Donald Trump’s “MAGA” movement is arguably more unpopular than ever before, with only 24% of the electorate (and 12% of independents) viewing it favorably. But if you looked at the 2024 presidential polls for a Biden-Trump rematch, you probably wouldn’t be able to figure that out. Virtually every polling aggregation shows an extremely tight race, with Biden doing several points worse than he did in 2020.
Most observers have explained this by pointing to the fact that Biden is comparatively old and extremely unpopular. Only 41% of survey respondents approve of him according to the FiveThirtyEight aggregate, and he will be nearly 82 years old at the start of his potential second term. Any incumbent with these numbers is going to face some negative headwinds, and current polling simply reflects the public’s general dissatisfaction with Biden.
But the rush to explain and analyze the early numbers obscures an important reality: early head-to-head polling isn’t useful for predicting elections. The graph below visually represents how irrelevant polls taken today are: at this point in time, the average polling error is on the order of 8 percentage points.
The graph above suggests that presidential polling historically only becomes fairly predictive around June of the election year, which is when the primary season ends and the general election matchup is confirmed, allowing voters to better frame the issues at stake and settle on a choice.
Interestingly, the past few cycles suggest that voter attitudes are more or less set by this point in time; presidential polls from June have historically been roughly as predictive of the final result as those taken at the end of October. The post-June curves on the graph can generally be explained by polls shifting in response to events like debates, conventions, and other campaign-related headlines, but the data suggests that these short-term shifts are less signal than noise and often represent nothing more than fleeting bounces.
What does all this mean for how we should read today’s polls, then? We are approximately 17 months away from the upcoming presidential election. The average polling error at this point in time has been roughly 8 percent in margin over the last few cycles. That’s the difference between a Democratic and a Republican landslide in today’s hyper-polarized era. The variability suggested here is immense, and it renders head-to-head polling at this stage almost entirely useless from both a forecasting and statistical perspective. It should also be noted that even come election day, by when the debates, the issues, and the messages have been settled in the minds of voters, there is still a nearly three-percentage point average error.
Early presidential forecasts should instead place more weight on different fundamental factors like historical results, trends, candidate favorability, and economic indices. From this angle, both sides have reasons to feel bullish this early into the campaign. Democrats could argue that an improving economic outlook, Trump’s legal issues, and the electorate’s general rejection of Trump-aligned candidates last November will benefit Biden in 2024. Republicans, meanwhile, could point to the President’s unpopularity and advanced age, both of which are serious concerns for swing voters.
Neither Trump nor Biden is particularly strong at the moment, and both candidates have very real and significant vulnerabilities that expose them to the risk of underperforming. Anyone could make a decent case for either candidate winning or losing, and perhaps even by a comfortable margin.
None of these arguments, however, should be based around head-to-head polling. Because this far out, it means almost nothing.
I’m a software engineer and a computer scientist (UC Berkeley class of 2019 BA, class of 2020 MS) who has an interest in machine learning, politics, and electoral data. I’m a partner at Split Ticket, handle our Senate races, and make many kinds of electoral models.
My name is Harrison Lavelle and I am a political analyst studying political science and international studies at the College of New Jersey. As a co-founder and partner at Split Ticket, I coordinate our House coverage. I write about a variety of electoral topics and produce political maps. Besides elections, my hobbies include music, history, language, aviation, and fitness.
Contact me at @HWLavelleMaps or harrisonwlavelle1@gmail.com