Ashley Koning analyzes an unprecedented election, the voters and the polling industry
The polls correctly predicted that Joe Biden would become the 45th president of the United States but stopped short of getting the rest right – leaving the public and media asking what happened?
Ashley Koning, director of Rutgers University’s Eagleton Center for Public Interest Polling (ECPIP), a unit of the Eagleton Institute of Politics, says that while polling isn’t dead or useless, pre-election polls should not be considered the only tool in making accurate election predictions and discusses why it’s time for serious reflection on the industry and what we should expect in the future.
What is the significance of this election and Joe Biden’s win?
Everything about the 2020 U.S. presidential election was unprecedented. It took place amidst a pandemic, with voting procedures altered across the country because of it. And yet, there was record-high early voting and voting by mail. The candidates set an all-time record for voter turnout, in general, with both candidates beating President Obama’s record from 2008 and with President-Elect Biden now receiving the most votes for president of all time. President Trump, on the other hand, joins a small club of U.S. presidents who have only served one term, becoming the third to do so since World War II.
At the same time, we are in an era of hyper-partisanship and at an all-time high for political polarization, and this kind of political landscape will be difficult for the president-elect to navigate. Biden will have to take action on his promise of unity and find commonality between Democrats and Republicans. He may be able to do so on the coronavirus, which is the country’s most pressing issue right now, but even the pandemic has become a political issue.
What are the major issues that Americans want addressed first?
Exit polls currently show that voters, overall, were twice as likely to choose the economy over the coronavirus when asked about what issue factored most in their vote choice for president. Trump voters were especially likely to select the economy as their top issue as well as crime and safety. Biden voters were most likely to say racial inequality, the pandemic and health care. While the economy is certainly connected to the coronavirus, the pandemic has become highly politicized itself, with Republicans and Trump supporters much more in favor of putting the economy over containing COVID, while Democrats and Biden supporters feel just the opposite.
How did polling compare to the 2016 election?
Heading into 2020, pollsters attempted to resolve a number of problems with the polls in 2016. First among them was trying to better represent the non-college educated white vote that propelled President Trump to victory in 2016. Pollsters have now factored in education when using a statistical tool called “weighting,” which makes sure that a sample accurately reflects the population under study by balancing it on key demographics. Surveys tend to have too many college-educated respondents in them, and while education was not an indicator of vote choice in previous elections, non-college educated voters significantly departed from college-educated voters and heavily went for Trump over Clinton.
Another problem in 2016 was statewide polling, particularly in key battleground states. While national polls that year pretty accurately reflected the popular vote, statewide polls had a historic amount of error -- whether due to not being done frequently enough, not being done well or not being done close enough to Election Day with many voters still undecided up until the last minute. This time around, a number of highly credibly pollsters frequently polled battleground states right up until Election Day. Unfortunately, polls of some of the most crucial battleground states seem to be suffering the same fate as 2016 according to current vote counts, but just how off the polls were is something we will not know until all the votes in very state are counted.
As the American Association for Public Opinion Research states, "When all the votes are counted, some of the hundreds of polls conducted on this election will come close to the final vote percentages, but some will not. This has been the case in every election. The issue is how the polls collectively performed in describing the official results of the 2020 election."
Does voter turnout affect polls?
The unprecedented nature of this election in terms of who and how many turned out, as well as how they voted, may have contributed to what we saw in pre-election polls versus the vote counts thus far. Taking a sample of a population limits us in estimating support for smaller, yet crucial, subgroups, such as the Cuban-American vote in Miami-Dade County, which proved vital to Trump’s win in Florida. But until we have final vote counts in every state -- which will take some time -- we will not definitively know how the pre-election polls fared. We may see a slightly larger polling error than normal, but we simply can’t tell at this point. Pollsters' underestimation of two consecutive presidential elections -- an underestimation not seen in polling the 2018 midterms -- is particularly troublesome, however. And it may hint at a deeper issue that specifically stems from accurately representing President Trump’s supporters, whether because of coverage error problems, social desirability bias or nonresponse bias.
What does the future of polling hold?
This is something the polling community is still trying to figure out and on which it does not want to make a rushed judgment given that final vote counts are still in progress. A lot of the resolutions -- like more representative samples, better coverage or eliminating response bias -- are easier said than done. Polling takes a significant amount of time and money, and against a backdrop of continually evolving technologies and low response rates across the board, it’s becoming increasingly difficult to do. These problems are made even worse if certain types of respondents are purposely not answering surveys or not revealing their true attitudes more so than others and as polls themselves become increasingly politicized by both politicians, the press and the public.
Polling is a vital part of our democratic process, connecting the voice of the people to their policymakers and providing an assessment for some issues that may never get their day at the ballot box yet are nonetheless just as important. As survey researchers begin to investigate 2020 in the coming months, we all must have patience with the time it takes to find the answers and how these answers may affect the future of the industry. But above all, no matter who you are or what you believe, please pick up your phone when we call. We would love to hear from you.