Site Metrics and Web Analytics by NextSTAT

Thursday, November 8, 2012

So Did Social Media ‘Predict’ the Election?

Nate Silver’s being hailed as a Big Data hero after correctly predicting an overwhelming majority of states in Tuesday’s presidential election. But how did social media sentiment fair as a barometer of the electorate?

Before the election was called, Mashable reported that Twitter sentiment in the hotly contested battleground states was split: it favored Barack Obama in Michigan, Pennsylvania, North Carolina, Ohio, Florida, New Hampshire and Iowa, while Romney had the lead in New Mexico, Virginia, Colorado and Nevada.

So did the Twitter sentiment “predict” the election? Not really: Michigan, Pennsylvania, Ohio, New Hampshire, Iowa, New Mexico, Virginia, Colorado and Nevada and likely Florida — where the results haven’t yet been called — went to Obama, while only North Carolina went to Romney.

This does, however, depend on how you look at the data: Twitter’s pre-election chart showed a trend in Obama’s favor across the board just prior to Election Day:


According to social media analysis frim Attention, powered by Tracx, Obama also dominated the online conversation at large, taking up 64% compared to Romney’s 36% across Twitter, Facebook, Tumblr, blogs and message boards.

Social media sentiment is far from a scientific poll — after all, only a minority of registered voters in the U.S. are actively posting about politics.

However, Twitter is fond of the fact that its Political Index tends to track Gallup polls, a sign that it could potentially be taken seriously sometime down the road as more voters use the platform to talk politics. And as sentiment analysis gets smarter and better every day, we could be looking at a new and valuable data point for campaigns sometime not too long down the road.

Do you think social media can ever be taken seriously as an elections predictor?

Source : mashable

0 σχόλια:

Post a Comment