In 2016, then-candidate Donald Trump’s ascension to the White House surprised many Americans. Many people were positively sure that Hillary Clinton would win the election and one of the major reasons was polling.
Polls showed Clinton up by comfortable margins in many states and showed her handily winning the election. Although she secured the popular vote, she failed to win the electoral college, leading many pollsters to re-analyze how they do their work.
This year, polls were off once again and in some cases by even worse margins than the last presidential election. So what happened? Charles Franklin, director of the Marquette Law School Poll, says that while some polling was better this time, some polling was much worse.
"Last time we were off by 7 [points], this time we were off by 3 or 4 [points], so I think we did a little bit better this time. But the polling industry as a whole seems to have had bigger errors this year than four years ago and that’s very disturbing," says Franklin.
It's unclear why these polls would be so inaccurate this year and to some extent, Franklin says these errors may be unknowable. But these inaccuracies may point to an issue that would be hard to combat: a failure to capture an accurate picture of voting trends because of partisan differences in who chooses to participate in polls.
"It raises huge challenges, and how do we address that? I want to be clear that this is something that pollsters worry about constantly and are constantly trying to check for," says Franklin.
Although the Marquette Law School Poll has had consistent participation from Republican voters, he says there could still be specific kinds of Republicans who don't respond and that specifically Trump supporters may be less likely to respond.
"Those are hard things to know for sure. They are things we look for. But if that turns out to be part of the cause of the errors this year, I think that will be a big challenge for how to do polling that reaches those folks," he says.