Why Were The Michigan Primary Polls So Wrong? Look To Millennials' Phone Habits, A Wonky 2008
DETROIT (CBS Detroit) How did Bernie Sanders manage to pull off that upset victory over Hillary Clinton in the Michigan Primary?
Pollster Ed Sarpolus of Target Insight says the polls missed a number of independents who decided to vote at the last minute for Sanders.
"They're not a typical frequent voter so when you poll them the week before their intention is probably not to vote," Sarpolus said. "Bernie Sanders spent a lot of money at the end, he came into the state, a lot of young people, new people, at the last minute kids were like 'I'm going to go out and vote.'
"Pollsters were not able to catch that unless you were actually polling the night before or that specific morning."
Sarpolus says Sanders benefited from the biggest primary turnout since 1972. He says turnout was especially high among young people, most of whom voted for Sanders.
Sanders also benefited from stronger than expected support among African-American voters, which is typically a Clinton stronghold. Hillary Clinton didn't spend a lot of time in west Michigan, and that's where Sanders captured more African American voters, Sarpolus said.
Clinton won Wayne, Oakland and Macomb counties, and Sanders won just about every other Michigan county. "She didn't count on west Michigan," Sarpolus said, which offset what was happening in Detroit.
If the turnout had been more traditional in terms of age and location, Sarpolus said Clinton would have won Michigan.
"What happened was, we had a large turnout out in west Michigan, specifically Kent County. As a matter of fact, I was getting calls by 9:30 last night, I was looking at Kent County, it was such a large turnout. It was a whole new voter out there," Sarpolus said.
There's also talk out there that pollsters were relying on traditional models of poll research like phone surveys, which failed to account for the fact millennials don't have landlines. There's also the issue of Michigan's wonky primary election in 2008, which brought out only about half as many voters as came out in 2016.
"Trying to predict what that turnout would look like based on half as many voters in a weird election eight years ago means there was a good chance the polls would be off...Which, we now know, they were," the Washington Post wrote.
