If you’re reading this post on election day, please vote and help others to vote!
Otherwise, I’m writing today for a few reasons. One, to solidify and timestamp my thoughts on the 2024 election. Two, to give people a few more tools to understand the election results. Three, to help Democrats, progressives, Kamala Harris supporters and others have a little more confidence going into election night results.
I’ve broken up my supporting evidence into parts. The style is minimal commentary to give context and help readers understand the content presented. Enjoy!
Here’s my predicted 2024 election map. Hopefully this article will help explain why I think Kamala Harris will win the election.
- Polling
- The State of Polling Going Into Election Night
- Candidate Net Favorability
- Donald Trump Diminishing Support
- Some Problems with Polling in 2024
- Outlier Polls
- Previous Polling Misses
- Campaign Advantages
- Enthusiasm
- Donor Advantage
- Cash / Fundraising Advantage
- Ground Game Advantage
- Swing State Advertising and Budgets
- GOP Messaging / Anti-trans Advertising
- Crossover Support From Republicans
- The Electorate and Early Voting
- Early Voting – What Do We Know?
- Polls of People Who Have Already Voted
- Young Voters
- Reverse 2016 / 2020 2.0 / 2012 2.0
Polling
The State of Polling Going Into Election Night
Polling of the 2024 presidential election suggests a tight race. Kamala Harris has a slight lead according to most polling averages, particularly in the “Blue Wall” states of Michigan, Pennsylvania, and Wisconsin.
Election analysts typically look at three kinds of polls to understand the presidential race — national, state, and US House District polls. While Harris has a small lead in national polls, state and district polls seem consistent with a lead of 3-4 points nationally.
Polls are trying to poll the entire electorate. White voters, for example, are typically easier to poll accurately as they are often such a large portion of the population being polled. A lot of The Discourse during the 2024 election has covered whether or not Black and Latino voters will support Democrats at similar rates to 2020. We have a lot of evidence that they will. Here is evidence about Black voters:
Latino voters:
A state-by-state breakdown of support among Latinos. Florida Latinos favor Republicans, although that won’t hurt Democrats in the electoral college since Florida is not currently a swing state.
Candidate Net Favorability
A candidate’s favorability is related to but not the same as that candidate’s level of support. While Donald Trump has been historically unpopular, both Hillary Clinton and recently Joe Biden have also had poor net favorability (determined by subtracting the percentage of people unfavorable to a candidate from the percentage favorable to a candidate).
Kamala Harris has seen a dramatic rise in favorability since announcing her candidacy as the Democratic nominee in July:
Poll after poll shows a significant advantage of favorability for Harris / Walz compared to Trump / Vance.
Much of the spring and early summer campaign and media discourse focused on Joe Biden’s age, which ultimately contributed to his decision not to run for re-election. What was once Trump’s advantage on age and health is now to his deficit:
While favorables are somewhat less important with current supporters of a candidate, they are crucial with undecided voters. It may be hard to believe that anyone is undecided in the presidential election, but most polls in the summer and early fall found 5-10% of voters were undecided. Harris has the advantage with late deciders:
Donald Trump Diminishing Support
Political movements have a lifespan — a rise and fall. We have evidence that Donald Trump’s unexpected rise gave significant strength to the movement, but has failed to win more than a particular base of voters. Commenters have often said that Trump’s support has a “ceiling” of around 46%. We’ll see where that number lands in 2024.
Here’s another example of a movement losing support more than a century ago (and there are many more):
Here’s another more recent example. Bernie Sanders inspired tens of millions of people with a popular movement in 2016. Even though Sanders was unsuccessful that year, a new energy inspired many new candidates and organizers which are the heart of the Democratic Party today:
Yet in 2020, Sanders again tried to win the Democratic nomination. With more competition, he was unsuccessful in building a larger base and growing the movement further, even after Elizabeth Warren dropped out of the race.
The reality is that Trump’s largest base is non-college educated white people (and more staunchly, men). His base is diminishing rather than growing, and he has little chance of increasing his base of support.
Some Problems with Polling in 2024
Polling will always have error, either derived by sampling error or error related to modeling who will vote in the election. In this section I’ll provide evidence that pollsters may be underestimating support for Kamala Harris to address polling failures in both 2016 and 2020 which did not accurately measure Donald Trump’s support.
The New York Times wrote a terrific piece on the decision that most pollsters are making to weight their responses on “2020 recalled vote.” That means that they ask respondents who they voted for in 2020 (Donald Trump vs. Joe Biden) and then weight their responses to match voting results from that election. The method often boosts the support score of the losing candidate, and has a net effect of showing a tighter race.
Another controversial decision pollsters make is when deciding their Likely Voter (LV) sample. While a Registered Voter (RV) sample has a straight-forward weighting (matching registered voter populations), LVs involve many more assumptions. Some of those assumptions in 2024 are mostly correct, while others are highly implausible and some border on impossible.
Another choice is to change the projected weighting of racial breakdown in the projected electorate:
Some polls have few controls to model the partisanship of respondents taking their polls, or make assumptions about the makeup of who votes based on party.
Sometimes poll-to-poll variation is based solely on sampling different sets of voters rather than a measurable change in opinion:
Quinnipiac polls have been quite “bouncy” this cycle:
Another important factor in understanding poll results is the “herding” effect. No pollster wants to look like their poll was significantly wrong, so there’s a strong incentive to tweak Likely Voter models to produce a closer race, or one that is close to an existing polling average or previous result. While the poll result may look convincing, it does little to inform us about the state of the race.
Polls should have random sample variation, and when they don’t, that is suspicious.
Another factor mentioned earlier is the divergence between national, state, and district level polls, which makes it harder to understand the true state of the race:
On a final note, my point in sharing this section is that while polling can tell us a lot about the state of the race, it can’t tell us everything. We should understand the limitations of polling and their failure modes to give a more informed opinion on who might be winning the race.
Outlier Polls
Previously I talked about polling and sample modeling which can act as a forcing function on the resulting data. What about polls that stray far from previous polling averages or election results? The polls are Outliers. While outlier polls may be confounding for pollsters, the public can learn quite a bit from the direction of these polls even if the magnitude is less accurate.
I present to you the famous Selzer poll which has broken everyone’s brain since its release a few days ago, showing a result of Harris +3 in Iowa. That’s a change from her previous poll of Trump +4 this summer and the 2020 election result of Trump +8 vs. Joe Biden.
Ann Selzer is a terrific pollster which a strong record of accuracy and finding trends:
Seltzer’s latest poll shows some of the starkest gender divides we’ve seen in 2024:
Another massive outlier from New Hampshire:
We have several outliers from the Midwest. What might they tell us?
We’d be hard-pressed to find this kind of signal when pollsters model-away their outlier samples. Will they ultimately be right? I can’t wait to find out.
Previous Polling Misses
Most people know and understand that polls can be inaccurate. Understanding that effect will be critical to understanding possible outcomes in 2024.
Since 2016, polls have underestimated Donald Trump’s support significantly. Looking at the graph below, polls in 2020 were not significantly more accurate than those in 2016.
However, polls have not always underestimated Republicans. Just two cycles prior to 2016, polls were underestimating Barack Obama and Democrats’ support, especially in 2012:
So, pollsters do try to make adjustments to modeling and methodology to address previous inaccuracies. What does that mean for this year? Hard to say, although we know pollsters are trying not to undercount Donald Trump’s support, and do show a race similar to 2020, yet tighter.
Another interesting fact about polling misses:
The 2022 midterm saw polling error go in the reverse direction. So will polling error this year look more like 2020 or more like 2022?
Historical polling errors:
We all just respond to incentives:
Campaign Advantages
Kamala Harris has run one of the strongest campaigns in history. And she built a lot of that strength in just three months.
Enthusiasm
Similarly to favorables, enthusiasm for a candidate can significantly increase individuals’ willingness to vote and volunteer in campaigns. I think it’s accurate to say that no one was looking forward to a rematch of Joe Biden vs. Donald Trump. Kamala Harris’ entry into the race changed everything.
Moving away from the data to more anecdotal evidence, I’ll share various stories about campaign rallies and crowd size. We all know Donald Trump is obsessed with crowd size, and it likely doesn’t have much predictive value for the election. But the difference between the two campaigns is stark.
Donor Advantage
Kamala Harris has exploded campaign donations since entering the race, raising an unheard-of billion-plus dollars, mostly through smaller donations. Those small donors are key to people’s investment in a candidate and winning campaign.
At least Trump and the Republicans have their billionaires.
Cash / Fundraising Advantage
Aside from the number of donors, what about cash raised and spent itself? Harris has a sizable lead there, too.
Not only does Harris have the advantage in the presidential race, Democrats have an advantage in House race spending which should help turn out more Democrats.
Ground Game Advantage
“Ground Game” refers to campaigns’ teams in the field contacting voters, convincing them if necessary, and getting them to ultimately turn in their ballots. Field teams are secondary to good polling, but an excellent field operation and volunteer canvassing program can shift states 1-2% which is the same range that nearly every swing state polling suggests.
Kamala Harris seems to have built a fantastic field operation building off of previous successes. 2020 was a strange year due to COVID which prevented typical Democratic field operations (i.e. door knocking) from reaching full capacity. I’m cautiously optimistic that 2024 will be a great success.
The Harris campaign has reported huge numbers of volunteers helping:
Independent polling suggests that Democrats have had significantly more success in contacting voters:
Conversely, Donald Trump’s campaign chose to outsource its canvassing operations, largely to Elon Musk. All available reporting suggests the effort is not going well and will not be nearly as effective as what Democrats are doing.
Swing State Advertising and Budgets
Kamala Harris has a significant advantage in funds contributed to TV and online advertising. Note this does not count “Super PACs” which are outside funding groups (often funded by billionaires) which cut into Harris’ advantage.
GOP Messaging / Anti-trans Advertising
Poll after poll shows that voters’ top concern are economic issues. While Harris also devotes time to other important issues to Democrats like abortion access and democracy, those aren’t the top messages to a more general swing-state audience. Harris has been hammering the the issue on which so many voters decide their vote.
Additionally, the campaign put out some really smart ads targeting the biggest cities in swing states:
Contra to the Harris campaign strategy, Trump’s strategy has focused on negative ads:
What kind of negative ads has the Trump campaign run? Well, the kind that upset people.
The difference in messaging has caused Harris to rise in polling questions about the economy:
Then, in a widely publicized rally in New York at Madison Square Garden, racist attacks and jokes turned more voters away from Trump, including a joke calling Puerto Rico a “floating island of garbage.”
Crossover Support From Republicans
Take them or leave them, the Republican Party has gone fully-Trump in the past eight years. You can certainly argue that signs of the party’s authoritarian, anti-science, anti-people positions have been brewing for decades. But some older Republicans are deeply repelled by Trumpism. Kamala Harris has positioned herself to embrace anti-Trump Republicans without making any policy concessions. This is about democracy.
I don’t have any illusion that Republicans en masse will reject Trump. Partisans will partisan. But if these stories are any indication, just a one or two percent shift of Republicans away from Trump would provide a meaningful margin in swing states.
Aside from Republican elected and former elected officials, I see numerous posts on social media about Republicans voting for Trump. Are they real? Maybe. Are they common? Probably not. Will they make a big impact? Also, doubtful. But they might be notable so I’ll include a few here.
What other data points might we have about who Republicans are voting for? Nikki Haley voters might offer a clue. Or, Trump voters are more likely to vote on election day rather than vote early. I guess we’ll have to see. Here are some maps of Wisconsin:
Or maybe most Republicans are just voting for the Republican.
The Electorate and Early Voting
Early Voting – What Do We Know?
If there’s one potential source of error in trying to predict election results in 2024, it may come from analysis of early voting. The 2020 election was without precedent in regards to the number of Americans who voted early. Since that time, we’ve also seen a significant reversion to the mean — that it appears most voters are voting later this year.
I do think there are a few things we can learn from early voting — does it look like any previous year, or similar patterns we’ve seen between 2020 and 2022? Where are the early votes coming from? How many previous 2020 voters are shifting to earlier voting who voted on election day previously? How many are simply waiting until election day to vote? Let’s dive in with the caveat that any conclusions may only be telling half of the story.
We’ve already found two opposing factors — comparing a gender shift with an increase in early voting by Republicans. Which effect is bigger?
I think you mean may be good for Harris!
Voters in urban centers appear to be voting later this year:
Republicans had a strong start in early voting, but on the eve of election day, Democrats gain a slightly larger advantage than 2020. Now who will show up on Election Day?
Pennsylvania
Let’s dive into individual state data.
Democrats in Pennsylvania started with a significant advantage in Vote By Mail (VBM) ballots requested, as well as ballots returned.
However, by day 15, the return rate had narrowed significantly even though Dems had turned in nearly twice as many ballots.
On Election Eve, return rates had nearly equalized. Dems hold a significant edge, and based on past Independent vote splits, their advantage going into election day is about 500k votes. All analysis suggests that this puts Dems on relatively equal footing with Republicans, the latter being more likely to vote on Election Day.
What about individual county returns?
And city returns? Philadelphia probably needs a large election day turnout to provide a large number of Democratic votes statewide.
The biggest question is whether early GOP voters are new voters, or if they’re previous election day voters that have shifted to voting early:
A lot of signals suggesting there’s room for new early voters and possibly new Election Day voters for Democrats.
Michigan
Some potential bright spots for Democrats in Michigan which is polling the best for Democrats of the swing states.
The data we have so far suggests that Detroit had low-ish turnout in 2020 and will beat that mark in 2024.
Wisconsin
Is this turnout good? Hard to say!
Milwaukee is a big source of Democratic votes in Wisconsin:
A week before the election:
Are enough people voting? OMG
And things looking better on Election Eve as long as there’s decent turnout on election day:
Nevada
Nevada may be the biggest source of words and controversy in 2024’s early voting season.
Republicans grew a sizable advantage in early voting, hence the Democratic freakout:
How many mail ballots will ultimately be returned?
As early voting started to finish, Dems and Others started gaining ballot share:
Early In-Person (EIP) voting:
Some decent election day cannibalization of voting from Republicans compared to Democrats / Others:
North Carolina
North Carolina was the only swing state that Democrats did not win in 2020, so the default of “like 2020” may not be good enough for a Democratic win.
Given reasonable Dem performance in early voting, another useful stat:
Large turnout:
Again, seeing a fairly decent shift of GOP Election Day voters to early voters:
Looks like some folks are optimistic:
Arizona
I don’t have much for Arizona besides what other people are saying: Dems winning the state largely depends on GOP crossover and large Unaffiliated voter turnout:
Georgia
Georgia was an unexpected win for Joe Biden in 2020. Can early voting tell us if Kamala Harris will win here? Probably not! But it might give us some clues.
Based on 2020 results, you might expect Georgia to left of North Carolina:
Turnout in Georgia is high and consistent with recent elections:
However, turnout percentage is tough to pin down as some parts of the state have rapidly growing populations:
So who’s winning in turnout?
Everyone wants to know, at what rate are Black voters voting?
Larger turnout from non-white voters in general:
Young voters are definitely voting:
Where are the votes coming from and how many left to vote on Election Day?
What does the gender ratio look like?
Where does all this early voting land?
Texas
Bonus early voting — Texas, which is not a swing state.
A lot of Texas’ early vote increase is also population increase. So we’ll have to see how that causes the state to lean demographically and electorally.
Polls of People Who Have Already Voted
We’ve talked about polls. We’ve talked about early voting. What could be better than polls of people who early voted? These data sets give us a little more insight into how votes are breaking down of people who have been verified to have voted (although I can’t guarantee all pollsters are verifying ).
The takeaway is that Kamala Harris is dominating the early vote. However, given this is an important presidential election, it’s highly likely those who have not yet voted and are enthusiastic to vote will end up turning in their ballots which will drive these numbers closer.
In general, we’re trying to compare how “red” the electorate is vs. vote share to identify how much crossover support Harris is getting from non-Democrats
Young Voters
Young voters are a key Democratic constituency. We have some reason to believe they’re motivated to vote this year.
If I’m reading this graph right, Gen Z will have a decent chunk of new voters in 2024. NYT survey said they expect 13% of the electorate 18-29 which would be historically low in recent elections.
There’s significant upside for Harris if young voters exceed normal turnout.
2020 early voting was very high, so young voters are doing well in 2024 in comparison:
Additionally, young voters are the most likely to be new voter registrations:
Gen Z is really coming into the fold as a significant voting bloc. Gen-X tends to be the most pro-Trump while Gen Z is the most pro-Harris:
Reverse 2016 / 2020 2.0 / 2012 2.0
Many folks are trying to figure out — what election offers the best parallel to what 2024 will be? a Reverse 2016 would show Trump underperforming his polls and running an overconfident but weak campaign. 2020 2.0 would be pulling narrow wins out of key swing states. 2012 2.0 would be a big polling miss in Harris’ favor and a landslide victory. While will it be?
Reverse 2016
2020 2.0
2012 2.0
Thanks for reading! Questions, comments, or otherwise can be emailed to me Eric Budd at ericbudd@gmail.com or on social media through Bluesky at @ericmbudd.bsky.social, on Twitter @ericmbudd or on Mastodon at @ericmbudd@toot.bldrweb.org.
Learn more Eric Budd.
*Note – I recognize that many of the sources used here are from men. If you have recommendations for women doing great work in this space, please send them to me!
Effective Discontent — Eric Budd 