WASHINGTON (UPI) — Polling errors, including miscalculating the young adult vote and privacy regulations that limit robocalling, led to erroneous poll results to show Democratic presidential hopeful Hillary Clinton with a big lead over Bernie Sanders in the days up to Michigan’s primary, pollsters say.
Sanders took more than 50 percent in the vote in Michigan on Tuesday to Clinton’s 48 percent, far from his projected loss by 15 to 20 percentage points. Similarly, polls showing Clinton taking an easy victory in Iowa in February were wrong.
The Michigan upset has left pollsters scrambling for answers, but the reality is it is impossible to pin it down to just one factor, Harry Enten, a senior analyst at FiveThirtyEight, said. FiveThirtyEight aggregates polls as a basis for its election predictions.
Enten told NPR three main factors came into play for the missteps in calculating Sanders’ popularity among voters: Young voters, black voters and Donald Trump.
Enten said pollsters miscalculated how younger voters and black voters would respond in the voting booths. He said it’s difficult to predict the path young voters will take because they are less likely to answer pollster questions. At the same time, many pollsters don’t call cell phones because strict privacy regulations require hand-dialing cell phone numbers — but not landlines — something that is costly and time consuming. While black voters in Michigan supported Clinton, they supported her in smaller numbers than anticipated by pre-election polls, Enten said.
Enten said the foremost reason why the polls were off by so much is the assumption Clinton would sweep the primary so voters turned their attention to disarming Trump.
“It’s an open primary,” Enten said of Michigan. “We can go vote on the Republican side, and maybe we’ll vote to stop Trump by voting for John Kasich, or maybe we think Donald Trump is such a weak general election candidate, as a Democrat, we want him to face Hillary Clinton, so we’re going to go vote in a Republican primary for Donald Trump, so that’s one reason. ”
Compounding the problem was the huge voter turnout in Michigan — it broke a 1972 record for the state. With some 1.2 million people turning out to vote Tuesday and a 2008 change in the way Michigan’s delegates are awarded, it was difficult for pollsters to predict the outcome, especially when past primary results in the state play a key factor in the polling outcome.
Another lesson learned was in Iowa, where Clinton — predicted to be the winner by a landslide — won 49.9 percent of the vote to Sanders’ 49.6 percent. The polling problem here was pollsters didn’t follow through.
Tom Jensen, director of the Democratic firm Public Policy Polling, said firms must continue intensive polling after the last debate and through the night before an election, which might be cost prohibitive. Late-day polling may better reflect voters’ decisions and allow pollsters to look for emerging trends.
“Polling’s never been more expensive so I don’t know how many organizations really have the resources to do more than they already are,” he said.
As for the future of polling for the upcoming primaries, John Della Volpe said there will likely be more faulty polls to come. Volpe is the director of polling at the Institute of Politics at Harvard University.
“There aren’t a lot of economic incentives to get it right,” he said. “A TV station is going to hire a local pollster, and they’re going to poll an election once or twice, and they’re not going to get paid a lot of money. If they’re right, they’re going to get paid, and if they’re wrong, they’re going to get paid.”