Leading up to last year’s presidential election, many national pundits were predicting a dead heat, a dramatic ending to one of the most polarizing races in recent history.
The result – incumbent President Barack Obama beating challenger Mitt Romney by a 126-electoral vote margin – did not live up to the hype.
University of Virginia political science professor Larry J. Sabato and Princeton neuroscientist Sam Wang attempted to answer why the pundits missed the mark at a forum in Nau Hall Thursday night. The problem, they said, is the emphasis pundits put on narrative – a tossup is far more interesting to write about than a comfortable victory by an incumbent.
But the proliferation of data and poll aggregation could change the way people look at elections, they said.
“I’ll bet you in 2016, the coverage will depend more on data-based facts, instead of pundit predictions because the pundits were pretty embarrassed,” Sabato said.
Sabato and Wang are two respected analysts from completely different backgrounds. Both have amassed online followings by analyzing polling data and making predictions, but that’s where their similarities end.
Sabato is a political scientist who has predicted presidential, House and Senate races with 98 percent accuracy in the last four election cycles. His blog, Sabato’s Crystal Ball, combines hard data with punditry and predictions.
Wang’s path to becoming a political analyst was a bit more unusual. He’s a neuroscientist by trade and an amateur political junkie who became annoyed with the way media outlets and pundits make their predictions. News outlets don’t do a good job of analyzing the data in a way that separates good from bad information, he said, and pundits rely too much on gut feelings. He recalls getting into an argumentative email exchange with one pundit who said Mitt Romney had “momentum” going into Election Day.
“To put on the physics hat for a second, my idea about momentum is moving in one direction and continuing to move in that direction,” Wang said. “He thought I was a real bean-counter for being so literal in my definition of ‘momentum.’”
Instead, Wang relies on aggregated polling results – not looking at one or two of the most reputable, but looking at them all, including partisan polls, and drawing conclusions from aggregated data. One or two polls may be skewed, but the aggregation usually gives a fairly accurate picture of what’s happening on the ground, he said. It’s a practice he drew from his work.
“When we’re in the lab and we’re looking at data, we aren’t in the habit of looking at one data point at a time,” he said. “We want to look at all the data at once.”
It has worked well for Wang – he leads the Princeton Election Consortium, which also blogs its predictions and analysis online. The consortium correctly predicted 10 of the closest Senate races in the last election.
Wang and Sabato said they are worried about the prospect of poll manipulation, though. Political parties are beginning to see the power of polling, and they’re paying more and more money to commission private polls favorable to them. Aggregating the results may mitigate the impact, but manipulation can distort the averages.
Wang predicts the “arms race” between data manipulators and analysts could escalate in coming elections. His approach – and other analysts’ approach – may have to be refined in the future.
“I think nerds are going to have to be vigilant, and that will come to a question of data curation,” he said.
Sabato, who also writes about the historical context of elections, said he is noticing a few general trends. First, he said, presidential elections are becoming easier to predict because the country is more polarized – people vote more along party lines. Most moderates, he said, lean strongly in one direction or another.
He believes Republicans will gain about a half-dozen seats in the 2014 House elections. Democrats may get more of the popular vote overall, as they did in 2012, he said, but gerrymandering in the 2010 redistricting process will hold them back in the next election cycle.
“I think the chances of Democrats taking over the house in 2014 are next to nil,” Sabato said.