It’s important to put the 2020 performance of election polls into perspective. The errors were consequential, to be sure, and very sizable compared with most election polls. But they were not particularly large in absolute terms – just a few percentage points, on average, on each candidate’s support.
It’s also crucial to understand that most polls are
not about elections. Instead, they aim to capture people’s views on the key issues of the day, the realities of their lives, such as their employment, lifestyle activities or health care, where they get their information, and other subjects.
This kind of polling is important because it provides an independent source of information about what Americans experience and what they believe – without which we’d be left with whatever pundits, parties and public officials claim the public is thinking.
Fortunately,
the news is better when it comes to this type of issue polling. We have a lot of evidence that polls that are weighted to match the population on important demographic categories do a good job.
And when it comes to issue polling, errors in accurately representing the share of Biden or Trump voters
make very little difference in the results. For example, small errors in candidate support are less meaningful when trying to understand how people feel about abortion, the budget deficit, foreign aid, climate change or other issues. Plus, just because someone supports a certain candidate doesn’t mean they agree with that candidate on every issue.
In sum, pollsters recognize the challenges they are facing and are working hard to ensure that their work continues to accurately represent the views of the public. One way this is happening is through the growth of different ways of doing polls, some of which may be less reliable than others. Experimentation is a good thing, but it makes the job of a poll consumer – you – much harder. Our last lesson will focus on how you can become a more astute reader of the polls.