Though his Super Bowl predictions didn’t pan out, Nate Silver still had a great year. His algorithm (which incorporates various polls), led him to correctly predict not just the outcome of the presidential election, but the state-by-state breakdown, as well.
But polling might not be the only reliable indicator of electoral success. I recently sat down with J.D. Chang, founder of TrendPo, and Morgan Muchnick, TrendPo advisor and president of The M2 Group, to discuss emerging ways technology can be used to help predict election outcomes.
“We’re seeing a correlation between politician buzz and election results,” says Chang. “On Election Day, national politicians who had more buzz and momentum according to TrendPo Rank won all 11 of the gubernatorial races and 24 out of 33 Senate wins, including the two Senate races that Nate Silver missed.”
What does he mean by “buzz”? According to their website, TrendPo uses an “algorithm that scrapes National Media, Beltway Media, Facebook pages, Twitter profiles, and sentiment analysis to create a daily rank of all national politicians.” In other words, candidates who generate positive earned media — as well as Facebook “likes,” etc. — are more likely to win.
Though correlation does not necessarily imply causation, aside from using the data to predict elections, individuals candidates can also use the data to determine which specific messages resonate.
For example, Chang tells me that Arizona Gov. Jan Brewer has a tremendous Facebook following, and Chang and Muchnick believe this is for a variety of reasons — including the fact that her team creates cool Facebook images. The two saw a spike her her Facebook activity, and traced it back to this posting:
Still, the ability to use news and social media to predict elections provides arguably the most interesting potential for growth.
TrendPo also allows you to compare politicians against one another. Like polls, this is just a snapshot, subject to change. Here, we see that Joe Biden is winning the media war against Chris Christie, Marco Rubio, and Paul Ryan. But, unlike polls, you can run this chart every day:
According to Chang, this is an idea whose time has come: “Political polling is an antiquated process. Its high costs, sampling bias, and lagging speed to breaking news is criticized in every election season. We think we have an alternative to measuring how politicians are performing in day to day buzz and future election chances.”