Forget Gallup, Zogby, Rasmussen—how can you trust their numbers? They're biased; they're inconsistent. Rather than depend on polls these days, we're more and more turning to polling aggregators like Real Clear Politics or statistics whiz kids like Nate Silver. But why not use social media to measure a candidate's chances in the mid-term elections?
That's exactly the aim of Crimson Hexagon, a Harvard-developed analytics tool that CNN will use Tuesday to measure social media chatter in real-time. The company (which you can read about in more detail here) has given Fast Company readers a sneak peek at the results in two battleground states. Here's how the opponents in California and Nevada stack up online.
Over the past month, Crimson Hexagon has monitored the public's tweets about the Senate race between Sharron Angle and majority leader Harry Reid. RCP poll averages show Angle is up on Reid by a slim margin; on the Web, Angle has more followers on Twitter and fans on Facebook. However, according to Crimson Hexagon, which analyzed more than 45,000 tweets and found 32,209 opinions, pro-Reid and anti-Angle opinions make up 55% of the chatter on Twitter. The pro-Angle and anti-Reid opinions make up 45% of the tweets.
In California, where incumbent Barbara Boxer is up by just a few percentage points against Carly Fiorina, Crimson Hexagon found the race is even tighter than polls have shown. Of the 21,324 mentions of the race culled from Twitter the blogosphere, 10,456 opinions revealed pro-Boxer opinion to make up only 26% of the Web talk, whereas pro-Fiorina opinion captured 37%. However, anti-Fiorina opinion made up 24% of the tweets, while anti-Boxer opinion represented just 14%, bringing the totals to a near dead heat.
So what kind of impact do these figures have on the election? When I spoke recently with e-campaign expert Mindy Finn, I asked her whether social media analysis might be more accurate than polling data.
"I wouldn't want to compare social media to poll numbers, just because poll numbers seem to be way off this election cycle," she explained. But Finn added that measuring data from Facebook and Twitter was not "100% reliable." Indeed, Crimson's data wasn't filtered by location, meaning online opinion wasn't limited to California and Nevada.
Will Reid top Angle in a surprise upset, as Crimson's analysis suggests, surprising traditional pollers? Will the Fiorina-Boxer race end in a neck-and-neck finish? We'll be tracking the data—and social media chatter—and have an update after the election.