Who won in Iowa — Facebook, Twitter or the polls?

The presidential primary season, which finally got down to actual voting with Tuesday’s Iowa caucuses, isn’t just the official process of choosing a nominee for president. This year, it’s also being watched as an unofficial referendum on social media’s accuracy in taking the pulse of the electorate.

How does social media compare with traditional polling methods? And which of the two biggest platforms fares better, Facebook or Twitter?

With one of 50 states reporting, it’s difficult to project a clear winner, but the caucuses did provide a glimpse at the wisdom of tracking the online crowd.


Related story:

Social media can predict the future!


Start with the official results: Mitt Romney won the caucuses by eight votes, edging the late-surging Rick Santorum, 30,015 to 30,007. Ron Paul came in third with 26,219, followed by Newt Gingrich with 16,251, Rick Perry with 12,604, Michele Bachmann with 6,073, and John Huntsman with 745.

How does that match up with the social media buzz?

As the caucuses got under way Jan. 3, the social metrics analyst Sociagility had Paul as the clear front-runner, based primarily on his Facebook presence, Venture Beat reported.

Sociagility’s co-founder Niall Cook told VentureBeat that Paul had the greatest social media reach and, despite having fewer “Likes” than some other candidates, led in most metrics on Facebook. 

“The popular vote for the nomination is going to be won or lost on the most popular social network, Facebook,” Cook said.

So maybe Facebook shouldn’t take itself to Vegas just yet. Twitter, on the other hand, might have shown a better feel for the way the pre-caucus trends were unfolding.

The technology news site Mashable and Global Point Research compared Twitter activity in advance of the caucuses against a national poll to see how they matched up. The data was taken from an NBC News-Marist poll and Twitter activity between Dec. 27 and Dec. 30 and found that Twitter mostly reflected the NBC poll with one big exception — Twitter showed a much more pronounced rise for Santorum.

The NBC poll had Romney in the lead, followed by Paul, Perry and Gingrich about even, and then Santorum, Bachmann and Huntsman. (Most other national polls had similar results, though many of the final polls showed Santorum in third place. The Huffington Post collected and compared 11 of them.)

But in the Twitter data, Santorum was on fire, with about 35 percent of activity compared with Romney, who was next with about 21 percent.

Mashable pointed out that Santorum’s big Twitter numbers in the days before the caucuses didn’t necessarily mean he’d win, just that he was suddenly getting a lot of attention. And that seemed to be reflected in the votes, where Santorum finished higher than polls predicted.

Santorum, who had been an afterthought for much of the interminable debate season, had been gaining steam recently, and although the polls reflected his rise from nowhere to contender, Twitter’s data appears to have been more current. Global Point said its research shows that Twitter tends to be about two weeks ahead of polls in reflecting popular sentiment.

The idea of using social media activity as a predictor has been tried before, in forecasting such things as stock market performance, movie ticket sales and the spread of flu, often to good effect. And campaign managers will hone their skills at using, interpreting and, most likely, manipulating social media.

But it’s also worth remembering that, wisdom of the crowd notwithstanding, platforms such as Facebook and Twitter are designed to follow people and events rather than forecast them, and that their users might not always reflect the demographics of the general voting public.

And ultimately, of course, there’s only one poll that matters. It comes from that old-school technology called “voting.” 

 

About the Author

Kevin McCaney is editor of Defense Systems. Follow him on Twitter: @KevinMcCaney.

Reader Comments

Fri, Jan 6, 2012

Wondering why four states get to vote and choose the one republican who'll run against the incumbent. Why is this not something we all get to vote on? Why do a few get the privilege of making this decision?

Thu, Jan 5, 2012 dp dc

I am glad that it is recognized that social media cannot yet be used as a predicter of elections. Users are highly self-selective, not a random sample of voters. Social media is currently heavily weighted to the younger side of the electorate. As for the correlation between news mentions and elections, it is nothing valuable. Since they all are following the polls, the news media are writing about who they see at the top. Witness the news as Bachmann, Perry, Cain, Gingrich, and Santorum each rose. I would argue that a cause of media stories on front runners is the polls, not the other way around.

Wed, Jan 4, 2012 Niall Cook

Great analysis. There's a difference between analysing social media performance and predicting real world events and we certainly didn't want to attempt the latter. Having done our own post-election analysis, there's one channel you've missed: the good old media. There's an unbelievably strong correlation (0.9) between the number of 'news' mentions of each candidate in the week prior to the caucus and the final outcome. Who says the media are dying! Niall Cook, Co-founder, Sociagility

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above