Our first look at the 2024 elections ends with the biggest prize of all: the presidency of the United States.
Following big wins by Republican former president Donald J. Trump in the Iowa Caucuses and New Hampshire Primary, it is reasonable to assume a rematch of 2020, except now Democrat Joseph R. Biden, Jr. is the incumbent and Trump is the challenger.
***********
We start with a quick historic overview. Beginning with George Washington successful reelection in 1792, 61.5% of elected first-term incumbents have won reelection. The percentage rises to 70.0% in the 10 such elections since Franklin Delano Roosevelt first ran for reelection in 1936. The three losses were by Democrat Jimmy Carter in 1980, and Republicans George H. W. Bush in 1992 and Trump in 2020. Bush and Carter faced strong challenges from within their own party (Massachusetts Senator Edward M. Kennedy and commentator Patrick J. Buchanan, respectively) and economic downturns. In fact, Carter is the only elected first-term Democratic incumbent to lose reelection (in five opportunities) since Grover Cleveland in 1888. Trump quickly solidified his party behind him in 2020, but severely mismanaged the COVID-19 pandemic and its resulting economic dislocation, a key factor in his narrow loss.
Biden, meanwhile, is facing only token opposition from within his party and is presiding over strong economic growth as of January 2024. Thus, at least according to history, Biden should essentially be a lock to win reelection.
Which brings us back to Cleveland – the only defeated president to be renominated by his party four years later. Cleveland won his rematch against Republican Benjamin Harrison in 1892, giving some comfort to Trump supporters. Notably, though, Cleveland lost in 1888 despite winning the popular vote by 0.8 percentage points (“points”); he lost in the Electoral College because the 51 combined electoral votes (“EV”) of Indiana and New York narrowly flipped barely to Harrison.
In other words, Cleveland’s loss in 1888 arguably came “on a technicality.” By contrast, in 2020 Trump lost by 4.5 points and 74 EV – not a landslide, perhaps, but a clear victory nonetheless. So, while it is true that the only other presidential candidate in Trump’s position regained the presidency, the parallel is not quite exact.
There is also the election of 1984.
While the Democrats did not renominate Carter, who had lost by 9.7 points and 440 EV to former California governor Ronald Reagan, they did nominate his VICE president, Walter F. Mondale. This is the only time the losing incumbent vice president – someone twice on the same ticket as the losing president – was nominated for president four years later. Mondale, perhaps because of his association with Carter, struggled to win the nomination – then did about as well as you would expect, losing by 18.2 points and 512 EV, winning only the District of Columbia and his home state of Minnesota (barely).
Notably, Reagan was 73 years old in 1984, making him the oldest living president, just as Biden is now; in fact, Biden was four years older in 2020 than Reagan was in 1984. Meanwhile, I personally read the 1984 election as a vote for stability and continuity after 20 years of assassination, war, protest, social upheaval, resignation and presidential defeat; the victory of Vice President Bush in 1988 continued that trend. While COVID-19, Trump’s two impeachments and multiple indictments, and a Republican Party seemingly at constant war with itself, are not quite analogous to the historic upheaval of the 1960s and 1970s – a vote for Biden is still very much a vote for stability and continuity.
So, on balance, the historic record strongly favors elected incumbents presiding over good economic conditions – and has little to say about renominating presidents (or vice presidents) who lost in the previous election.
Biden should also be buoyed by recent history. Since 2000, the Democratic presidential nominee has won the popular vote five of six times – seven of eight going back to 1992 – and by an average of 2.6 points – which jumps to 3.3 points if you weight more recent elections higher.[1] Republicans won the presidency half the time starting in 2000 solely because of narrow Electoral College wins in 2000 and 2016. Basically, only one Republican – George W. Bush in 2004 – won both the popular vote and the Electoral College since his father won easily in 1988.
Turning to the last three presidential elections – from which I calculate my 3W-RDM measure of state-level partisanship – the weighted average is a 3.6-point advantage for the Democratic nominee. If this is the margin in the 2024 popular vote, Biden and Vice President Kamala Harris would almost certainly earn four more years in the White House (see Table 1 below), although eight states totaling 123 EV would have margins under four points; this variation is reflected in the range below each EV total. Not surprisingly, these eight states – Nevada, Michigan, Pennsylvania, Wisconsin, Florida, North Carolina, Arizona, Georgia – have been swing states for multiple election cycles.
Table 1: Projected Democratic state margins using multiple popular vote margins
| State | EV | 3W-RDM | D+3.6 (Ave. 2012-2020) | D-2.7 (Jan 2024 Polling) | D+0.3 (Hist + Poll) |
| District of Columbia | 3 | 82.7 | 86.3 | 80.0 | 83.0 |
| Hawaii | 4 | 29.0 | 32.6 | 26.3 | 29.3 |
| Vermont | 3 | 28.9 | 32.5 | 26.2 | 29.2 |
| Maryland | 10 | 26.2 | 29.8 | 23.5 | 26.5 |
| Massachusetts | 11 | 26.1 | 29.7 | 23.4 | 26.4 |
| California | 54 | 24.9 | 28.5 | 22.2 | 25.2 |
| New York | 28 | 20.2 | 23.8 | 17.5 | 20.5 |
| Rhode Island | 4 | 16.6 | 20.2 | 13.9 | 16.9 |
| Connecticut | 7 | 13.9 | 17.5 | 11.2 | 14.2 |
| Washington | 12 | 13.7 | 17.3 | 11.0 | 14.0 |
| Illinois | 19 | 13.3 | 16.9 | 10.6 | 13.6 |
| Delaware | 3 | 12.8 | 16.4 | 10.1 | 13.1 |
| New Jersey | 14 | 12.0 | 15.6 | 9.3 | 12.3 |
| Oregon | 8 | 10.1 | 13.7 | 7.4 | 10.4 |
| New Mexico | 5 | 6.3 | 9.9 | 3.6 | 6.6 |
| Colorado | 10 | 5.7 | 9.3 | 3.0 | 6.0 |
| Maine | 4 | 4.5 | 8.1 | 1.8 | 4.8 |
| Virginia | 13 | 3.9 | 7.5 | 1.2 | 4.2 |
| Minnesota | 10 | 1.8 | 5.4 | -0.9 | 2.1 |
| New Hampshire | 4 | 1.2 | 4.8 | -1.5 | 1.5 |
| Nevada | 6 | -0.5 | 3.1 | -3.2 | -0.2 |
| Michigan | 15 | -0.7 | 2.9 | -3.4 | -0.4 |
| Pennsylvania | 19 | -2.3 | 1.3 | -5.0 | -2.0 |
| Wisconsin | 10 | -2.4 | 1.2 | -5.1 | -2.1 |
| Florida | 30 | -5.5 | -1.9 | -8.2 | -5.2 |
| North Carolina | 16 | -5.8 | -2.2 | -8.5 | -5.5 |
| Arizona | 11 | -6.1 | -2.5 | -8.8 | -5.8 |
| Georgia | 16 | -6.5 | -2.9 | -9.2 | -6.2 |
| Ohio | 17 | -9.8 | -6.2 | -12.5 | -9.5 |
| Iowa | 6 | -9.8 | -6.2 | -12.5 | -9.5 |
| Texas | 40 | -12.0 | -8.4 | -14.7 | -11.7 |
| Alaska | 3 | -15.8 | -12.2 | -18.5 | -15.5 |
| South Carolina | 9 | -15.9 | -12.3 | -18.6 | -15.6 |
| Missouri | 10 | -19.0 | -15.4 | -21.7 | -18.7 |
| Indiana | 11 | -19.6 | -16.0 | -22.3 | -19.3 |
| Mississippi | 6 | -19.7 | -16.1 | -22.4 | -19.4 |
| Montana | 4 | -20.8 | -17.2 | -23.5 | -20.5 |
| Kansas | 6 | -21.3 | -17.7 | -24.0 | -21.0 |
| Louisiana | 8 | -22.3 | -18.7 | -25.0 | -22.0 |
| Nebraska | 5 | -25.1 | -21.5 | -27.8 | -24.8 |
| Tennessee | 11 | -27.2 | -23.6 | -29.9 | -26.9 |
| Utah | 6 | -27.6 | -24.0 | -30.3 | -27.3 |
| Alabama | 9 | -29.2 | -25.6 | -31.9 | -28.9 |
| South Dakota | 3 | -29.6 | -26.0 | -32.3 | -29.3 |
| Kentucky | 8 | -30.3 | -26.7 | -33.0 | -30.0 |
| Arkansas | 6 | -30.3 | -26.7 | -33.0 | -30.0 |
| Idaho | 4 | -34.8 | -31.2 | -37.5 | -34.5 |
| North Dakota | 3 | -35.4 | -31.8 | -38.1 | -35.1 |
| Oklahoma | 7 | -37.8 | -34.2 | -40.5 | -37.5 |
| West Virginia | 4 | -41.4 | -37.8 | -44.1 | -41.1 |
| Wyoming | 3 | -47.5 | -43.9 | -50.2 | -47.2 |
| TOTALS | 538 | 276 (226-349) | 212 (180-247) | 226 (212-276) |
Unfortunately, Table 1 also reveals the structural imbalance in the Electoral College: Democratic presidential nominees currently need to win the popular vote by at least 2.5 points to secure the minimum necessary 270 EV; going from most to least Democratic-leaning, Pennsylvania (D-2.3) and Wisconsin (D-2.4) are the tipping-point states, the ones that get a nominee to 270.[2] Vice President Albert Gore, Jr. and former Secretary of State Hillary Clinton won the popular vote by 0.5 and 2.1 points, respectively, in 2000 and 2016 but lost the Electoral College because of extremely narrow losses in one (Florida) or three (Michigan, Pennsylvania, Wisconsin) of these swing states.
Thus, while history says Biden should easily win reelection, the current polling and structural imbalance in the Electoral College tell a very different story.
***********
A total of 24 public polls assessing a hypothetical election between Biden and Trump have been conducted entirely after January 1, 2024. Nine polls assessed the election in multiple ways: first with only Biden and Trump, then again with those two plus other potential candidates, such as activist Robert F. Kennedy, Jr., Harvard University professor Cornel West, and former Green Party nominee Jill Stein.
Here are the weighted-adjusted polling averages (“WAPA”)[3] under three different conditions:
| Biden, Trump only (all 24) | Biden 42.4%, Trump 44.5%, Other/Undecided 13.1% |
| Biden, Trump only (w/other; n=9) | Biden 42.2%, Trump 44.6%, Other/Undecided 13.2% |
| Biden, Trump, Other (w/other; n=9) | Biden 35.8%, Trump 40.0%, Other 13.1%, Undecided 11.1% |
These data yield three conclusions:
- Biden currently trails by 2.1-2.4 points in a direct matchup with Trump, albeit with about 1 in 8 voters not sure for whom they will vote.
- Biden loses more (6.4 points) to other candidates than Trump (4.6 points)
- With third-party candidates in the mix, fully a quarter of the electorate is not yet choosing either Biden OR Trump.
Also, polling more than nine months from an election is not especially predictive, especially when the margins are well within the statistical margin of error (and the other/undecided is well over 10 points). Taking a weighted average of the 24 polls assessing Biden and Trump only (Biden-2.1 points) and the nine assessing Biden, Trump and at least one other named candidate (Biden-4.2) yields my current WAPA of Biden-2.7.
Curiously (and purely coincidentally), I suggested elsewhere that approval rating polling may be underestimating Biden’s strength by as much as six points. Adding six points to -2.7 gives you 3.3, roughly the average Democratic margin in the popular vote in recent presidential elections.
Still, I need to factor this early polling into an initial analysis of the 2024 presidential election. Using the average Democratic margin of 3.6 points, I calculated a time-weighted average (53.7% polling, 46.3% history) for both the 24 polls assessing Biden and Trump only (Biden+0.9) and the nine polls assessing Biden, Trump and at least one other candidate (Biden-0.9). I then calculated a weighted average (using number of polls as weights) of the two projections to obtain my initial estimate of the 2024 presidential election: Biden leads Trump by 0.3 points nationally.
Using the values in the last column of Table 1 (0.3-point popular-vote win plus state partisan lean), I calculated the probability of a Biden victory in each state (plus the District of Columbia).[4] Adding up the EV in states with Biden win probability greater than 50% gives him 226 EV (and Trump 312[5]), with Nevada (41%) and Michigan (40%) toss-ups, followed by Pennsylvania (29.6%) and Wisconsin (29.0%). Winning all four of these states would give Biden just enough EV – 276 – to win reelection. Indeed, this has been the most direct winning path for Democrats since at least 2008.
Meanwhile, nine states conducted a total of 14 state-level polls assessing Biden and Trump (and, sometimes, other) entirely after January 1, 2024: three in Michigan (3);[6] two each in California, New Hampshire and Pennsylvania; and one each in Georgia, Iowa, Nevada, New York and Texas. I was thus able to calculate an initial WAPA for these nine states and, using a somewhat arbitrary 3-point margin of error, estimate the probability of a Biden victory in these states.
Combining these with fundamentals-based probabilities, using an average weight of 44% polling/56% fundamentals,[7] in these nine states, while using the fundamentals-based probability in the other 40 states plus DC, suggests Biden will win 245 EV (as Pennsylvania’s 19 EV go to Biden), with Trump winning 293 EV, and thus the presidency. However, polls historically miss by up to four points towards one party or another, suggesting a Biden range of 195 – 276 EV – meaning the president is very much in the game based upon scant early polling.
Thus, using some very back of the envelope math, I estimate that – based solely on this combination of state fundamentals (based on national polls from January 2024) and state polling (ditto) – Biden has a win probability of 25%. Meanwhile, let us say that as an elected first-term Democrat presiding over an improving economy, Biden has at least a 90% probability of winning reelection.
Somewhat arbitrarily, I weight historic precedent 50% more than early polling. Thus, 3*90 + 2*25 = 270 + 50 = 320, and 320/(3+2) = 64%; this is my starting estimate, my Bayesian prior, of Biden’s win probability. This value falls between the 61.5% winning percentage for all elected first-term presidents and the 70.0% of such presidents starting in 1936, validating the methodology significantly.
All of which is a fancy way of saying that President Biden starts 2024 as roughly a 2-1 favorite to win reelection; in forecasting terms I translate this to “Lean Democrat.” Given that Biden does about 1.8 points better when no third-party candidates are tested, and given that third-party candidates tend to fade dramatically in polling, Biden is likely to net at least one point in the national polling over time. Furthermore, as more voters feel the economy (and their own post-pandemic lives) improving, undecided voters may well start to break for Biden, adding perhaps another point.
Of course, it is also possible Biden’s age and his handling of the Israeli assault on Gaza and the West Bank – or a surprise economic or foreign policy shock – widen Biden’s polling deficit. Trump still has a one in three chance to win back the White House, after all, which is the equivalent of needing either of two numbers to turn up on a fair die roll.
For all the uncertainty, though, I would much rather be Biden than Trump right now.
Until next time…if you like what you read here, please consider making a donation. Thank you.
[1] Weights: 2020 (6), 2016 (5), 2012 (4), 2008 (3), 2004 (2), 2000 (1). Adding 1992 and 1996 increases the weighted average to D+3.4.
[2] This imbalance vanishes if I calculate 3W-RDM using the last six presidential elections, however. A win of 0.2 points in the popular vote would likely suffice to make the Democratic presidential nominee the winner in the Electoral College.
[3] FiveThirtyEight recently reevaluated how it calculates pollster ratings. Essentially, they switched from a letter grade to a scale ranging from 0.0 to 3.0, with higher values denoting higher quality, based upon a combination of accuracy and transparency. Thus, I now weight pollsters by “pollster rating/3”. They also no longer present partisan bias estimates, so I no longer adjust topline values by those. The exception is if the poll was conducted for a partisan Democratic (-3), Democratic-tilted (-1.5), Republican-tilted (+1.5) or partisan Republican (+3) organization. I first questioned the value of their partisan bias estimates in the fall of 2022.
[4] Using the mean (-1.06) and standard error (5.77) of the differences between estimated and actual final state-level presidential election margins over the last three presidential election.
[5] For simplicity, I assume the 1st Congressional District of Maine votes for Trump and the 2nd Congressional District of Nebraska votes for Biden.
[6] Average pollster rating = 1.5, which translates to a C/C+.
[7] Weights are 50% for both “fundamentals” and polling on January 1, 2024, after which the polling gets a bit higher each day, using the midpoint of the most recent poll field date. That said, I also estimate a margin of error from the total number sampled in a state and calculate a polling weight based on mean pollster quality. On balance, this drops polling weight below 50% for some time. While these adjustments substantially lower polling weight at the state level, they would have a negligible effect at the national level.
Just Bear With Me... 
One thought on “A Somewhat Wicked Early Look at the 2024 Presidential Election”