Big Data

Election Tech: Data from social shows big Sanders jump from Jan 17 debate

The latest Twitter data coming out of the January 17 US presidential debate for Democratic candidates reveals a correlation between debate exposure and online activity.

Image: William Stodden/Stata

Presidential campaigns are won by orders of increments, not magnitude. In 2016, social media is an important tool that allows candidates to fine tune messaging, and get a better understanding of the electorate.

As the Democratic race tightens, some of that subtext and nuance are being revealed within Twitter data. During the 2016 presidential campaign, TechRepublic is reporting on how technology, big data and social media are shaping the outcome of the race. Last week we analyzed Twitter activity and extracted three takeaways related to presidential candidates and the first Republican debate of 2016. This week, we repeated the process and took at look at Democratic Twitter activity around Sunday's debate.

Comparing last week's data to this week, we can draw a few conclusions. Here are the three most significant Twitter data takeaways from January 17's Democratic debate.

Takeaway 1: @BernieSanders won big

@BernieSanders added a whopping 18,073 followers during the Democratic debate. Sanders' debate Twitter growth trumped Trump's performance during last week's GOP debate by nearly 8 thousand followers. @HillaryClinton added 9,098 followers during the debate, and @MartinOMalley tallied 4,993 new followers.

Clinton had 44% higher new followers Sunday night compared to the same amount of time on Thursday Night (during the Republican debate).

O'Malley had 3,741% more followers Sunday night, compared to Thursday Night.

Interestingly, Sanders' account was added to twice as many Twitter lists twice as O'Malley, and several times as many lists as Clinton's account.

Image: William Stodden/Stata

Takeaway 2: @RealDonaldTrump is (still) winning big

The Republicans also earned several thousand followers on Twitter during the Democratic debate, and @RealDonaldTrump's growth remained robust. On Twitter Trump continued to pummel his opponents, and tacked on 10,214 new followers. @TedCruz, added 2,324 followers, and @MarcoRubio earned 1,648 new followers on Sunday evening. It certainly didn't hurt that Clinton and Sanders mentioned him several times.

Kasich's and Christie's new follow rates were down 79%. This may indicate that Christie's momentum coming out of last week's GOP debate was primarily among lower information voters. It appears Christie was added because few people were paying attention to him before, and started to as a result of last week's debate.

Takeaway 3: Debate exposure matters

National television exposure on the debate stage may have had a positive impact on Martin O'Malley's Twitter account. Over the course of the debate, O'Malley's account gained followers at a clip faster than his Democratic counterparts, experiencing relative growth of 3.98%, more than twice Sanders relative growth of 1.53%, and Clinton 0.17%.

Analysis of relative growth must be accompanied by the obvious disclaimer that candidate follower growth rate is relative only to to their own account. Relative growth is simply a measurement of percentage of new followers during and after the debate. Because O'Malley's Twitter account is smaller than both Clinton's and Sanders' accounts, he requires fewer followers to demonstrate a significant uptick in growth velocity. His account growth was large enough to suggest that broadcast face-time— even on a debate late on a holiday weekend—is still significant.

Image: William Stodden/Stata

Click here to automatically subscribe to the Election Tech 2016 newsletter

TechRepublic's Election Tech 2016

This is a study and analysis of Twitter data only. Specifically this is an index of candidate account data sampled prior to and after the final Democratic debate before the 2016 Iowa caucus. This is not an analysis of electorate data. This study is not a poll or prediction.

As expected, during the debate Democrats gained more followers overall. The Republicans were not debating, and—just as the Democrats were last week—GOP data was included in our sample as a control. In this context, their followers can be seen as people following politicians on Twitter.

Our charts were produced in cooperation with political scientist William P. Stodden, using Stata, a common data analysis tool. There was no magic involved with how we acquired our data. We pulled snapshots at approximately 5:30 pm Eastern Standard Time on January 17, and at 8:30 am on January 18, and included follower count, listed count, followed by, to following ratio, and other publicly available data available on candidate accounts.

Over the course of the campaign, and at major events like the Iowa caucus and the New Hampshire primary, we will gather data snapshots of publicly available candidate social media data. Throughout 2016 we will grow a historic archive that we hope will inform a number of stories about election technology.

We'd love to hear from you, if you're a data scientist, TechRepublic reader, or just curious about the correlation between social media and American politics. Please leave your ideas in the comments below, or ping us on Twitter @TechRepublic.

See Also:


Dan is a Senior Writer for TechRepublic. He covers cybersecurity and the intersection of technology, politics and government.

Editor's Picks

Free Newsletters, In your Inbox