Jadrian Wooten, left, and Ben Smith analyzed a billion tweets and found the
popularity of a pundit hinges more on whether he or she is confident than right.
(Photo courtesy of Washington State University)
PULLMAN, Wash. – It would be nice to think the pundits we see yelling on TV and squawking on Twitter are right all the time. It turns out they’re wrong more often than they are right.
Now two Washington State University economics students have demonstrated that it simply doesn’t pay as much for a pundit to be accurate as it does to be confident. It’s one thing to be a good pundit, but another to be popular.
“In a perfect world, you want to be accurate and confident,” says Jadrian Wooten. “If you had to pick, being confident will get you more followers, get you more demand.”
Wooten made his discovery with Ben Smith, a fellow doctoral candidate in economics. Smith originally wanted to test the accuracy and confidence of stock market pundits, taking inspiration from stock watcher and CNBC host Jim Cramer, whom Wooten describes as “the yelling genius that he thinks he is.”
But stock predictions rarely come with a date at which one could say a pundit was right or wrong. Sporting events do, so Smith made a software program to sort through more than 1 billion tweets for predictions of the 2012 baseball playoffs and World Series and the 2013 Super Bowl.
The program pulled out tweets with team names, nicknames and expressions commonly associated with predictions, like “beat.” Where they might rate the confidence of a television pundit by how loudly he or she yelled, a scale of word strength pegged words like “vanquish,” “destroy” and “annihilate” as expressions of confidence.
Their hypothesis: Pundits have a false sense of confidence because that’s what the public, seeking to avoid the stress of uncertainty, craves.
“They’re trading away some of their accuracy to be a Jim Cramer,” says Wooten. “‘I might not be right all the time but I can yell louder than this other guy.’”
Wooten and Smith looked at both professional pundits – celebrities with verified Twitter accounts – and amateurs claiming some sports expertise. Both were worse at predicting than the 50-50 odds of a coin toss. Professionals were right 47 percent of the time, a hair better than the 45 percent accuracy of amateurs. But the professionals were more confident, scoring a .480 confidence rating to the amateurs’ .313.
And confidence pays – far better than accuracy.
If a professional pundit accurately predicted every game of the baseball playoffs and series, Wooten and Smith estimated his or her Twitter following would increase 3.4 percent. An amateur would get 7.3 percent more followers.
But a professional whose confidence knows no bounds would increase his or her following by nearly 17 percent and an amateur would see a nearly 20 percent rise in followers.
The outlier of the field could be Nate Silver, the statistician and New York Times political blogger. He’s both cautious and accurate. But owing in large part to his correctly calling all 50 states in the recent presidential election, he’s popular.
By and large, say Smith and Wooten, pundits get a better audience through confidence and the excitement it generates.
“There is some psychological literature on the idea that people hate uncertainty,” says Smith. “The fact that people don’t like uncertainty would suggest that they don’t like the idea of a Nate Silver sort of person standing up there and saying, ‘I’m only 90 percent sure.’”
“I like to think of it like a roulette wheel,” says Wooten.” If you have somebody just placing bets, that person is kind of boring. But if you have someone going, ‘Oh, yeah! It’s red!’ and they are confident, that’s the person that you are interested in.”
Smith and Wooten outlined their findings earlier this year at the 50th Annual Meeting of the Academy of Economics and Finance. They plan to publish a paper on their method in an open-source journal, helping researchers, business people and others pose all sorts of questions to the vast amounts of data on the Internet.