the race for the Labour leadership. I used BuzzSumo’s advanced Facebook analysis feature to explore the data. What was surprising was just how many more ‘likes’, ‘shares’ and ‘comments’ there were for posts mentioning Jeremy Corbyn.
The chart below set outs out all the interactions for public posts mentioning Jeremy Corbyn, ‘likes’ are the blue line, ‘shares’ the green line and ‘comments’ the yellow line. As soon as the election started posts mentioning “Jeremy Corbyn” were liked or shared and this rose quickly to over 500,000 “likes” or “shares” a month. By August public Facebook posts that were published mentioning him were shared or liked over 1m times.
By comparison the figures for the two main alternative candidates in the election were significantly lower throughout the campaign. Yvette Cooper’s results are on the left and Andy Burnham’s are on the right. The volume of “likes” for posts mentioning them topped out at around 40,000 a month each. Combined the “likes” and “shares” were less than a tenth of that for Jeremy Corbyn.
Thus just to summarise in August, the last month of the campaign, there were:
- Over 1m “likes” and “shares” of posts mentioning Jeremy Corbyn
- 55,000 “likes” and “shares” of posts mentioning Yvette Cooper
- 50,000 “likes” and “shares” of posts mentioning Andy Burnham
Thus in August posts mentioning Jeremy Corbyn were getting over ten times the “shares” and “likes” of posts mentioning the other two main contenders combined.
It should be noted the electorate for the Labour Leadership was relatively small compared to general elections. A total of 422,000 voted and Jeremy Corbyn won with 251,000 votes.
The U.S. Republican Presidential Primary
In reality there are many factors that might help us predict the result of elections and at best social media engagement is just one factor. Social media interactions can also be both positive and negative. In the case of Donald Trump in the U.S. some posts are liked or shared because they are negative. Thus looking at total interactions is not enough we would need to look at some form of sentiment analysis if we were trying to establish support. Alternatively we can just look at engagement with their own Facebook page. On the assumption the author does not post anything negative about them themselves we can reasonably assume that “likes” and “shares” will be broadly supportive of the author.
If we use Sumo Rank (BuzzSumo’s free Facebook page analyser) we can see the interaction with posts published on Donald Trump’s Facebook page.
You can see the full analysis of Donald Trump’s page here. What the chart shows is that in July posts on his page gathered around 6m “likes” and “shares”. This compared to just over 3m “likes” and “shares” for posts published in the same month on Ben Carson’s Facebook page, see the chart below. However, we can see that total interactions for posts published by Ben Carson jumped in August to overtake those for Donald Trump. This engagement appears to be in line with the latest polls that show Ben Carson and Donald Trump are now neck and neck in the polls.
You can see the full analysis of Ben Carson’s page here.
Clearly the number of “likes” and “shares” are not the best indicator of how an election is progressing. However, if you
- get vastly different levels of engagement for politicians,
- sudden jumps in engagement, or
- a sharply rising trend
it does suggest there is something to be explored, if only to understand why the engagement is happening.
Opinion polls remain the best indication of an election’s progress and they are increasingly being refined, though there are occasions when they are also wrong, such as in the last UK general election. With social media we have a new measure of engagement around a topic or a politician. It may be biased towards certain groups and demographics but it can be an indicator of something resonating with an audience and engaging them both positively and negatively.
I suspect we will see a lot more analysis of social media engagement over the coming years. Could social engagement predict an election result in the future? Time will tell.