A lot [of] things went wrong in 2020. And presidential polls were no exception. Joe Biden was supposed to win the 2020 presidential by eight points, according to the polls, which were wrong. He won by five points. He was supposed to win Wisconsin by 10 points. Instead, Biden eked out a victory there with less than 1 percent of the vote between him and incumbent President Donald Trump. The polls were very wrong in Wisconsin. The polls also had Biden winning Florida. And North Carolina. Here’s why the polls ended up missing the mark in 2020, and what’s being done about it.What's interesting is that polls weren't off significantly in 2018, when Donald Trump was not on the ballot. That makes me think a lot of the problem with polls is the result of the kinds of voters Trump inspires to vote for him. As CNBC noted, a lot of them do not want to talk to pollsters at all. FiveThirtyEight ascribed this to low levels of social trust: "Social isolation, or the lack of social integration, has long been thought to reduce willingness to participate in surveys. Americans who feel alienated or isolated from society do not feel compelled to participate in surveys out of a sense of civic obligation." These individuals probably also vote infrequently. If pollsters are looking for likely voters, they might miss or underweight low-trust voters who like Trump in their surveys because without Trump on the ballot, low-trust Americans aren't likely to vote. That means that they probably won't be messing up poll results in 2022, so people will be less likely to ask this question next year. That might not be the case in 2024 if Trump is on the ballot again, but a successful impeachment and conviction would prevent that, as it would make him ineligible to run for office again. One can only hope.
CNBC asked many of the same experts back in October Are Polls Trustworthy?
New polls are being released daily, answering how many Americans favor President Donald Trump or Vice President Joe Biden to win the 2020 presidential election. A majority of Americans don't believe these pre-vote polls are accurate, pointing to the misrepresentation of polling results from the 2016 presidential election. Experts contend the process of polling has become more precise and conclusive over time, but that they poll how people will vote, not how the electoral college will count votes. Find out how the polling industry makes money and how to understand polling results ahead of November 3rd by watching the video above.In addition to reinforcing the points about how contacting people and getting them to respond are getting more difficult, making polling less reliable in general, this video opened with how political polling is just one segment of market research. That should have occurred to me, but leave it to the cable channel watched by people who think they own the country to point it out to me. Another fact that surprised me, but probably shouldn't, was that political polarization is starting to spill over into questions that had no political dimension before Trump. That's causing problems for market research that the pollsters hadn't encountered before. Still, the answer to CNBC's question is that it's in the polling firms' interest that the answer is yes, so they're working on making polls more trustworthy. I hope they succeed so Americans can have more faith in polling and the outcomes of our elections.
*There was one — Kamala Harris in Vogue — but I wasn't seeing my sources cover that yet, either. Maybe later this week.
Political polling could only possibly be reliably accurate if it were able to get a random sample of the voters. It can't. It instead has to rely on convenience samples, the sets of voters who can be reached, and who then agree to answer the polling questions. Since the pollsters report that they get <10% of their target responders to actually respond, it does not require a very large differential bias in the difference between the preferences of convenience sample and those of the general population of voters to create potentially huge differences in the poll vs the actual results.
ReplyDeleteAnd that seems to be the problem, as the differential does not include as many Trump voters as the actual electorate.
DeleteThis comment has been removed by the author.
ReplyDeleteDeleting a duplicate of your comment above or posted something and decided it wasn't worth asking?
DeleteThet're not the only ones who don't talk to pollsters. We basically unplugged our home phone last sprint during the primaries and still haven't plugged it back in. We were getting 30-40 political calls a day.
ReplyDeleteI got so many spam calls during the pandemic that I stopped answering my phone unless I recognized the number. That means I didn't answer any calls from pollsters, even though I would have actually liked to answer those.
Delete