What Have Pollsters Changed Since 2016?
In the aftermath of 2016, a majority of Americans have at least some doubts about the accuracy of polls, though it’s worth noting that the presidential polls were not that off four years ago, historically speaking. Still, in 2016 some pollsters failed to account for factors that ended up being crucial, like the importance of a voter’s level of education in predicting their political preferences. So we asked a number of well-known pollsters about what they had changed since 2016 as they sought to get a better read on the electorate’s intentions.
Perhaps most importantly, close to half of the 15 pollsters we talked to told us they now weight their samples by education. This adjustment could help deal with a real problem the polling industry had in 2016, when surveys tended to underrepresent voters with little or no college education. This was especially true among white voters, and so some pollsters such as Ipsos and the Pew Research Center have gone even further to weight by education attainment within racial groups, too.
Some pollsters have also tried to make sure they have a more representative sample based on where people live, as more heavily populated areas tend to be more Democratic. Marist College has tried to account more for whether people live in a metropolitan area while NBC News/Wall Street Journal is now more exacting about the share of its sample that lives in urban, suburban and rural areas.
Lastly, pollsters are also trying out new ways of contacting people, partly because of what happened in 2016 but also because of the increasing costs of high-quality polling. Some have moved to sampling from voter registration lists instead of random-digit dialing, which can help ensure you’re talking to someone who really might vote. (Those lists can also provide additional details on respondents, such as their party registration.) Some firms have also increased the share of respondents they contact by cell phone, as 96 percent of Americans report owning one. And others are also trying new ways of reaching respondents, such as texting them questions.
We’ll have to wait to see how accurate polling is this year, but just remember: Polls are our best tool for measuring public opinion, but they’ve always had a margin of error.