As co-editors of the special theme on the 2024 Election for Harvard Data Science Review, we decided that it was important to incorporate both an academic and industry look at multiple aspects of the 2024 presidential election. From a high-level discussion of different statistical and theoretical models by one of the most popular and well-known forecasters, Decision Desk HQ (Donnini et al., 2024), to a qualitative method by one of the most well-known future predictors, Allan Lichtman (2024), we explore both ends of the prediction spectrum in this special theme. A leading polling expert takes a look at what possible failures (i.e. uncertainty) could lead to their model being incorrect for the 2024 election (a venerable and admirable forward-thinking approach rarely seen by those that are otherwise so ‘certain’ in their predictions; Gelman et al., 2024). We also speak with a secretary of state that is using innovative methods to ensure election integrity (Simon et al., 2024). And finally we take a look at whether two of the main predictors for who will win an election can ever truly be right; if the voter turnout models that pollsters and modelers rely so greatly on even do a good job of predicting voter turnout (Ansolabehere et al., 2024) and whether or not polling itself can actually capture the true voting patterns of the American public anymore (Bailey, 2024a). We hope this theme will further illuminate some of the pitfalls and possibilities that our political system holds and help readers see the differences in predictions amongst some of the most capable minds in election polling and modeling.

Given the success (or failure) of the predictions from the 2020 Election special theme, we decided to give the analysis another try, but it is worth noting if anything looks familiar that some parts of our previous editorial were so good that we don't mind repeating them! To see our predictions from the 2020 presidential election, as well as those from some authors also featured in this 2024 special theme, please see take a gander at our previous editorial (Vittert et al., 2020).

Our co-editor meeting finalizing this special issue back in June 2024 took place the morning after the debate between President Biden and former President Trump. And just as this special theme will show a diverse range of predictions, we as co-editors had a plethora of different opinions. Ansolabehere thought Biden’s stubbornness would keep him in the race, still had Biden’s chances as 60/40 in favor of a second term because Biden’s liabilities were already baked into the polls, and believed that the debate didn’t show the American people anything they didn’t already know about Biden. Enos was in the middle between us co-editors and shared Vittert’s wide-eyed ‘shock’ at Biden’s performance. While he also thought Biden would stay in the running, Enos felt that the debate had flipped the odds, and that the election was now in Trump’s favor to win. Vittert took the other end of the spectrum as she felt that the debate was too shocking and that Biden was going to be bounced out, but that Trump would probably still prevail due to what would be infighting by the Democrats for whoever would steal away the nomination. While all three of us were wrong to some degree, we all had a few things right, including correctly predicting a J.D. Vance vice presidential pick by President Trump and correctly guessing that it wouldn’t really help him much… We will get to our personal predictions now that we actually know who is running—but first, let’s take a look at the pieces we have carefully curated for you that we think give a truly diverse overview of this election.

Before a discussion on what the pollsters are predicting, we felt the need to investigate two of the most important variables that will affect the prediction of the election.

Let’s start with arguably the most important variable: polling. Man, we have seen polls that are all over the place this election cycle. And many times there is a perfectly good explanation: they are hooey-filled polls. Campaigns (and entities that support those campaigns) will release their ‘own polls’ that regularly get picked up by the media that can show something wildly different than the more independent groups polls. It can be very difficult for the public to differentiate polls that have intentional bias in them, for example, a campaign trying to show their candidate winning when they definitely aren’t (everyone wants to vote for the winner) or losing when they definitely aren’t (sympathy donations). We have all heard the numerous mumblings about how pollsters were so wrong in 2016 and 2020, with the inevitable question of why should we trust them this time? Michael Bailey (2024a), a professor of American government at Georgetown University, takes a satirical spin on helping us decide if we should really listen to any polls at all (for an academic look at these same principles, please see Bailey’s previous article in HDSR [Bailey, 2023] or buy his book on the topic [Bailey, 2024b]).

As of writing this, Bailey says it’s so close that picking one side or the other for a prediction is a bit crazy. At least in public.1

Now onto the second most important variable (or maybe the most in some cases): voter turnout. Campaigns can spend all the money in the world (and this one is shaping up to be the most expensive election season in history [Bryner & Glavin, 2024]), can have the best message, and frankly can be a demonstrably better candidate, but if your voters don’t turn up to vote, you might as well have lit your money on fire. At this point in the race, candidates tend to not be aiming their ad dollars at undecided voters, but rather at getting their decided voters out to the ballot boxes on election day. The level of voter turnout that occurs is really what decides the election. (Many arguments have been made that if every American who is eligible to vote actually voted, then Democrats would always win by a landslide.) Voter turnout is also one of the main variables used to help predict who is going to win. Stephen Ansolabehere, Jacob Brown, Kabir Khanna, Connor Phillips, and Charles Stewart III (2024) take a look at whether the leading voter turnout models have any meat on their bones in this special theme.

As of writing this, Jacob Brown, the corresponding author of Ansolabehere et al. (2024), predicts “that the swing state polling averages are within the margin of error, so this race is a toss up. I’ll bet on whichever candidate I can get better than a 1-1 payout on.”

Having looked at two of the most important issues in prediction, we can now get to the larger picture: Who will win the 2024 U.S. presidential election?

Decision Desk HQ is the leading provider of real-time election results, race calls, forecasting, and notably the only provider to cover U.S. elections from the presidency and Congress to the county and city level. DDHQ was the first news outlet to call the election for Donald Trump in 2016 and the first to call the election for President Biden in 2020.2 In the HDSR special theme on the 2020 Election, DDHQ detailed their model for forecasting both presidential and congressional elections (Ram et al., 2020), and they have continued innovating in this space, expanding to the creating of a Live Primary Model that gives real-time, primary election night analysis in Donnini et al. (2024). As primaries become arguably more and more important in this country (both the Democratic and Republican parties in many states are choosing to close their primaries so that independents can’t vote in them, potentially changing the type of candidate chosen in the primary), having clear analysis on elections that really weren’t paid that much attention to previously is integral to our political system.

As of writing this, DDHQ predicts “that neither candidate should be surprised if they win or lose.”

From yet another perspective, a more qualitative approach than quantitative, we turn to Professor Allan Lichtman of American University. Lichtman has correctly predicted nearly every presidential election result of the U.S. presidential election since 1984, using his own “Keys to the White House” historical-based index system (Lichtman, 2024). The 13 Keys form a true/false criteria based on historical correlations with presidential elections from 1860–1980, using statistical methods adapted from the work of geophysicist Vladimir Keilis-Borok for prediction earthquakes (Keilis-Borok & Lichtman, 1981). While Lichtman himself states that the Keys do not perform a regression equation or use “horse-race polls” or “presidential approval ratings” (Lichtman, 2020), the statistician among your co-editors would disagree. While there are no Excel spreadsheets being loaded into R, we perform regression all the time in our everyday lives; we measure how variables move in relation to each other, from everyday decisions to the White House. In effect, this is what Lichtman’s Keys are doing—with a majority of negative Keys lined up against Donald Trump.

As of writing this, Lichtman is forecasting a Democratic win with Kamala Harris becoming the first female president.

From an entirely different perspective, we approach an industry-academic collaboration of The Economist with Andrew Gelman, Ben Goodrich, and Geonhee Han (2024). In the 2020 special election theme for HDSR, Andrew Gelman and co-authors Merlin Heidemanns and G. Elliott Morris (Heidemanns et al., 2020) discussed the model they had built in conjunction with The Economist that used fundamental predictors (U.S. economic growth factors, presidential approval, polls, etc.) but also spent a considerable amount of time to improve specific features within the model, such as state-level trends and nonresponse bias—aspects that many felt the 2016 pollster predictions did not take into account. In Gelman et al.’s (2024) piece for the 2024 Election theme, the authors take a hard look at both widely acknowledged and, more interestingly, the most unacknowledged sources of uncertainty in The Economist’s state-by-state election forecast. These range from third-party issues to replacement of a candidate (which did happen!), to the general concept of polling errors that arose in 2016 and 2020, as well as concerns specific to this 2024 election (the first presidential election where a major-party candidate has been convicted of a felony, for example). Their hard look at the ‘things that can go wrong’ in modeling is a testament to their dedication to getting it right.

As of writing this, Gelman is predicting “whatever The Economist forecast says.”

Finally, we turn to a topic that might end up being front and center in this election (as it was in 2020): election integrity. Our theme co-editor Liberty Vittert and founding editor-in-chief of HDSR, Xiao-Li Meng, sat down for a conversation with Secretary of State Steve Simon of Minnesota to discuss issues relating to this topic (Simon et al., 2024).3 After the 2020 U.S. presidential election, there were widespread claims that the election was unfair, insecure, and in some instances, outright stolen. This outcry put election officials under intense scrutiny, particularly those in the secretary of state offices across the country. We delved into these topics and more to understand how data science is helping to ensure election integrity, but also the risks posed by AI now and in the future to our election system.

And with that, we bring you to our own personal thoughts on the 2024 presidential election outcome. As statisticians and political scientists, we have been both watching the polls right along with you all and have had the benefit as well as strained eyes of many nights reading over pollsters’ models. For what it is worth, and to have in all perpetuity, we have decided to give you our (off-the-cuff, never-said-it-if-it-didn’t-turn-out-the-way-we-think) opinions. (The internet isn’t forever, of course…)

Co-editor Ryan Enos, a professor in the Department of Government at Harvard University, feels that this election is a really close coin flip, weighted in one direction, but that really it all depends on who you think is currently benefiting from the polling error.

As of writing this, Enos’s ‘seat of the pants’ prediction is a Democratic win.

Co-editor Stephen Ansolabehere, also a professor in the Department of Government at Harvard University, is pleading the fifth (sort of). He thinks that this election will end up in litigation because it is so close, or in other words, that the coin in the coin toss will land on its edge.

As of writing this, Ansolabehere’s ‘cuff of his shirt’ prediction is pleading the fifth.

Our final editor, Professor Liberty Vittert of Olin Business School at Washington University in St. Louis, agrees with Enos’s analysis, but simply doesn’t think that pollsters have changed their ways enough, nor that they have accounted for the nonresponse bias (shy voters for Trump), hence the polling error is undercounting Trump as it has in the past two presidential elections.

As of writing this, Vittert’s ‘collar of her jacket’ prediction is that Trump will have one of the greatest comeback stories in presidential election history and will clinch a win.

We hope you all enjoy reading the nooks and crannies of this special theme as much as we do, and we will keep you updated on social media (@libertyvittert and @ryandenos on X) with the changing of our pollsters—and our predictions—in the coming days leading up to the election.

Disclosure Statement

There are no financial disclosures by any of the co-editors.

Stephen Ansolabehere, co-editor on this special theme, is an author on one of the published articles.

Liberty Vittert, co-editor on this special theme, is the senior data scientist for Decision Desk HQ, which has a published article in this theme. There is no financial compensation for this position.

References

Ansolabehere, S., Brown, J., Khanna, K., Phillips, C., & Stewart III, C. (2024). Forecasting turnout. Harvard Data Science Review, 6(4). https://doi.org/10.1162/99608f92.62881547

Bailey, M. A. (2023). A new paradigm for polling. Harvard Data Science Review, 5(3). https://doi.org/10.1162/99608f92.9898eede

Bailey, M. A. (2024a). Murder at polling manor. Harvard Data Science Review, 6(4). https://doi.org/10.1162/99608f92.19b0a307

Bailey, M. A. (2024b). Polling at a crossroads: Rethinking modern survey research. Cambridge University Press.

Bryner, S., & Glavin, B. (2024, October 8). Total 2024 election spending projected to exceed previous record. OpenSecrets. https://www.opensecrets.org/news/2024/10/total-2024-election-spending-projected-to-exceed-previous-record/

Donnini, Z., Louit, S., Wilcox, S., Ram, M., McCaul, P., Frank, A., Rigby, M., Gowins, M., & Tranter, S. (2024). Election night forecasting with DDHQ: A real-time predictive framework. Harvard Data Science Review, 6(4). https://doi.org/10.1162/99608f92.ccb395f0

Gelman, A., Goodrich, B., & Han, G. (2024). Grappling with uncertainty in forecasting the 2024 U.S. presidential election. Harvard Data Science Review, 6(4). https://doi.org/10.1162/99608f92.a919e3fa

Heidemanns, M., Gelman, A., & Morris, G. E. (2020). An updated dynamic Bayesian forecasting model for the US presidential election. Harvard Data Science Review, 2(4). https://doi.org/10.1162/99608f92.fc62f1e1

Keilis-Borok, V. I., & Lichtman, A. J. (1981). Pattern recognition applied to presidential elections in the United States, 1860–1980: The role of integral social, economic, and political traits. Proceedings of the National Academy of Sciences, 78(11), 7230–7234. https://doi.org/10.1073/pnas.78.11.723

Lichtman, A. (2020). The Keys to the White House: Forecast for 2020. Harvard Data Science Review, 2(4). https://doi.org/10.1162/99608f92.baaa8f68

Lichtman, A. (2024). The Keys to the White House: Predicting the 2024 winner. Harvard Data Science Review, 6(4). https://doi.org/10.1162/99608f92.a390dc78

Ram, M., Shor, M., Williams, K., Jarugula, S., DeRemigi, D., Alduncin, A., & Tranter, S. (2020). Forecasting the 2020 U.S. elections with Decision Desk HQ: Methodology for modern American electoral dynamics. Harvard Data Science Review, 2(4). https://doi.org/10.1162/99608f92.9663befd

Simon, S., Meng, X.-L., & Vittert, L. (2024). AI and elections: A conversation With Secretary Steve Simon of Minnesota. Harvard Data Science Review, 6(4). https://doi.org/10.1162/99608f92.a10bcaeb

Vittert, L., Enos, R. D., & Ansolabehere, S. (2020). Predicting the 2020 presidential election. Harvard Data Science Review, 2(4). https://doi.org/10.1162/99608f92.fed3dc89

©2024 Liberty Vittert, Ryan D. Enos, and Stephen Ansolabehere. This editorial is licensed under a Creative Commons Attribution (CC BY 4.0) International license, except where otherwise indicated with respect to particular material included in the editorial.