Trump on course for narrow victory in 2024
Key takeaways:
Donald Trump would beat Joe Biden in the electoral college by 292 votes to 246 if a presidential election were held today
Biden is still on course to win the popular vote by 49% to 48%.
Despite Trump’s lead, the race in the battleground states is extremely close, with four states forecast to be won by less than 2% of the vote, and Nevada and Michigan projected to be won by Biden.
Amid calls from senior Democrats for Biden to step down as prospective nominee, the poll reveals Trump would also beat potential candidates VP Kamala Harris and Governor Gavin Newsom.
For mobile users, the dashboard exploring the data is available here.
Stack MRP projections
Following the successful use of MRP (Multilevel Regression and Poststratification) modelling by Stack in the UK in 2019, the US in 2020 and 2022, and Spain in 2023, this post summarises the results of the first Stack MRP predictions for the 2024 US presidential election. Stack’s latest MRP predictions reveal that Donald Trump is on course to narrowly beat Joe Biden in the electoral college, despite Biden holding a lead in the popular vote. Specifically, we forecast Trump winning 292 electoral college votes to Biden’s 246 (see the dashboard at the bottom of this page).
The poll shows Trump on course to win back four key swing states that Biden won in 2020: Arizona, Georgia, Pennsylvania and Wisconsin. All of these states were won by Trump in the 2016 election before flipping to Biden in the 2020 election, preventing Trump from the same path to the White House. These states are vital for Trump’s re-election. These four states had the closest margins out of all states in 2020, and Biden managed to win Arizona, Georgia and Wisconsin by less than one percent of the vote. In our current projections, however, we are seeing significant swings towards Trump in Georgia and Pennsylvania. In Georgia, Trump leads by 3.3% in our projection, while he is also projected to win Pennsylvania by 2.3%.
Despite the Trump campaign’s lead, the battleground states remain extremely tight. Four states are forecast to be won by less than 2% of the vote. Under the current estimates, Trump would win the vote in Arizona by 1.4% and in Wisconsin by 0.9%. This far out from the election, these are very small differences within a margin of error, and it is far from impossible that the Democrats will be able to close the gap by November 5th. You can explore all these results at your leisure in the dashboard below.
Trump and Biden are considered to be the most likely candidates for president for their respective parties. However, both are facing challengers as well, with a number of primary competitors for Trump and calls from senior Democrats to step aside for Biden. We tested whether candidates other than Trump or Biden would perform better for their respective parties. However, our MRP model shows no advantage to either party in changing their candidate. Donald Trump would beat both Gavin Newsom and Kamala Harris in the electoral college whilst Joe Biden would beat Governor Ron DeSantis by a score of 359 electoral college votes to 179.
“Our research is the largest exercise of its kind so far this cycle and we can confidently say that as things currently stand, if Donald Trump is selected as the Republican candidate, he is likely to win. Despite recent calls for change, our polling also shows that neither party would benefit from a change in candidate, President Trump would beat both of Biden’s possible replacements by an even greater margin.”
Methodological note
Finally, a brief note on methodology. Stack interviewed approximately 15,000 respondents from across the US about their political opinions between October 12th and November 3rd. These surveys then powered our tried-and-tested MRP model. MRP modelling is a statistical technique that uses large-sample national polling data to create estimates of public opinion at lower geographic levels. This model effectively used census data to generate a geographically detailed view of the make-up of the US electorate. It then estimates the relationship between these demographic and geographic variables and voting behaviour to project vote intention results at the state level.