Nate Silver-Led Statistics Men Crush Pundits in ElectionJonathan D. Salant and Laura Curtis
Nate Silver was right. The Gallup Poll was wrong.
Silver, the computer expert who gave Obama a 90 percent chance of winning re-election, predicted on his blog, FiveThirtyEight (for the number of seats in the Electoral College), that the president would receive 51 percent of the popular vote as he called each of the 50 states, including all nine battlegrounds.
“Nate Silver, right,” said Bill Burton, who moved from the White House to the pro-Obama super-political action committee Priorities USA Action.
Gallup’s daily national tracking poll put Republican nominee Mitt Romney ahead by five points until it was suspended for Hurricane Sandy, and a final national survey released Nov. 5 gave the Republican a one-point advantage.
“These polls are designed only to measure what is happening at the time of that poll in terms of the national popular vote” and are not “designed to be predictive,” Gallup editor-in-chief Frank Newport said.
With the count in Florida still to be finished, Obama was leading Romney nationwide by two percentage points, 50 percent to 48 percent, the Associated Press reported, and won a decisive Electoral College victory.
Two university-based pollsters joined Silver in correctly predicting Obama’s win, and one of them will be dead-on about the electoral vote tally.
Drew Linzer, an assistant professor of political science at Emory University in Atlanta and a former pollster based in California, predicted yesterday morning on the website votamatic.org that Obama would end the race with 332 electoral votes and Romney 206.
Of Silver, Linzer wrote in that post, “his most likely outcome is still Obama 332, followed by 303 and 347, just like me.” Linzer also wrote that his model for votamatic.org had been predicting since June the Obama win with 332 electoral votes.
Sam Wang, a Princeton University professor of neuroscience, posted his final prediction -- that Obama would likely receive 303 electoral votes to Romney’s 235 -- on the school’s election blog at 2 p.m. yesterday. He reduced Obama’s total from 332 based on late polls yesterday.
Florida, with its 29 electoral votes, hasn’t been called by the Associated Press. Its outcome will determine which of those professors’ final forecasts was accurate.
The Republican-leaning Rasmussen Reports poll also had Romney winning the popular vote by one point. It missed on six of its nine swing-state polls. Rasmussen is an automated poll, meaning that it cannot call mobile phones and relies instead on an online polling tool to reach those without landlines. Rasmussen also adjusts data to reflect political party identification, which other pollsters say can change from survey to survey.
Rasmussen Reports had Obama winning Nevada and New Hampshire, tying Romney in Ohio and Wisconsin, and losing in the other five, including North Carolina.
“Nationally, we projected a toss-up and that’s what happened. We projected Ohio would be a tie, and it was very close,” president Scott Rasmussen said. “I believe that what happened is that the polls were right.”
Among other national polls, Pew Research Center and ABC News/Washington Post put Obama up by three points, while the NBC News/Wall Street Journal survey had him ahead by one point.
Silver infuriated conservatives with his model, which uses a number of measurements and calculations, including attention to state polls.
Regardless of the national surveys, Obama maintained a steady lead in most of the swing states in the last month, most notably Ohio, without which a Republican has never won the presidency.
U.S. elections are won in the Electoral College, where each state receives votes equaling the total of their two U.S. senators and their number of House representatives. The aggregate data from Real Clear Politics and Pollster.com also showed an Obama advantage in all of the swing states except North Carolina.
“Unless America abandons the Electoral College, the national polls just aren’t meaningful, although we all love the horse race,” said Rogan Kersh, provost at Wake Forest University in Winston-Salem, North Carolina.
Silver had Obama losing North Carolina, which the president won four years ago, and winning the other eight swing states. Silver also predicted Romney’s win in Indiana, the only other state that Obama won in 2008 and lost in 2012.
The NBC/Journal/Marist College poll called seven of the nine swing states, missing North Carolina, where it had Obama in a state that he lost, and Colorado, which it called a tie. The CBS/Times/Quinnipiac University survey predicted four of the five swing states it polled in, missing only in Colorado when a mid-October poll had Romney up by one point. CNN successfully predicted the winner in Colorado, Ohio and Nevada, though it had Romney winning Florida.
Quinnipiac had Obama ahead in Pennsylvania, as did Rasmussen. Romney made an unsuccessful last-ditch effort to compete in the commonwealth, which hasn’t backed a Republican presidential nominee since 1988. A poll by the Morning Call of Allentown and Muhlenberg College also had Obama leading.
Other automated polls correctly predicted most of the swing state results. The Democratic-leaning Public Policy Polling had Obama winning eight swing states and Pennsylvania, and tied in North Carolina. Another automated poll, SurveyUSA, had Obama winning Colorado, Nevada and Ohio, losing North Carolina and tied in Florida. Online poll YouGov didn’t poll in North Carolina, predicted a tie in Florida and had Obama ahead in the other seven states.
To continue reading this article you must be a Bloomberg Professional Service Subscriber.
If you believe that you may have received this message in error please let us know.