Arc 3 Communications recently participated in Big Data Week in Atlanta. The goal of Big Data Week was to hold community events focusing on the social, political, technological and commercial impacts of Big Data. One of the many discussion items of the week among data scientists, marketers and social media experts was the impact of social data, primarily in the examination of consumer behavior and the ability to predict sales and craft marketing messages.
However, social data is also a powerful tool for understanding the American voter, and like traditional phone surveys can help in the understanding of voters’ opinions on an issue and their intensity. Social data which consists of the analysis of tweets, posts and discussion items on social media channels, forums and blogs, may also be a future indicator in predicting voting behavior. A recent study by social scientists at Indiana University of 800 competitive Congressional races in 2010 and 2012 revealed that getting more tweet mentions relative to your opponent meant higher vote margins. The social scientists found that the frequency with which a Republican is named correlates with the Republican vote margin in the subsequent election, independently of other factors such as incumbency, media coverage, partisanship and demographics.
While the ability of tweets to predict the vote is a topic of much discussion, the use of social data as a tool in understanding the opinions of voters and how to develop targeted messaging to them is well established after the last election cycle. The Obama 2012 campaign through its 200 digital and 50 analytics team members was able to utilize social data to develop effective message development and targeting strategies while also identifying advocates for the campaign.
In the area of advocacy, the Obama campaign developed an algorithm to score over 5oK accounts by political id on Twitter. This identification of accounts favorable to the President utilized the text of the tweets and did not look at @mentions. The analysts at the Obama campaign discovered that @mentions referencing the President were largely negative. However through text analysis of tweets, the Obama campaign was able to identify Twitter users favorable to the President. The Obama campaign also looked at Twitter influence (number of tweets and followers) to score and identify favorable accounts. The Obama campaign then sent a direct message on Twitter from President Obama to the scored accounts asking them to advocate for the campaign, which in turn many of them ended up doing.
Another powerful use of social data for advocacy by the Obama campaign was the development of a Facebook App that enabled supporters to match their Facebook friends with the Obama campaign’s various lists and then send a message to undecided voters in key swing states asking them to take action such as sharing viral content, signing up for the campaign or voting on Election Day. This tactic known as Facebook targeted sharing, was developed by the Obama team to offset the 50% drop in landline usage among 18 to 29 year olds. Despite being difficult to reach by phone, 85% of this same group were friends with an Obama 2012 Facebook App user. Through the Facebook targeted sharing app over 600,000 Obama supporters were able to reach over 5 million people by Election Day. 20% of the 5 million reached took some action, such as registering to vote. Hundred of thousands used the tool in the days leading up to the election. The Obama campaign’s Facebook targeting app represented a significant advancement in voter contact – overlaying traditional voter file data with social graph information and relationships.
The supplementation of a voter file with social data will become the norm by the 2016 election cycle and innovative campaigns are already taking this task on for 2014. One of those chief innovators on the Republican side is U.S. Senator Mitch McConnell of Kentucky whose campaign is creating a statewide voter file that adds social data so that the campaign can track and target individual voters based on what they’re saying online. According to Senator McConnell’s campaign, it has already matched the voting records of close to 40 percent of Kentuckians – more than 1 million voters – with their online persona. Senator McConnell’s campaign plans to monitor in real time the social and online outputs of voters in Kentucky in order to understand voter perceptions about issues and the candidate. Through this process, the campaign will be able to measure voter sentiment and the level of engagement on various issues.
Social data is the future of campaigns and a powerful new tool that can help a candidate understand voter opinions on key issues and serve as an advocacy tool. In time, social data may be able to accurately predict voting behavior in a race. As campaigns begin to experiment with social data, it will be critical that the proper metrics and algorithms are established. Campaigns will need to ensure that they connect their social media and social data operations to their off-line grassroots and get-out-the-vote operations in a meaningful way that translates into supporters and votes. This will become an increasing challenge for national and statewide campaigns in the immediate future.