Every day seems to bring more bad polling news for President Trump. From internal Democratic polls showing Joe Biden in a strong position in places such as Missouri to major national media polls with eye-popping results in places such as Texas, even the president’s allies have become more vocal about his need to change strategy in the face of overwhelmingly negative data.
Tough numbers for Trump are not entirely new, though they represent a sizable slide from his position pre-coronavirus. Despite poll after poll showing Trump trailing by enormous margins in critical states, the public still seems reluctant to predict that Biden will be the winner. Trump voters, in particular, are likely to think their candidate will again beat the odds, shock the world, and rally an unpolled coalition to victory in November.
After the surprise of 2016’s Trump victory, it is understandable to be skeptical of the growing consensus that Trump is cruising toward a massive defeat. There are two key components to this skepticism: the belief that it is possible to recover in time for November and the belief that the polls may be continuing to underrepresent Trump voters.
Let’s tackle the first component and take a look at how the polling averages stack up over time. How bad is it, really? To put these numbers in context, consider the margins held at this point in time in previous presidential races. In 2008, at the same point in time, Barack Obama led John McCain by around 7 points in an election that was an absolute blowout for Democrats. At the start of July 2016, Hillary Clinton led Trump by an average of 4.5 points, according to the RealClearPolitics polling averages.
In today’s RealClearPolitics average, Trump trails Biden by around 9 points. If those numbers held, it would be an absolute wipeout for Republicans at the federal level. It is true that at one point in early 2016, Trump trailed Hillary Clinton by nearly 11 points. But once he was the presumptive nominee, he never trailed her by margins like what we are seeing versus Biden this year.
It is, of course, possible for Trump to close the gap, given that this is the year 2020 and fortune has taken nothing off the table. But Trump has shown no desire to change strategy, and COVID-19 shows no sign of abating.
All of which brings us to the belief that Trump is actually in better shape than advertised. This is not just a view held by many Trump supporters; it includes Democrats who felt burned in 2016, assured that their candidate was going to win.
Setting aside for the moment that the national-level polls were largely correct in 2016 (the final RealClearPolitics average had Clinton up by 3.2 points, her actual popular vote victory was 2.1 points), it is certainly true that some state-level polls portrayed a wildly incorrect picture of how the “blue wall” states would break. Trump outperformed his polls in Pennsylvania by three points, Michigan by about four points, Ohio by about five, and most egregiously in Wisconsin by seven. While many states had polling that was right on the money (Virginia and Florida come to mind), those Midwest states blew up the predictive models.
The American Association for Public Opinion Research investigated the causes of the polling error, and one key problem was polls in 2016 overrepresenting college-educated voters. Given Trump’s strength with non-college-educated voters, polls that didn’t include enough of them overstated the Clinton lead.
As a result, polls in 2020 should adjust their samples for education level. Many today do, such as the New York Times Upshot/Siena College polls that showed Biden with double-digit leads in Michigan, Wisconsin, and Pennsylvania.
But not all polls have made that adjustment, or they are still grappling with how to do so. In one notable recent example, polls conducted by the Atlanta Journal-Constitution showed bad news for Trump in the state of Georgia; it was only when they updated their methods to weight for education that the poll moved toward Trump and their poll release acknowledged that fact, noting that “the results of this poll may appear to be slightly more conservative than our previous polls.”
Additionally, pollsters who want to do the right thing and account for education level face a debate over what the makeup should look like. As the industry report notes, “Current Population Survey data and voter file analysis show a whiter, less-educated electorate than the exit polls. Thus, polls weighting to past exit poll parameters may have missed the mark in 2016.” This can make it harder for a layperson to know if a poll that says it is weighted by education is truly hitting the mark.
It is possible some polls are not capturing enough non-college-educated voters, and it is also possible that Trump could stage a stunning comeback (or Biden could have a dramatic collapse). Being skeptical is never a bad idea, especially in the face of bold, confident prediction. But healthy skepticism of polls can’t explain away double-digit deficits in critical swing states.
