Why the 2024 Polls are So Confusing Right Now
In just the last week, we have seen polls with Biden up big and down big against Trump
Last week was yet another ride on the PollerCoaster (this is my not-so-subtle pitch to check out my new Crooked Media podcast). In quick succession, a Bloomberg/Morning Consult poll found Biden losing to Trump in all six states by between 3 and 8 points. Then Quinnipiac University released a poll that showed Biden beating Trump 50-44 nationally. And then CNN rained on our very brief parade with a national poll that showed Biden losing to Trump 49-45.
This week, we also got polls that showed
And early this morning, NBC News released a new poll that showed Trump beating Biden 47-42 — a 3 point increase for Trump since November.
What the hell is going on? Why are the polls all over the map? And what, if anything, does it tell us about the outcome of the General Election?
1. Polling is an Educated Guess
There are two rules for riding the Pollercoaster. First, polls are not supposed to be predictive. They don’t tell us what WILL happen. They try to tell us what is happening right now — during the polling period. Second, every poll is based on an educated guess about who is going to vote. When the polls are wrong — like they were in 2016 and 2020 — it’s because pollsters were wrong about the makeup of the electorate.
In 2016, the New York Times's Nate Cohn gave the raw data from a poll of Floridians to four different well-respected pollsters and asked them to estimate the results of the poll using their methodology and models. And lo and behold, the four pollsters (and Cohn) got widely different results.
I would encourage everyone to read Nate’s write-up of the experiment because it’s one of the best explanations of the least understood and most opaque parts of polling.
The process of weighting the results to match one’s model of the likely electorate is rarely explained to the public, so it’s hard to understand exactly why the results differ, but that’s one main reason they do.
2. An Unpredictable and Unprecedented Election
Pollsters make predictions about the electorate based on what voters tell them, demographics and the turnout in recent elections. In hindsight, it’s not surprising that the polls were so off in 2020. There had never been a modern election conducted in the middle of a pandemic. There were huge disparities between how seriously the parties judged the risk of voting in-person and the voters’ willingness to use a raft of new voting options like vote by mail, drop boxes, and early voting. The 2024 election has a similar set of unprecedented variables. For the first time in the modern era, a former president is running to reclaim his job. There is a historic level of dissatisfaction with both candidates. A record number of voters are expressing interest in voting for a third-party candidate. Finally, one of the candidates is facing the prospect of being convicted of a crime and sentenced to prison before the election.
What does all of this mean? Will the U.S. continue the trend of increasing turnout? If there is a demobilization, which side will it affect? What will the large swath of voters who don’t like Biden or Trump do? Will they eventually pick Biden or Trump, vote third party, or just stay home?
All of this creates challenges for pollsters. In their defense, the pollsters were historically accurate in 2022, but this is a much more difficult task.
3. The Third Party Factor
There hasn’t been this much interest in third-party candidates since Ross Perot ran for president twice in the 1990s. A Quinnipiac poll from last summer found that 47% of voters would consider voting for a third-party candidate. In addition to Trump and Biden, Robert F. Kennedy Jr. and law professor and activist Cornel West are mounting Independent bids. Jill Stein is once again seeking the Green Party nomination. No Labels, a shadowy front group for the whims of rich people, is getting on the ballot in a number of states and contemplating running a bipartisan ticket.
Pollsters are dealing with this dynamic by testing a Biden v. Trump matchup and then another question that includes the third party candidates. This is all very confusing for a couple of reasons. First, every pollster is treating the No Labels candidacy differently. Some use possible candidates like West Virginia Senator Joe Manchin or former Republican Governor Larry Hogan. Others describe No Labels as a bipartisan organization. Secondly, we have no idea where these candidates will be on the ballot. National polls that include all five candidates are inherently imprecise because most if not all of these candidates are absent from the ballot in the majority of states. A poll might show that RFK Jr. or Manchin might draw enough support in Wisconsin to tip the election to Trump, but it won’t matter if they aren’t on the ballot. The best polls will be state polls from high quality pollsters that include the candidates qualified for the ballot in their state. We are a long way from pollsters being able to field those sorts of polls.
4. Young People and Independents
With all of the above caveats about the imprecision of polls this early in an unpredictable election, there are two specific subgroups that are bedeviling pollsters — Independents and young voters. Not coincidentally, these are the two groups that are hardest for the pollsters, campaigns, and the media to reach. When I see two polls with wildly different results, I immediately reference the cross-tabs to see how Independents and young people told the pollsters they planned to vote.
In the Quinnipiac poll that has Biden winning by 6, he is beating Trump among Independents by 12. In the CNN poll, Biden is trailing Trump by 4 with Independents. In the Bloomberg/Morning Consult swing state poll, Biden is down 8 with Independents. In the NBC poll, Trump is winning Independents by 19 points!
The youth vote is even more all over the map. John Della Volpe, a pollster who specializes in tracking young voters, pointed out on Twitter:
What It All Means
These days, partisans gravitate to the polls that support their chosen narrative. Trump “truths” the polls that show him winning. Haley supporters and Biden adherents tout the ones that show Biden winning. I love seeing polls that show Biden up. However, it’s best not to get too worked up over an individual poll, and instead look at the overall average and broader trends. That advice, of course, is hard to adhere to in an election with so much at stake.
The broader trends of the recent polling tell me two things. First, Biden, at minimum, has stabilized, and things may be looking up as people’s views on the economy improve. Second, the polls are consistent with what we have long assumed — this is a very close and winnable race that will come down to less than 100,000 votes across a small handful of states. Our task is the same if the polls show Biden up by 2 or down by 5. We have to do the work of persuading voters, which is why you should immediately go to Vote Save America and sign up to do something other than worry about polls.