2024 Election Forecast

| President | Senate | Governors | House | Results |

For this year's forecast, I'm trying out some new techniques in an effort to improve from the last election model. Similar to last time, the model will primarily emphasize the probability of an outcome occurring and the uncertainty of the forecasted result. A core weakness of previous forecasts was that the outcomes between different states were correlated in ways that the model had a tendency to underestimate the chance of an upset, and this effect was most notable in the 2022 house forecast. Hence, the methodology for forecasting individual states has not changed, but how the individual probabilities are aggregated is significantly different. Some results from running this model on previous presidential elections is shown here.

The final results from this new methodology shows President Trump with a narrow lead in most battleground states but also shows Vice President Harris with a slightly higher chance of winning the election. This result sounds counterintuitive, but it can be explained by the upper Midwest being very close in the polls, President Trump's distinctive but small polling leads in the Southern and southwestern battleground states, and the new model's assumptions about how polling error between individual states are related.

The new model shows a significantly higher probability of a Harris win than the previous model and other public model-based forecasts. I think that the reason for this effect in this model is that while certain states tend to be correlated with each other, say Wisconsin and Michigan, the model takes into account that there are some states that have the exact opposite pattern, where if we assume that the overall popular vote is held constant, then the states tend to move in opposite directions in historical trends, such as Wisconsin and Texas. This effect is taken into account in the new model, but not the old model. An assumption of the new model is that historical inter-state moves are indicative of how polling error can be correlated, and this assumption is based on the notion that polls tend to have a tendency to model the electorate based on the last election and as a result have similar errors when an electoral realignment occurs. This year, the Midwestern battleground states (Michigan, Pennsylvania, Wisconsin) and the Southern battleground states (Arizona, Georgia, Nevada, North Carolina) follow this anticorrelation pattern, and given that Harris only needs to win one of these two combinations to win, this phenomenon increases the probability of a Harris victory. On the other hand, Trump needs to win some combination of states from both of these groups to win, which overall lowers the chance of a Trump win when considering these correlations.

This phenomenon, in addition to the fact that Trump appears to be overperforming in heavily urban areas that favor Democrats, makes it a real possibility that Trump wins the popular vote but loses the electoral college. This year, it is not so clear that a split between the populate vote and the electoral vote would necessarily disadvantage Harris, though it remains more likely than not that the margin of the tipping point state will be more favorable to Trump than the national popular vote.

The latest polls going into the election shows Vice President Harris and President Trump within a point in the three Midwestern "blue wall" states of Michigan, Wisconsin, and Pennsylvania. If the polling is accurate in all other states, then losing the blue wall means losing the election for both campaigns. This virtual tie in the blue wall state polls, plus the extraordinarily narrow polling lead that President Trump holds in Arizona, Georgia, Nevada, and North Carolina, effectively means that the distribution of the results of the model is dominated by inter-state covariance assumptions rather than by distinctive polling leads in particular battleground states. This is the primary reason for why adding up the raw ratings of individual states is indicating a Trump win in the electoral college, but the overall likelihood of Harris winning the election remains somewhat higher than what intuition would suggest. Given that the Midwest is in the pathway to victory for both candidates and the history of polling errors in recent years in the Midwest, a distinctive possibility in the election outcome is President Trump again overperforming in the Midwest and winning the election. I think that it is likely that betting markets are pricing in this effect, since all major markets are showing a higher likelihood of a Trump win than poll-based models.

A note copy-pasted from previous years, I'm putting here the important caveat with reading probabilistic models: Just because it's more likely than not to happen doesn't mean that it will. Furthermore, this forecast is not meant to be an endorsement of a political candidate, party, ideology, or system, but rather meant to be an academic exercise on predicting election outcomes.

As with the last election, there will be a live election results stream on Twitch during election night, starting at 6:00 PM EST on November 5, with a live model based on partial results and commentary focusing on the political geography of the data. In 2020, I forecasted the reasonably high probability of Biden's victory while Trump led the overnight vote count in key Midwestern battleground states, and I'm hoping to repeat that this year.

So here are the results:

Maine and Nebraska award electoral votes both for the statewide winner and for the winner in each of its congressional districts.

Overall

Aggregating state-by-state estimates together produces an overall simulation of the Electoral College. The numbers used in this simulation are from the results of the 2020-NN-M model:

The remaining 3.1% is the possiblity of a 269-269 tie.

The median result is Harris winning 270-268.

The most likely result is Harris winning 270-268.

An 80% confidence interval ranges from Harris winning 325-213 to Trump winning 326-212.

To give a better sense of distribution, despite the headline probabilities being very close, the most likely outcomes are: (1) Harris barely winning 270-268 by claiming Michigan, Pennsylvania, and Wisconsin, (2) Trump sweeping all major battleground states (AZ, GA, MI, NC, NV, PA, WI) to produce a 312-226 Trump victory, (3) a 292-246 Harris victory, which involves Harris winning the Midwestern battlegrounds of Michigan, Pennsylvania, Wisconsin, plus Nevada and one of either Georgia or North Carolina, (4) Harris winning 308-230 by winning all major battleground states except Arizona, (5) a 301-237 Trump victory by winning all major battleground states except Arizona, (6) Harris winning 276-262, which involves Harris winning Michigan, Pennsylvania, Wisconsin, and Nevada, and (7) a 269-269 tie in the electoral college. These scenarios account for about 31 percent of the outcomes.

History

This table shows the results of the simulation over time.

Below are the results of the same simulation, but with President Biden at the top of the Democratic ticket, which was run up until July 21.

For comparison, here are the results using the 2018 methodology that was used for the 2020 election.

2018 model with President Biden, up until July 21:

Analysis

The key states to watch are the following:

Arizona (Leans Trump). President Biden is the only Democrat to have won the Grand Canyon state since President Clinton's reelection in 1996. However, Democrats have found more mixed success at the statewide level in recent years, including key wins in the 2022 midterms. That being said, Biden's margin of victory was an exceedingly narrow 0.3 percent, and Arizona as a whole remains a highly competitive state. Most of the vote in Arizona comes from Maricopa County, which includes Phoenix, and President Trump's underperformance among suburban voters here was crucial to his narrow loss in the state. The most likely path for Trump to flip Arizona's 11 electoral votes back is through suburban voters in Maricopa County, and Arizona is likely a must-win state for Trump to 270 electoral votes. While results come in on election night, Arizona is not a state that we can model easily due to its concentration of vote in a single county.

Florida (Likely Trump). Over the past decade, the Sunshine State has shifted to become increasingly out of reach for Democrats to win statewide in a way that has caused it to lose its status as a bellwether state. Two key shifts are responsible for this phenomenon: the first is that support for Democrats among rural voters in Florida has collapsed, and the second is a shift in support among Hispanic voters in South Florida, particularly Miami-Dade County, towards the Republicans. In order for Vice President Harris to flip the state, Democrats would have to reverse these trends. Florida has voted with the winner since 1996 until it voted for President Trump's reelection in 2020. Florida's 30 electoral votes are considered must-win for President Trump, though its electoral climate is much more favorable to Republicans than it was four years ago.

Georgia (Leans Trump). President Biden is the only Democratic candidate to have won the Peach State since then-Gov. Bill Clinton won it in 1992. The key to Biden's 0.2 point victory in 2020 was overperforming among urban and suburban voters in the Atlanta metro area. In order to flip the state, President Trump would need to either win back some traditionally Republican voters in suburban Atlanta or turn out slightly more rural voters who make up the core of the GOP base in the Deep South. Georgia is a state with a heavy urban-rural divide, with Atlanta and its suburbs heavily favoring Democrats and much of the rural areas favoring Republicans, with some Democratic pockets in between as part of the Black Belt. President Trump has to win Georgia's 16 electoral votes to have a realistic chance of reaching 270.

Iowa (Leans Trump). Despite being a traditional battleground state, the Democrats have effectively conceded the Hawkeye State following former Secretary of State Hillary Rodham Clinton nearly double digit loss in 2016. That margin was repeated by President Biden in 2020, despite President Obama having comfortably carried Iowa twice. Iowa's trend towards the Republican Party mostly resembles the course of its southern neighbor Missouri, which was also a longtime battleground state until 2012, and is part of a much broader trend of rural Midwestern voters realigning with the Republican Party. Since 2016, President Trump has carried traditionally Democratic voters in rural eastern Iowa, and it is a phenomenon that is expected to repeat this cycle. However, a last-minute bombshell poll from reknown state pollster J. Ann Selzer shows Vice President Harris leading in Iowa, and if that result is anywhere near true, that could imply Harris wins throughout the Midwest, especially in Michigan and Wisconsin.

Maine (Likely Harris). Maine is one of two states, alongside Nebraska, that splits its electoral votes by congressional district, and there is a demographic split between the more urban, Portland-based first congressional district, and the second congressional district that covers the rural interior. President Trump won the second congressional district twice by comfortable margins following a nationwide shift of rural voters away from the Democratic Party in 2016, and is favored to win the single electoral vote again. Meanwhile, the first district is not contested and is expected to go for Vice President Harris, though how two districts balance out will determine the two statewide electoral votes, which currently narrowly favors Vice President Harris. Vice President Harris must win the two statewide electoral votes to have a realistic path to 270.

Michigan (Leans Harris). As one of the three Midwestern "blue wall" states that have come to decide presidential elections since 2016, the Wolverine State again finds itself at the center of attention. Since George Bush's 1988 landslide, no Republican has ever won Michigan with the exception of President Trump's upset in 2016 due to winning over traditionally Democratic rural voters and working class union households. President Biden flipped Michigan back in 2020 by winning back some working class voters while also cutting into traditionally Republican suburban voters, particularly around Grand Rapids and Traverse City. The key to winning Michigan remains appealing to surburban voters and working class voters, which both campaigns are attempting to do. Vice President Harris likely needs to win Michigan to win 270 electoral votes. Without Michigan, Vice President Harris would have to win a large swath of Southern states to be able to win reelection. Current polling suggests that Harris is slightly ahead, but Trump could easily overperform again in the Midwest and pull off a win in Michigan.

Minnesota (Likely Harris). In a pattern similar to the Midwestern blue wall states, Minnesota has voted Democratic in every presidential election since 1976. President Obama won this state by large margins in 2008 and 2012, but Minnesota saw closer margins in more recent elections due to realignment of rural voters and working class voters across the Midwest. Despite Republican attempts to make a play for Minnesota, Democrats have managed to hold on. In the event of Trump overperforming in the Midwest, it's conceivable that Minnesota is in play. Minnesota is considered a must-win state for Vice President Harris.

Nevada (Leans Trump). Although no Republican has won Nevada since 2004, the Silver State features as a key battleground for working class Hispanic voters. President Obama won Nevada twice by wide margins, though President Trump has held the margins much closer by appealing to working class voters and Hispanic voters. Nevada has a history of inaccurate polling that tends to underestimate Democratic candidates, and a better than expected posting for Vice President Harris remains a real possibility. President Trump hopes the trend of Hispanic voters towards the GOP can flip Nevada, but Republicans still have to contend with the formidable Democratic turnout machine in Clark County established by the late Senate Majority Leader Harry Reid. Given the concentration of votes in one particularly populous county, Clark County, which contains Las Vegas, it will be hard to model this state on election night, and it may take a long time to know a winner in Nevada as it was in 2020 and 2022. While current polling data suggests a narrow Trump edge, it's easy to see this rating turning out to be wrong due to Nevada's history of polling errors.

New Hampshire (Leans Harris). Despite awarding only four electoral votes, New Hampshire could matter in a close election. In 2000, Vice President Al Gore could have won without Florida by winning the Granite State, and former Secretary of State Hillary Clinton won it by only a fraction of a point. Winning moderate surburban voters is the key to winning New Hampshire, and President Trump has historically struggled to win that vote. Currently, polls are suggesting that Vice President Harris will narrowly hold on to New Hampshire.

North Carolina (Leans Trump). Similar to Georgia, North Carolina has a large urban-rural divide, with Democratic urban centers around the Triangle and Charlotte and Republicans favored in much of the rest of the state. The Tar Heel State has been a fleeting target for Democrats in recent years, where every Democratic campaign has heavily invested in the state since President Obama in 2008, yet 2008 was the only time the Democrats were able to win. In 2016 and 2020, President Trump was able to win statewide by running up large margins in predominantly white rural areas even while Democrats posted a strong performance in the Triangle. Vice President Harris faces an uphill battle to win the state due to President Trump's strength among rural voters, which make up a significant portion of the state's population.

Ohio (Likely Trump). Similar to Iowa, Ohio has recently been losing its status as a core presidential battleground. The Buckeye State has voted with the winner of every presidential election from 1964 to 2016. That trend broke in 2020 when President Trump won Ohio comfortably due to his hold on traditionally Democratic working class voters in the Midwest despite losing nationwide. Ohio is now significantly more Republican leaning than the nation overall, and the Democrats no longer see Ohio as a top target. President Trump is a strong favorite to win again in Ohio.

Pennsylvania (Leans Trump). Similarly to Michigan, Pennsylvania is one of the three "blue wall" states in the upper Midwest that have come to define the most recent presidential elections. President Trump narrowly flipped the Keystone State in 2016 by winning over rural voters and traditionally Democratic working class voters in northeastern and western Pennsylvania. However, Trump underperformed among suburban voters in the Philadelphia suburbs, which President Biden capitalized on to win the state back in 2020. Biden also performed better than former Secretary of State Hillary Clinton among blue collar workers, which played a key role in his 2020 win. This year, it is again expected that working class voters and suburban voters will decide the state and perhaps the presidency, and both campaigns see Pennsylvania's 19 electoral votes as critical to victory. For Vice President Harris to win, Harris must hold Pennsylvania or hold some of the southern battlegrounds of Arizona or Georgia. Similarly for President Trump to win reelection, Trump can flip back Arizona and Georgia and one of the blue wall states. Current polling suggests a dead heat with maybe just barely a slight edge for Trump, but Pennsylvania remains highly volatile going into the election.

Texas (Likely Trump). With 40 electoral votes, Democrats have been eyeing the traditionally Republican state with more interest in recent cycles. While suburban voters around metropolitan areas have trended away from Republicans during the Trump years, particularly around Austin, Dallas, Houston, and San Antonio, Texas has remained a reliably red state where Democrats have found limited success. Part of this is due to countercurrents in the Rio Grande valley, where rural Hispanic voters have moved in the direction of President Trump. Democrats have not won an election statewide since 1994, when some down ballot Democrats were able to hold on to office riding the coattails of the late Gov. Ann Richards (D), despite Richards losing the gubernatorial race to George W. Bush, and no Democratic presidential candidate has won Texas since 1976, when former Georgia Gov. Jimmy Carter swept the South. In order for Vice President Harris to flip Texas, Democrats must vastly improve margins among urban and suburban voters in Texas while maintaining support from Hispanic voters, and a Democratic win in Texas likely implies a Harris landslide nationwide. It remains unlikely for this to happen, and President Trump is heavily favored to win.

Wisconsin (Leans Harris). Along with Michigan and Pennsylvania, Wisconsin is one of the three "blue wall" states that were traditionally Democratic until President Trump's upset victory in 2016 due to rural voters shifting towards the GOP. Similarly to Iowa, Democrats historically did well with rural voters in Wisconsin, particularly southwestern Wisconsin, but these areas have moved to be increasingly out of reach for Democrats. In 2020, President Biden won the Badger State by outperforming in traditionally Republican suburbs, particularly those around Milwaukee, despite Trump holding most of the GOP gains in rural areas. Wisconsin was decided by less than a point in both of the last two elections and played a pivotal role in Trump and Biden's victories respectively, and is heavily contested by both campaigns this year. The election in Wisconsin will be decided by how much Trump can improve Republican margins in rural areas and how much Harris can improve Democratic margins in urban and suburban areas. The polls show Harris very narrowly ahead, but Wisconsin has seen large polling errors the last two presidential cycles, and it's not hard to imagine Trump overperforming the polls here again to pull out a win.

Find a bug or inconsistency? Feel free to shoot me an email!

<< Back to Elections Home