Recent headlines are full of the latest polls and election forecasts. “Biden leads Trump in Wisconsin,” declares CBS; “Biden advantage Pennsylvania, Michigan, Wisconsin,” says CNN; “Biden is favored to win the election,” according to FiveThirtyEight.com.
But can the polls be trusted?
After 2016, polls were criticized for underestimating Donald Trump’s chance of victory. They got some things right. Nationwide surveys correctly predicted Hillary Clinton would win the popular vote. Polls predicted she would win it by 3 percent, and she won it by 2 percent.
However, in many regions of the country, especially in battleground states in the Midwest, statewide polls persistently misread support for Trump. An analysis of what went wrong found state polls failed to factor in education, for example. People with more years of education were more likely to support Clinton and also more likely to participate in polls, leading to an overrepresentation of Clinton supporters in the samples.
What other factors contributed to polls miscalculating Trump’s chances? And what did pollsters learn from the mistakes of 2016?
Wednesday at 9 a.m., MPR News host Kerri Miller discussed how polls work, how to interpret them and whether they should even be used to predict an election outcome that can only be known after the ballots are counted.
Margie Omero, is a principal at GBAO Strategies and co-hosts The Pollsters podcast.
Seth Masket is director of the Center on American Politics at the University of Denver. He is also the author of the “Learning from Loss: The Democrats 2016-2020.”
Before you keep reading ...
MPR News is made by Members. Gifts from individuals fuel the programs that you and your neighbors rely on. Donate today to power news, analysis, and community conversations for all.