Facebook, Twitter Election Results Prove Remarkably Accurate

The Influence Project

Candidates in yesterday's mid-term elections fought hard to gain voters and leads in polls, but what about Facebook fans? According to figures released this morning, popularity on Facebook provided a surprisingly accurate portrait of House and Senate races.

Facebook's political team said that candidates with more Facebook fans than their opponents won 74% of House races and 81% of Senate races. Though Facebook accurately predicted most elections, some big upsets included Christine O'Donnell, Meg Whitman, and Sharron Angle, all of whom had far more fans on Facebook, yet lost their elections.

Still, Facebook is rapidly becoming a social voting platform, one that campaigns will certainly focus on in 2012. A remarkable 12 million people clicked the "I Voted" button this election cycle, a massive 122% increase from 2008's 5.4 million voters.

We're still analyzing how Twitter stacked up—stay tuned—but in Nevada's major contest between Angle and Harry Reid, the micro-blogging platform may have been far more accurate than Facebook—and even scientific polls. In the days leading up to the election, most every poll showed Angle ahead of Reid; however, data compiled by Crimson Hexagon, which uses a sophisticated algorithm to analyzes Twitter conversations online, showed Reid ahead of Angle 55% to 45%.

Reid ended up winning, proving polls wrong, with 50.2% of the vote, compared with Angle's 44.6%.

Add New Comment


  • Josh Weinberger

    No one can quibble with the underlying point here — social networks are increasingly a factor in elections.

    But there's not much weight in noting that "12 million people clicked the 'I Voted' button this election cycle, a massive 122% increase from 2008's 5.4 million voters" -- after all, Facebook itself has grown at a far greater rate over the last two years.
    Aug08 - 100M
    Apr09 - 200M
    Sep09 - 300M
    Feb10 - 400M
    Jul10 - 500M
    I don't know how many of those members (at any of those points) were registered U.S. voters, but it's a bit misleading to call a 122% increase "massive" in the face of the truly massive growth of overall membership.
    If anything, on a percentage basis, I wouldn't be surprised if the level of political activism actually dropped.
    That is:
    Share of eligible 2010 Facebook members who clicked "I voted" < Share of eligible 2008 Facebook members who clicked "I voted"

    Anyone have figures on that?
    Even back-of-the-napkin estimates?
    Here are mine -- and please don't treat these figures as anything other than the biggest of ballparks.

    5.4M in 2008 was roughly 5% of Facebook's 100M+ membership at the time. (again, that's full membership, not just U.S. registered voters, and using the Aug08 membership total, not Nov08, which I don't have available.)
    12M in 2010 is roughly 2.4% of Facebook's 500M+ -- and, again, that's the Jul10 membership figure; Facebook has added 100M every 5 months or so, which would mean the 12M "I voted" clicks represent an even smaller share.

    As for Crimson Hexagon - I'm not familiar enough with its methodology to accurately judge, but if I'm reading the company's materials correctly, CH makes no allowance for the geographical source of the chatter it evaluates. In other words, it includes sentiment from nonvoters and voters alike, out-of-staters as well as Nevada residents — making its results incredibly unreliable as a gauge of "sentiment among likely Nevada voters."
    The fact that the algorithm happened to produce a result similar to the actual outcome may be nothing more than a fluke. (And, by the way, not all ** that** similar -- Reid's actual 5-point margin is far smaller than the 10-point drubbing Crimson Hexagon was predicting.)