Originally published on LinkedIn.com on May 6, 2016 By Ben Zeidler
This post was written in conjunction with the Customer Intelligence team at Tenthwave, including Eric Schwamberger, Luciano Calvaruso, Rebecca Beyer, and our social listening partners at Talkwalker.
In this 2016 election season, social media is taking center stage as one of the core components of running a successful campaign (especially on a tight budget). It offers real-time intelligence, operates in massive amounts of volume, can easily be earned rather than bought, and is able to be geotagged. As a result, it offers a much different type of currency than what traditional endorsements typically provide to the candidates. While social media has been a part of campaigns before, 2016 represents the first election in which social media is at the forefront of the decision making process.
Barack Obama ran a fabulous Digital Marketing campaign in 2012, deftly employing CRM and email Marketing, in becoming our first “Digital President.” However, he only used Social Media as an add-on, spending only 4% of his 2008 Internet Marketing budget on different channels. A lot has changed in the 7 years since his first election, with the vast majority of his press team now focused on social communication:
“The current 14-member staff of the White House Office of Digital Strategy is slightly larger than the entire press secretary’s office of George W. Bush in 2005 (12) or even Obama’s in June 2009 (13),” said Martha Joynt Kumar, a Towson University political science professor.
Obama is considered to have become the first “Social President” during his tenure, but the 2016 race we’re now experiencing is widely considered to be the first Social Media election.
Recently, Tenthwave was lucky enough to host Justin McConney, Donald Trump’s Director of New Media, in our offices for a fireside chat with UK based industry media publication The Drum. McConney spoke at length about Trump’s approach to social media, and how it offered certain advantages to the candidate.
McConney’s assertions appear to be backed up by date from The New York Times, which shows the strength of Trump’s ability to earn media, despite the fact he is 8th among campaign spenders. But how does that earned media (heavily earned through social) translate – if at all – into votes?
As we combed through the last 6 months worth of social data on all relevant mentions related to Donald Trump (roughly 11.6 million in total) via our digital listening partner, Talkwalker, we had a core methodology and goal in mind – to understand the impact of social identity factors, such as
- topic (immigration, etc.),
- impetus (appearance, quote, etc.),
- medium (social, traditional, etc.),
- geography (localized, national, etc.),
- and sentiment –
and how those factors correlate aggregated poll data (in this case, the RCP Average from noted quality site Real Clear Politics). Put more simply: does social identity drive votes?
Two insights immediately became clear.
- First, that a candidates reach in social has a definitively positive correlation to their poll rankings. This may have been expected and predictable, but our data was able to prove it out.
- Second, and more surprisingly, that sentiment in social is oftentimes inversely correlated to poll numbers (i.e. negative mentions correlate to positive poll numbers). In the case of Trump, negative sentiment actually has a higher correlation to polls (3X) than positive sentiment, suggesting that when he makes a polarizing statement, the conversations generated are creating more benefit than harm to his poll numbers. This was an incredible finding, and essentially means that Trump needs three positive mentions to equal the good done to his campaign by one negative mention. Could these negative Trump mentions be creating a groundswell among Trump supporters, encouraging them to come out to the polls? Could they be the reason he has all but clinched the nomination?
In the case of Trump, negative sentiment actually has a higher correlation to polls (3X) than positive sentiment, suggesting that when he makes a polarizing statement, the conversations generated are creating more benefit than harm to his poll numbers.
An interesting wrinkle in the sentiment data is that it typically takes about 5 days for the social surge related to the sentiment-driving statement to maximize their impact in polls. Remember that social is real-time, but polls are not. Negative sentiment increases more significantly over the 5 days following an event, but all mentions (both positive and negative) increase over that period. Companies like Dataminr and political pollsters make money on being real-time and ahead of the story, but it appears that when it comes to social influence, real-time is a few days ahead of the larger poll-related impact.
We looked deeply at correlation between medium and positive polling, as Trump is known for his ability to utilize different channels to get his message out. When Trump was mentioned on YouTube, Instagram, Facebook, or Twitter, that had a higher correlation to positive polling numbers than more “traditional” mediums such as blogs and newspapers. Radio and TV, once strongholds of any campaign, showed almost no correlation to gains in polls.
What does all this data mean for a candidate looking to use social to drive votes? It’s still unclear what the power and value of a social endorsement is, especially with Trump’s negative mentions having greater value to his campaign than positive ones. Only time will tell if the volume of people endorsing a candidate in social will impact the election, but its undeniable that social reach (whether negative or positive), driven by social identity, is related to a candidate’s ability to rally voters in the early-mid stages of the primaries, and in many cases, newer forms of media like Instagram may provide a better proving ground for rallying those voters.
Social Identity and the Changing Face of Politics: a summer of data.
Next, during the post-convention phase of early August, we’ll reflect more specifically on each party’s nominee. We’ve got some fantastic data around which candidates are winning which key topic areas, which have the highest adjusted net sentiment (positive-negative), and geographic implications of sentiment and reach. All the data we’ve explored to date will be updated again, and we’ll attempt to further correlate different types of mentions with some of the first head-to-head polling results.
I’m the VP, Customer Intelligence at the Customer Obsessed Digital Agency®, Tenthwave. If you use twitter, please follow me @BenjaminZeidler (if you like ramblings on digital, sports, and food) and @Tenthwave (if you like a daily dose of interesting finds).