Can Facebook predict the winner of the 2011 NH presidential primary?

Last week, looking at Twitter as a predictor the Iowa caucus results or building tools to do so felt like the shiny object of the moment in the tech blogosphere. Some were more over the top than others but the genre is well established: adapt the numbers of followers and fans a candidate has on Twitter and Facebook to the classic “horse race” coverage of political campaigns. Used that way, it feels like the latest in a long list of reasons to conjecture that political coverage is broken.

As Micah Sifry suggested at techPresident, however, you shouldn’t believe the social media hype.

“There are so many ways that such changes might NOT be indicative of anything, or indicative of the opposite, that it almost seems silly to list them. A candidate might gain followers because he’s entertaining to his opponents. She might gain followers because of something outrageous that she says. The same with retweets. As the saying used to go, “a link is not an endorsement.” At best, it’s a very low-level indication of interest, an invitation to start a relationship that campaigns need to convert into real support. Similarly, we shouldn’t take big numbers of followers or “likes” as proof that a candidate has a really engaged base. Over on the Huffington Post, Alan Rosenblatt demolishes the notion that Newt Gingrich 1.4 million Twitter followers means he’s popular among Republicans. For starters, half of those accounts aren’t even in the United States. Newt’s numbers are a sign of online longevity and notoriety, not much more. And who can forget when Herman Cain was topping the Facebook “buzz” charts?”

Today at techPresident, Nick Judd published a feature on Twitter obsession that went even deeper into the issue and how problematic trying to prognosticate about the campaigns using social media can be. It’s worth reading the whole thing but here’s the relevant bit:

“The thing about attaching numbers to people’s names is that it usually makes them want to make the number go up. Call it gamification if you want. The truth is that it’s human nature, and as more people pay attention to social media, it is creating a sort of downward behavioral spiral. Candidates wanting more points on the social media scoreboard are urging supporters to tweet and post to Facebook on their behalf — spreading borderline spam on social networks and doing nothing to make the campaign season less of a horse race when that doesn’t necessarily have to be the case. “

Judd suggests that instead, political reporters and tech journalists (and those of us who try to straddle those worlds, on some level) should be evaluating the success — or failure of campaigns based upon better metrics.

“…why not evaluate them in the public online sphere based on metrics that say more about their potential as a public official? Especially over the past two years, information of this type has become easier for programmers to get to. The unfinished bit is making a competition in which a candidate can ‘win the Internet’ by accumulating the most small-dollar donors, or by making the most in-state campaign stops of an hour or more in a week, or by staking out the most detailed position on oh, I don’t know, the future of American Internet infrastructure.”

I think Judd is offering political journalists a great way to differentiate themselves from the pack of stressed, overworked correspondents chasing the same story in the same place. As primary season goes into hyperdrive, I can’t help but hope that some take up his suggestion.

I’ve been thinking a lot about the intersection of social media and government over the past couple years. Governance has proven to be a genuine challenge, at least with respect to the current architectures for digital participation we have. That could well change, as the civic Web moves from idea to reality. To make the point in a succinct way, policy makers seeking to leverage technology in governance in the 21st century might do better to look to the wiki, not the tweet.

Campaigns, however, particularly well suited to quicksilver sociality of media, with its capacity to share links to viral videos, rapid responses or pithy quotes. There’s a reason that reporters of many stripes, political or not, watch Twitter carefully now: it tends to be where news breaks first.

All that being said, I saw some data tonight that made me wonder, yet again, about social media’s potential value as a leading indicator, getting out ahead of the polls that campaigns and the media traditionally use to gauge how the candidates are doing. While Twitter still has low penetration in the overall population of voters, a majority of Americans online are now on Facebook.

Earlier this evening, I got off the phone with Jan Rezab, CEO of Socialbakers, whose company has been crunching social data from Facebook around the GOP presidential candidates. They’ve been porting the data into an infographic on elections at socialbakers.com.

There are several ways to read the data, said Rezab, including the overall status of a candidate, where they’ve been gaining followers over a long period of time. If we look deeper, said Rezab, the best thing is geometrics, looking at fan gains, people talking about the candidates, how many people do they reach and what topic they talk about.

So here’s the interesting data point — and yes, I may well have buried the lede — social data from Facebook shows former Utah Governor John Huntsman is trending sharply upwards, much as former Pennsylvania Senator Rick Santorum did before Iowa. Hunstman gained 1182 fans yesterday and, as of 7 PM EST tonight, more than 2000 fans, a 69% increase. The same is also true of people talking about him on Facebook, with some 8800 people on the 8th and approximately 11,0000 today, a 25% increase.

That’s “quite huge growth,” said Rezab. “He’s the one that’s trending right now.”

Will growth be a leading indicator for Huntsman finishing 3rd or even 2nd? Hard to say. The gold standard for political calculus is Nate Silver at 538, who is now at the New York Times. Silver blogged today that “Huntsman has momentum but lacks time“:

“Two new polls out Sunday night showed a favorable trend for Mr. Huntsman. A Public Policy Polling survey, which had a very large sample size, gave him 16 percent of the vote, enough to challenge Ron Paul for second place. Another poll, by the University of New Hampshire, had Mr. Huntsman at 11 percent of the vote, although this was improved from his 7-percent standing earlier in the week.”

Will that be enough to keep him viable and in the race? It’s hard to know. I’ll refer readers back to Silver’s analysis of Huntsman’s prospects on that count.

Here’s one more thing to consider: relevant academic research. Responding to a question on Twitter about whether Facebook was was predictive, Girish J. “Jeff” Gulati, a politics professor at Bentley University, replied that the growth in Huntsman’s “likes” and “mentions” was not. Rather, “our 2008 data suggest it would be response to positive news coverage of his debate performance and gains in NH polls,” he replied. According to “Social Networks in Political Campaigns: Facebook and Congressional Elections 2006, 2008,” co-authored with Christine B. Williams, suggests a causal relationship is credibel, if not confirmed by their research data and methods:

Preliminary analysis of the Iowa and New Hampshire contests (Williams and Gulati, 2008) suggests that Facebook supporters matters even more than candidate visits and television ad buys, and together these indicators explain very high percentages of the variance in candidates‟ vote shares. The predictive power of the model for all New Hampshire voters was lower than for the model of all voters in the Iowa caucuses, however, and the regression coefficients for Facebook support were not statistically significant. The model for the youngest age cohort in New Hampshire was highly predictive, and there was a very strong and highly significant relationship between actual vote share and Facebook support among 18-24 year olds. Another approach would be to include as an independent variable the percentage voter turnout for 18 to 29 year olds in each state‟s nomination contest. If Facebook support is simply a surrogate for young people‟s heightened political engagement in the 2008 election, its independent effect would wash out.

If future research can confirm these findings and demonstrate a causal connection between online strategies and votes, then Facebook and other social networking sites will be an essential tool in enhancing the democratic process. These sites go beyond simply communicating the campaign‟s theme and information on how to make participating easier. Active engagement by the candidate and a well maintained site can make the candidate more accessible and seem more authentic. It also can encourage a more professional discussion among supporters. In addition to personalizing the candidate, Facebook puts a face on the candidate‟s other supporters and facilitates interpersonal connections around activities other than politics. And because Facebook organizes members by regional and organizational networks and gives greater access to profiles in one‟s own networks, offline meetings and connections are a real possibility. As membership in traditional civic associations declines, we see in these networks a new frontier for cultivating social capital, which candidates, elected officials and civic leaders can tap when they want to mobilize citizens for political action.

Will rapid changes on Facebook predict the winner of the 2011 NH presidential primary? There’s little doubt 2008 Facebook was a very different place, in terms of scale, functionality and integration into the lives of Americans online, than 2012 Facebook is today. Tens of millions of citizens have tablets and smartphones now and access Facebook from where ever they are. As we saw last Sunday, television networks and Facebook are increasingly integrating debates.

There’s “definitely a lot more going on Facebook and other social media this time around,” wrote Gulati in a follow up email. “In fact, every two years we see so much growth and change. I’ve been studying this since 2002. The beauty of studying this field is that there always exciting and you are never short of ideas for a new study. I still am sticking with my view that it is not a strong predictor of future success at the polls.”

Gulati listed three rationales for this assessment:

1. Much of social media strength is a function of longevity on Facebook and national visibility. “Gingrich, Romney, Paul, and Bachmann start out strong simply because they ran before and carry over their supporters or are members of Congress that are known nationwide,” wrote Gulati.

2. Some of the candidates who are stronger on social media are in that position because they are shut out of traditional sources. “Ron Paul and Buddy Roemer are cases in point,” wrote Gulati. “So my point is not to look at social media in isolation of other media.”

3. Social media isn’t integrated with the campaign. “From our interviews with Congressional candidates and staff in 2008 and 2010 (~150 interviews), we found that the staff really does not know what to do with these likes and followers,” wrote Gulati. “So even if a social media consultant is succeeding in growing the candidate’s online presence, it really does not go much past that. I think that as the number of followers grows, it actually will be harder for campaigns to truly engage with them because there are just too many. A similar problem was happening in Congress with regards to e-mail overload. I’m not sure if they really have resolved the problem.”

Gulati attributes rapid changes in likes or mentions to external events, which is makes sense. The one billion dollar question, is whether a social media following will lead to offline action, in this case results at the polls:

Of course, your post was referring to the change in supporters rather than actual numbers. I think clearly that indicates some buzz around the candidate. But I still see it as a reaction to external events as in 2008 rather than social media pushing overall numbers mostly because I haven’t seen any evidence that social media strength can move poll numbers or actual results. I think one reason that so many of us who study social media give social media such an elevated prominence because that’s all we see or what we see on a regular basis.”

Do you think the growth in user base and political involvement will lead other researchers to different conclusions? Or, as Sifry and Judd eloquently argued, should the idea or the practice of looking at politics from the perspective of a social prism alone simply be consigned to the “dustbin of twistory?”

Comments are open. And, even more important, the first votes of the presidential primary season will be cast as the eyes of the nation turn towards Dixville Notch .

About Alex Howard

Alexander B. Howard is a DC-based a technology writer and editor. Previously, he was the Washington Correspondent at O'Reilly Media, where he covered the voices, technologies and issues that matter in the intersection of government, technology and society. If you're feeling social, you can follow him on Twitter, like him on Facebook or circle him on Google Plus In addition to corresponding for the O’Reilly Radar, he has contributed to the Huffington Post, Govfresh, Mashable, ReadWriteWeb, National Journal, The Atlantic, CBS News and Forbes. He graduated from Colby College with a bachelor's degree in biology and sociology. Currently, he is a resident of the District of Columbia, where he lives with his greyhound, wife, power tools, plants and growing collection of cast iron pans, many of which are frequently used to pursue his passion for good cooking.

5 thoughts on “Can Facebook predict the winner of the 2011 NH presidential primary?

  1. Alex,

    A solid piece. I don’t disagree on the crux of your argument – that coverage of politicians on social media shouldn’t just be about the number of fans or followers. That’s clear.However, that’s not what I’ve been doing at Mashable. Our Twitter sentiment analysis with Globalpoint wasn’t just about number of followers or tweets. the weighted metric they used attempted to include interaction as well. That may or may not be perfectly quantifiable yet, but I think perfection will come as a result of experimentation.

    The Roemer piece, also, was much more than “look at how many fans Buddy Roemer has gotten!” It was a look at how he got those fans and the unique ways his campaign was using social to get a message across when more traditional media chose to ignore him.

    Along the same lines, I think Gulati’s criterium are off the mark. Bit-by-bit:

    1. This should, of course, be accounted for in quantitative analysis of social media success. Little disagreement there, except on the point that it can be done, I suppose.

    2. Yes, candidates who are shunned by traditional media often turn to social media because it’s accessible and cheap. Of course all forms of media are to be considered, and it would be interesting to start comparing spending on social vs. traditional media. However, look at Obama ’08 – certainly no lack of traditional media efforts, but also a hugely successful social campaign.

    3. That’s becoming less and less the case, depending on which campaign we’re talking about. I find – and this is circumstantial – that the more organized campaigns know exactly how to use social to achieve goals.

    Let’s continue the conversation. It’s an important one to be having, and I would love for it to result in better understanding of social media in politics and better coverage of politics in general.

    Thanks,
    Alex Fitzpatrick
    Mashable

      1. Of course. I think it’s a really important conversation to be having – how we can best cover digital in politics without generating white noise. Looking forward to seeing the debate continue and have an impact on our work.

        Politico has jumped on the game too, joining forces with Facebook.

        Thanks,
        AF

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>