Abstract

This article revisits my predictive model, the Keys to the White House, which I presented to HDSR readers 4 years ago. In 2020, the model predicted that Joe Biden would defeat Donald Trump, primarily due to Trump’s failed response to the COVID-19 pandemic. This article demonstrates that the Keys model remains valid despite the turmoil of Trump’s felony convictions, Biden’s withdrawal from the election, and the nomination of a woman of color. It provides the rationale for predicting that Trump will fail to regain the White House in 2024 and Kamala Harris will become the next American president.

Keywords: prediction, keys, White House, election, Trump, Biden, Harris

1. The Keys Revisited

Four years ago, I introduced HDSR readers to the Keys to the White House, a historically based index system for predicting the results of American presidential elections that has been successful since 1984 (Lichtman, 2020). At that time, I predicted that President Donald Trump would lose the 2020 election. I noted, “The Keys predict that Donald Trump will become the first president since George H. W. Bush in 1992 to lose a reelection bid.” This contribution explains why the Keys have not changed despite the turmoil of Trump’s felony convictions, Biden’s withdrawal from the election, and the nomination of a woman of color. It explains the rationale behind predicting another Trump defeat.

2. The Keys Model

The Keys to the White House follow the premise that governing, not campaigning, primarily determines the outcomes of American presidential elections. The Keys comprise 13 true-false questions that gauge the strength and performance of the party holding the White House, with an answer of true favoring the incumbent’s reelection. If six or more of the keys are false, the incumbent party is a predicted loser; otherwise, they are a predicted winner. The Keys gauge the big picture of a president’s record, such as midterm election results, internal nomination contests, third-party challenges, the short- and long-term economy, policy change, social unrest, scandal, and foreign and military failures and successes. Only two keys relate to the presidential candidates. (See Table 1 for brief definitions of the keys.)

I developed the Keys in 1981 through collaboration with Vladimir Keilis-Borok, founder of the International Institute of Earthquake Prediction Theory and Mathematical Geophysics in Moscow. To develop our model, we reconceptualized presidential elections not as Ronald Reagan vs. Jimmy Carter, Republican vs. Democrat, or liberal vs. conservative, but in geophysical terms. Stability meant that the party holding the White House stayed in power, and the earthquake meant the party lost power. We began our study in 1860 when most African Americans were enslaved and women could not vote. There were no automobiles, airplanes, radios, or televisions, and America was still an agricultural nation dominated by White people of Northern and Western European descent. Thus, our analysis covered vast changes in American society, politics, demography, and economics.

To develop this model, we applied a simple pattern recognition algorithm known as the Hamming distance to two binary vectors. We coded elections from 1860 to 1980 where the incumbent party prevailed as Class I (0) and those where the challenger won as Class C (1). The second vector consisted of true or false answers to questions that are answerable before an upcoming election. The model phrases the indicators so that an answer of true favors stability and an answer of false favors earthquake. For example, Key 5 states, “The economy is not in recession during the election year.” Rather than randomly trolling through history, we followed my theory that presidential elections primarily reflect the electorate’s up-or-down vote on the strength and performance of the party in power. Thus, the Keys are grounded theoretically and empirically.

We computed the Hamming distance for each of 31 elections, defined as D = ΣXi, where X is coded for each key as 1, when turned false and against the party holding the White House, and coded 0 when turned true in favor of the incumbent party. For 13 keys, D, the Hamming distance, varies from 0 to 13, with higher numbers indicating a greater likelihood of an incumbent party defeat. Empirically, the threshold for defeat of the incumbent party is gauged as D = 6. A higher or lower threshold would have produced errors and, if used subsequently, would have invalidated several correct predictions. Each of the ultimately 13 keys added to the degree of distance between electoral stability and earthquake, as measured by mD(I) – mD(C), where mD(I) is the mean Hamming distance for elections classified as incumbent victories and mD(C) is the mean Hamming distance for elections classified as challenger victories.

These criteria resulted in the exclusion of about 15 initially proposed questions, including whether the economy is in a state of war or peace, whether the incumbent or challenging candidate is more centrist in policies, whether the incumbent party has held office for more than one term, whether the incumbent party gained more than 50% of votes cast in the previous election, and whether the incumbent party is Republican or Democratic. For additional elaboration on the method, see Gvishiani et al. (1980) and Keilis-Borok and Lichtman (1981). We initially identified 12 keys, but further analysis led to 13. All predictions are based on 13 keys.

3. A Unique Forecasting Model

The Keys system diverges from most predictive models by not relying on regression analysis but instead on an index model where each indicator or ‘key’ is equally weighted, thus avoiding the issue of fluctuating weights in future elections. Suggestions, for example, to weigh the two economic keys more heavily would have produced errors, such as predicting Hillary Clinton’s victory in 2016. In addition, research has shown that equal-weight forecasts are often more accurate than regression models with parameter weights across various data sets (Dana & Dawes, 2004; Graefe, 2015). If one or more keys are especially consequential in a particular election, the model self-corrects through trigger effects on other keys. For example, the fiasco in the Iraq War during President George W. Bush’s second term triggered Republican losses in the 2006 midterm elections and stymied policy innovation at home.

The Keys also incorporate judgmental variables alongside binary and quantitative indicators, despite potential criticism for perceived ‘subjectivity.’ The judgments required to turn keys, such as incumbent and challenger charisma, are not random but are tightly constrained by each key’s definition and the record of their turning since 1860. (Lichtman, 2024, devotes 30 pages to defining and turning the keys: 19–48; see also Tetlock, 2017.)

Critics frequently challenge Charisma/National Hero Keys 12 and 13 for their allegedly subjective application. However, as defined within the system, a candidate must have provided critical leadership in war to be considered a national hero, as exemplified by Ulysses S. Grant and Dwight D. Eisenhower. Candidates like George McGovern or John McCain, who performed heroically in war but played no leadership role, do not qualify. Similarly, a candidate only earns either charisma key by qualifying as a once-in-a-generation, across-the-board appealing candidate. Only a select few leaders have met these criteria. Among presidents since 1900, those tabbed as meeting the requirements of the charisma indicator include Theodore Roosevelt, Franklin D. Roosevelt, Dwight D. Eisenhower, John F. Kennedy, Ronald Reagan, and Barack Obama—all of whom won election to a second term, except for Kennedy, who died in office. In contrast, five of nine presidents since 1900 who lost Charisma Key 12 failed to win a second term: William Howard Taft, Herbert Hoover, Jimmy Carter, George H. W. Bush, and Donald Trump.

Despite much criticism of how I turned this key, Donald Trump does not qualify as a broadly inspirational candidate. Although a practiced showman, Trump appeals only to a narrow base. His presidential approval rating in the Gallup Poll averaged 41%, putting him at the bottom of all past presidents (Gallup, n.d.). In two elections, Trump lost the people’s vote by an average of 3% and a combined 10 million votes (Leip, n.d.). According to 538’s polling average for mid-October 2024, only 43% of Americans had a favorable opinion of Trump after his nearly a decade as a candidate and president (538, 2024).

The inclusion of judgmental indicators represented a purposeful trade-off between a limited degree of uncertainty and the greater benefits of including a much wider range of indicators than other forecasting models. In a special edition of Foresight: The Journal of Applied Forecasting devoted to the Keys to the White House, forecasting authorities Armstrong and Cuzan note that: “For most problems, regression analysis is limited in that only a few explanatory variables can be put into the model (perhaps three or four variables) because of limited data, measurement errors, and correlations among the explanatory variables (a problem called collinearity). Subjective indexes avoid these estimation problems. Given the many variables and the small amount of data, index methods would seem appropriate for forecasting presidential elections” (Armstrong & Cuzan 2006, p. 11).

The Keys have accurately forecasted the winners of presidential elections since 1984, including Donald Trump in 2016, with the possible exception of Al Gore in the contested election of 2000 (see Lichtman, 2003; Mebane, 2004). There is no specified time when the Keys fall sufficiently into place to call an election; they have correctly predicted outcomes years ahead of Election Day. The Keys predicted Ronald Reagan’s reelection in 1982, the 2008 defeat of the incumbent Republicans in 2006, and Barack Obama’s 2012 reelection in 2010. In subsequent years, however, the increasing polarization of the electorate has resulted in predictions during August and September of the election year (Lichtman, 2024).

4. An Unprecedented Election

Every presidential election is unique in some regard. A major party nominated an African American for the first time in 2008 and a woman in 2016. In 2016, for the first time, taped evidence documented a major party candidate bragging about sexually assaulting women. Voters elected a president in 2020 during the worst pandemic in U.S. history. However, no prior election matches the extraordinary circumstances of 2024. No major party nominee has ever competed for the presidency under a cloud of criminal indictments. Prosecutors indicted Trump in four jurisdictions, amounting to 91 felony counts. In the only trial completed to date, a Manhattan jury convicted Trump of 34 felony counts of falsifying business records to evade a campaign finance violation. Trump and Grover Cleveland (in 1888) were the only once-defeated presidents to compete for the Oval Office again. Among Democrats, after President Biden’s faltering performance in the Democratic debate, his party leaders forced him to exit the election in July and endorse Vice President Kamala Harris—unprecedented as a woman of color—for the vacated nomination. No president had ever withdrawn from reelection this deeply into the election cycle.

5. The Keys Prediction

Amidst this turmoil, the Keys remained constant, accounting for the late change in the Democratic ticket. Early in 2024, the incumbent Democrats had definitively lost two keys. They forfeited Mandate Key 1 because of U.S. House losses in 2022 and Incumbent Charisma Key 12. Biden’s withdrawal and the nomination of a younger candidate (age 59 at the time rather than 81) brought fresh enthusiasm to the Democratic campaign. Still, Harris had failed in her bid for the Democrat’s 2020 presidential nomination, was largely invisible as a vice president, and has only recently become the 2024 nominee. Thus, she fails to secure Key 12.

Trump’s legal troubles did not impact this tally of two negative keys. As in 2016, he did not meet the criteria for turning Challenger Charisma Key 13 against the incumbent party. Biden’s withdrawal cost his party Incumbency Key 3. However, by uniting Democrats around Vice President Harris’s candidacy, he avoided the loss of Contest Key 2, leaving Democrats with a three-key rather than a four-key deficit. In addition, the Democrats lost Foreign/Military Failure Key 10 because of the Middle East humanitarian catastrophe with no good end in sight. The final tally is that the Democrats lost four and secured nine keys. The model would predict the defeat of Harris’s incumbent Democrats only if they held seven or fewer keys, that is, a six-key or larger deficit. With the Democrats holding a two-key cushion, the Keys model predicts Harris will become America’s first woman president and its first president of mixed African and Asian descent.

The following nine keys line up in favor of the incumbent Democrats.

Contest Key 2: The Democrats have united in near unanimity behind Harris.

Third-Party Key 4: In recognition of his fading support, Robert F. Kennedy Jr. suspended his campaign. His endorsement of Donald Trump does not impact this key.

Short-Term Economy Key 5: It is too late for a recession to take hold of the economy before the election. The National Bureau of Economic Research, which provides the most reliable assessment of recessions, typically takes a few months to establish that the economy has fallen into a recession (National Bureau of Economic Research, 2020).

Long-Term Economy Key 6: Real per-capita growth during the Biden term far exceeds the average of the previous two presidential terms.

Policy Change Key 7: Biden has fundamentally changed the policies of the Trump administration in areas such as the environment and climate change, infrastructure, immigration, taxes, and women’s and civil rights.

Social Unrest Key 8: Despite sporadic demonstrations, social unrest has not risen to the level needed to forfeit this key: massive, unresolved unrest that threatens the stability of society as in the 1960s and early 1970s.

Scandal Key 9: Republicans in Congress have tried and failed to pin a scandal on President Biden. His son Hunter’s crimes do not count as scandal, which to do so must implicate the president himself and generate bipartisan recognition of wrongdoing.

Foreign/Military Success Key 11: President Biden and Biden alone forged the coalition of the West that kept Putin from conquering Ukraine and then undermining America’s national security by threatening its NATO allies. Biden’s initiatives will go down in history as an extraordinary presidential achievement.

Challenger Charisma Key 13: As explained, Trump does not fit the criteria of a once-in-a-generation, broadly appealing, transformational candidate like Franklin D. Roosevelt or Ronald Reagan.

However, even if subsequent developments in Ukraine most improbably flipped both the Foreign/Military keys against the White House party, the deficit would be five keys, one short of the six negative keys needed to predict the incumbent party’s defeat. Thus, despite polls showing an uncertain election, the Keys’ prediction of Harris’s victory would not change.

Table 1. The 13 Keys to the White House: Brief definitions.

Disclosure Statement

Allan Lichtman has no financial or non-financial disclosures to share for this article.

References

Armstrong, J. S., & Cuzan, A. G. (2006, February). Index methods for forecasting: An application to the American presidential elections. Foresight: The International Journal of Applied Forecasting, (3), 10−13. https://faculty.wharton.upenn.edu/wp-content/uploads/2012/04/13KeysbyArmstrong-Cuzan.pdf

Dana, J., & Dawes, R. M. (2004). The superiority of simple alternatives to regression for social science predictions. Journal of Educational and Behavioral Statistics, 29(3), 317−331, https://doi.org/10.3102/10769986029003317

538. (2024, October 14). Do Americans have a have a favorable or unfavorable opinion of Donald Trump? https://projects.fivethirtyeight.com/polls/favorability/donald-trump/

Gallup. (n. d.). Presidential approval ratings – Donald Trump. Retrieved October 10, 2024, from https://news.gallup.com/poll/203198/presidential-approval-ratings-donald-trump.aspx

Gvishiani, A. D., Zelevinsky, A. V., Keilis-Borok, V. I., & Kosobokov, V. I. (1980). Methods and algorithms for interpretation of seismological data. Computational Semiology, 13, 30−43.

Graefe, A. (2015). Improving forecasts using equally weighted predictors. Journal of Business Research, 68(8), 1792−1799. https://doi.org/10.1016/j.jbusres.2015.03.038

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

Leip, D. (n.d.). Atlas of American presidential elections. Retrieved October 10, 2024, from https://uselectionatlas.org/RESULTS/

Lichtman, A. J. (2003). What really happened in Florida’s 2000 presidential election. Journal of Legal Studies, 32(1), 221−243. https://chicagounbound.uchicago.edu/jls/vol32/iss1/8

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. J. (2024). Predicting the next president: The Keys to the White House, 2024. Rowman & Littlefield.

Mebane, W. R., Jr. (2004). The wrong man is president! Overvotes in the 2000 presidential election in Florida. Perspectives on Politics, 2(3), 525−535. https://doi.org/10.1017/S1537592704040320

National Bureau of Economic Research. (2020, June 8). Determination of the February 2020 peak in U.S. economic activity. https://www.nber.org/cycles/june2020.html

Tetlock, P. E. (2017). Expert political judgment: How good is it? How do we know? Princeton University Press.

©2024 Allan J. Lichtman. This article 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 article.