Startup from Huntsman Business School Succeeds at Predicting Political Outcomes
PoliticIt Algorithm Parses Social Media, Internet, Traditional Media Successfully to Predict Winners
Question: Can social media predict election outcomes?
Answer: Yes, at a Huntsman School startup at Utah State
Editor’s note: This press release about PoliticIt, a startup from Huntsman Business School students at Utah State, went out on Business Wire to thousands of press contacts and websites throughout the mountain west. It has been viewed more than 2,600 times and picked up by more than 440 news outlets, including such powerhouses as YahooFinance (which gets 152 million visits per month), MarketWatch, CBS MoneyWatch, and the Boston Globe.
LOGAN, Utah – It’s the puzzle so many are trying to solve—seeing the future. Can “Twitter sentiment” accurately predict the movement of Facebook’s stock price? Can the Romney campaign reliably forecast a Super Tuesday win based on social media trending?
Firms such as Gallup, A.C. Nielsen, Attensity, Netvibes, DataSift and Lithium are racing for the answer. But a small startup called PoliticIt in northern Utah got there first.
In the last month, PoliticIt examined its data on more than 90 state political convention elections all across the nation—contests for the Senate, the House and for governor—and it had correctly predicted the outcome in a remarkable 87 percent of those races. PoliticIt accurately called the results in 100 percent of the convention elections in Utah, including in three “upset” cases that traditional polling got totally wrong.
How did this six-month-old startup, created by five Jon M. Huntsman School of Business students and Professor John D. Johnson, develop a functioning political crystal ball before giant firms such as A.C. Nielsen?
Well, some firms analyze just one or two social media channels; DataSift and Attensity, for example, analyze just Twitter.
The more ambitious approach is to build an analysis machine that swallows “big data” from all of the sources—from all across the Internet, from many forms of social media including Facebook, YouTube, Twitter, InTrade, Wikipedia and more, and from the vast numbers of traditional media. Then the challenge is to understand that data, to identify the powerful “influencers,” to read the “word clouds” and to build a statistical model that predicts the outcome.
That’s what PoliticIt has done.
“We add over 30 megabytes of text and data every day to our database, which is now close to three terabytes,” said Josh Light, a co-founder and CEO.
After that, it’s all about the algorithm, as the Google boys, Brin and Page, like to say.
PoliticIt has a powerful algorithm that can measure “digital influence” by calculating an "It Score"—a number from one to 100 that signifies a candidate’s digital influence. These "It Scores" accurately predict, significantly in advance, both poll and election results, says Sterling Morris, a co-founder and chief marketing officer of the company.
In its next stage, PoliticIt will read its “data tea leaves” on specific geographic regions, down to the county level. It will identify those regions that are both trending negatively for a candidate and also open to change, if voters there are offered the right “influencer” intervention.
The Jon M. Huntsman School of Business at Utah State University seeks to inspire and equip students to become innovative, ethical leaders with refined analytical skills that will help them understand and succeed in the global marketplace. The Huntsman School of Business is one of eight colleges at USU, located in northern Utah. More information on the Jon M. Huntsman School of Business may be found on the web at www.huntsman.usu.edu.