Bulldog Bites

News and Views from the University of Redlands

Are election polls accurate?

U of R faculty members spoke about how the extent to which polling can predict election results during a recent virtual event sponsored by the Hall Network. (Photo courtesy of iStock.com/hermosawave)

That’s the question that Department of Political Science Professors Renée Van Vechten and Eric McLaughlin sought to answer at the virtual discussion, “Polling and the November Election: Are Polls Accurate?” held on Oct. 14 as part of the 2020 Hall Network Virtual Brown Bag series.  

Van Vechten, who specializes in California and legislative politics, and McLaughlin, whose research focuses on institutions in new democracies, led the discussion on how election polling has traditionally been viewed as a predictor of results—until the 2016 presidential election. In 2016, former U.S. Secretary of State Hillary Clinton was leading businessman Donald Trump on most polls, but ended up losing the election by a wide electoral-college margin.  

“You can never tell what's going to happen with 100% accuracy,” says Van Vechten. “In 2016, the gap between Trump and Clinton was within what we call the margin of error, which is about 3% for most polls—meaning that they were in a statistical dead heat,” says Van Vechten. “Clinton won the popular vote, as predicted,” she says, adding it’s important to remember that the national popular vote does not always translate to an electoral college victory.

With most polls currently indicating former Vice President Joe Biden ahead of President Donald Trump, the 2020 election is a different story, Van Vechten says. “Polls are showing that Biden has a lead outside the margin of error of most polls, so it would be very unusual if he actually loses the election—or at least the popular vote.” The electoral college outcome is tied to the results in each state, not the sentiments captured in national polls.

A poll’s accuracy largely depends on the representative nature of a sample. If large numbers of nonvoters or Trump voters are systematically underrepresented, for example, then that could skew the results of many polls, which would lead to false predictions.

The duo also led a discussion on the history of polling in the United States, and how these polls have been used to disseminate information. McLaughlin says people who view election polling as a barometer need to understand probability better. He adds that polling history shows that the way questions are posed color polls' outcomes.

“Polls always have random sampling error, but they also have bias, which we should try to correct for,” McLaughlin says. “It can be very easy to get lost in information.” 

The Virtual Brown Bag Series is sponsored by the Ken and Lynn Hall Network for Innovation and Policy, created in 2015 for advancing policy work at the University and with its partners. The first lecture in the 2020 series on Sept. 30 featured School of Education Professors Mikela Bjork, Brian Charest, and Nicol Howard, highlighting their ongoing work funded through a 2019 Hall Network Faculty Scholarly Seed Grant to develop a degree program in the School of Education with a focus on educational leadership and policy, and civic and community engagement.

In addition to the Brown Bag Series and the Faculty Scholarly Seed Grant, the Network offers student internships and faculty fellowships under the leadership of Professor Andrew Wall and Professor Greg Thorson, the Ken and Lynn Hall Endowed Chair in Public Policy.

The Virtual Brown Bag Series will continue at noon, Wed., Nov. 18, with Professors Annie Knox and Paul Jessup on “The history of education funding in California: Examining the stories of evolution.” The final event in the fall series is scheduled at noon, Wed., Dec. 9, with Wall, Thorson and Professor Adriana Alvarado on “Examining the Higher Education response to the COVID-19 pandemic.”

Learn more about programs at the University of Redlands