Seven Months To Go: One New State Rating
Arizona shifts towards Republicans, North Carolina is (still) barely a swing state, and some thoughts on the gap between the popular vote and the Electoral College.
Welcome, everyone, to my second swing-state update! Last time, I went through the seven swing states in detail, sharing my thoughts on their electoral trendline and giving each one a rating for this fall. Since that post, which came out a few days after Super Tuesday, nothing too significant has changed in the race. We’ve had a decent amount of new polling over the course of March, which has been broadly consistent with the polls we had prior to Super Tuesday. To the extent that there’s been any change, it would seem that Biden has slightly improved his standing in the polls.
Given the overall stability in the race right now, going through all of the swing states in depth again doesn’t feel warranted, at least not yet. For today, I’ll instead briefly share the latest polling average for each swing state, discuss why I’m changing Arizona’s rating, talk a bit more about North Carolina, and end with my thoughts on the gap between the national popular vote and the Electoral College.
So, let’s start with the latest polling. Below, I have the March polling average for each swing state, based on the polls that FiveThirtyEight has tracked. For any state, you can click on its name to see the list of recent polls:
Arizona: Trump +4.8
Georgia: Trump +5.0
Michigan: Trump +3.0
Nevada: Trump +4.7
North Carolina: Trump +6.4
Pennsylvania: Trump +1.6
Wisconsin: Trump +2.3
Note: most of these polls are sampling registered voters rather than likely voters. This will change as we get closer to the fall, by which time pollsters will screen their respondants to determine if they are likely to actually vote in the election. I mention this because it’s possible that polling averages will shift from that change alone.
If we order the swing states based solely on the simple polling average, we get some pretty interesting results, which differ quite a bit from the order we saw in 2020. Notably, based on the 2024 polling, Nevada is among the most Republican states, whereas in 2020, it was the second-most Democratic of these seven states, behind only Michigan. Also, Pennsylvania being the bluest swing state here is noteworthy, as it was only the third-most Democratic of these states in 2020, when it was a full point redder than Nevada and about 1.6 points redder than Michigan.
As mentioned, I do have one change to make to my state ratings, prompted in part (but not entirely) by the polls. I’m moving Arizona from Tilts Democratic to Tilts Republican. I don’t make this change lightly, as back in February, I wrote an entire post about Arizona, arguing that the long-term trends there have been quite positive for Democrats, and that it might be among Biden’s better swing states this fall.
I do still think that Trump is a poor fit for Arizona (relative to a more generic Republican candidate), which 2016 made clear, given how much the state swung away from Republicans from 2012. I also feel that in the near future, the state is clearly not returning to its old electoral habits from the 2000s, when it voted for the Republican candidate in every presidential election by anywhere from 6 to 10 points. But at this moment, it’s perhaps more accurate to think of Trump as having the slightest advantage in Arizona.
One thing that’s changed my thinking is the presence of the Green Party on the ballot in Arizona this fall. It’s hard to quantify the exact effect that could have, but it’s also not hard to see how it helps Trump. In fact, the third-party vote is arguably the single most important wildcard this fall, and nationally, it’s entirely possible that neither Biden nor Trump reaches 50% of the popular vote (in 2016, Clinton won about 48% of the popular vote, while Trump won 46%, and the rest went to third parties or write-ins). Below, I have a map from the Green Party’s website showing the states they have obtained ballot access in for the 2024 election.
This is a map that I hope to return to again in a future post, and as an aside, it’s worth noticing that Nevada and Pennsylvania could end up as the only two swing states without the Green Party on the ballot this fall. But in terms of Arizona, the state saw substantial levels of third-party voting in 2016, when nearly 7.5% of the statewide vote did not go to either of the two major parties, compared to about 6% nationally. That year, the Green Party received 1.3% of the vote in Arizona. In a state that was decided by 0.3 points in 2020 (when the Green Party was not formally on the ballot), the presence of the Green Party in 2024 could have an outsized impact, especially given the unpopularity of both major party nominees.
More broadly, I’m also coming around to the idea that Biden might really struggle to win again in the Sunbelt states of Arizona, Georgia, and Nevada this fall, which are quite a bit more diverse than the three northern swing states of Michigan, Pennsylvania, and Wisconsin. Perhaps Biden’s 2020 coalition, which reached record levels of support with college-educated whites while still retaining strong (though noticeably weaker) non-white support, was just enough to flip Arizona and Georgia blue. But, in the political environment of 2024, which could witness unprecedented racial depolarization and vote switching among non-whites, that coalition might just come up short in the diverse states of the Sunbelt. I’ll talk more about this in the second half of the post.
To give my usual disclaimer, no rating is final. If the polling changes in either direction, I’ll adjust Arizona again accordingly. But there’s enough reason right now to believe that the state could narrowly flip back to Trump, and that what we’re seeing in recent polls of Arizona (a state where Biden polled quite well throughout 2020) is part of a bigger realignment, in which the support of non-white voters is up for grabs by either party.
New Arizona Rating: Tilts Republican
The other state I wanted to bring up is North Carolina, which I currently have rated as Likely Republican. I’m not changing this rating today, but I did hope to justify it a bit more, as it’s one of the two states where I differ most from the conventional wisdom (Nevada being the other). For example, the website Inside Elections has North Carolina at Tilts Republican, while Sabato’s Crystal Ball has it at Leans Republican. These are both entirely reasonable ratings, and the state is undoubtedly going to be closely contested again in 2024. But for me, the uncertainty with North Carolina isn’t directional; it’s very likely going to be at least as Republican as it was in 2020. The real uncertainty with North Carolina is whether Trump wins it by 2, 3, or even 4 points (anything beyond that would be genuinely surprising).
What has shaped my thinking most is the fact that North Carolina has a tendency to follow shifts in the national popular vote in presidential elections, which it has done with surprising accuracy since 2008. Since then, wherever the popular vote lands, North Carolina has been roughly 6 points to the right of it. As such, the state generally does a good job of reflecting national shifts from one election to another, even though it’s somewhat more Republican than the country as a whole.
This behavior is probably due to the fact that the state is demographically comparable to the country, though North Carolina has a larger African-American population and smaller Hispanic population than the nation as a whole does. It also has a relatively balanced mix of urban, suburban, and rural populations, which isn’t true of all of the swing states. It’s educational attainment is even right at the national average, with about 35% of adults in North Carolina having at least a college degree. In short, the state is surprisingly representative of the US, which I think is part of the reason it does a good job of following changes in the national popular vote.
So, that’s a long way of saying that Biden could really only win North Carolina this year if a.) he improved on his national popular vote margin from 2020 or b.) something pushed the state to the left, even if for just one cycle, that made it defy Biden’s likely erosion in the popular vote this fall.
Neither condition seems too likely, and I think my rating reflects that. Without a doubt, North Carolina will be one of the nation’s closest states this fall. It’s apparently a state where Democrats have a high floor in presidential elections (i.e. getting to 48% of the vote is very doable), but winning outright seems especially difficult for the party. North Carolina has voted Republican in every presidential election since 1980, except for when Obama won it by 0.3 points in 2008. The national polling this year doesn’t seem indicative of the type of political environment needed to flip North Carolina blue.
To me, there’s no real contradiction between a state being rated as Likely Republican and being perennially close. In all of the recent polling that FiveThirtyEight has tracked in North Carolina, Trump’s lead has ranged from 3 to 6 points, confirming that it will be a somewhat close state, but not one that’s poised to flip, unless something truly significant about the election changes before November.
If any new developments emerge this year that suggest the state is truly in play, I’ll have no problem moving North Carolina to Leans Republican or maybe even Tilts Republican, but for now, it’s just not in the same category as the main six swing states.
So, with North Carolina staying at Likely Republican, and the Arizona rating change to Tilts Republican, my Electoral College map currently looks like this:
I’d like to think this map broadly reflects the current state of the race: Biden is a slight underdog, but there’s still much uncertainty, almost all of which is concentrated in just six or seven swing states. Compared to my previous map, which showed a nearly tied Electoral College, this one has Trump leading in electoral votes by a modest, but fairly narrow margin.
As I said last time, I wouldn’t read too much into the overall electoral vote count at this early stage. In my mind, the tilt states are maybe 55-45% or at most, 60-40% favoring that party. They should be thought of as being close to coin flips, with either party fully capable of winning. And of course, let’s not forget that winning any of these states by even a fraction of a point means winning all of their electoral votes. So, please don’t look at the map above and mentally assign states in the lighter shades to either candidate. We’re only on the cusp of April, and even come fall, I’d expect at least some of the swing states to still have a tilt rating.
In other words, embrace (or, at least tolerate) the uncertainty that winner-take-all electoral vote allocation brings with it. And while you’re doing that, feel free to leave your thoughts in the comments below. Particularly if you disagree with me about a certain state, or anything else I’ve raised, I want to hear it. Electoral analysis comes with plenty of assumptions, and being challenged on those is good for all involved.
The final topic I wanted to raise today is the popular vote, which I spend a lot of time thinking about it. Hopefully, you’re not like me, and you think about normal things. As we all know, the national popular vote doesn’t determine the winner of US presidential elections, but it’s worth trying to estimate, even in the broadest terms, what popular vote margin a candidate might need to prevail in the Electoral College.
Ever since 2016, when Clinton won the popular vote by 2 points while narrowly losing Michigan, Pennsylvania, and Wisconsin (the three states that would’ve most easily gotten her to 270+ electoral votes), it has been assumed that Democrats are inherently disadvantaged in the Electoral College. It’s an understandable claim, given that the 2000 election was another recent example of a split verdict, as Al Gore lost the election despite winning the popular vote by about half a percentage point (though Florida looms much larger than the Electoral College does for why Gore lost).
But it’s important to realize how much the bias in the Electoral College is dependent on the specific party coalitions in a given election. For instance, the Electoral College actually favored Democrats as recently as 2012, when Obama could’ve potentially won the election while tying or even losing the popular vote. That was in large part because he did relatively well with northern working-class whites, who are overrepresnted in some of the most important swing states (recall that Obama even won Iowa and Ohio in 2012, two states that have long since become uncompetitive). Obama’s coalition just happened to be very efficient and well-distributed from an Electoral College standpoint.
On the other hand, Biden’s coalition in 2020, which was broadly similar to Clinton’s in 2016 (though stronger with college-educated whites and weaker with non-whites, especially Hispanics) faced a greater Electoral College bias than even she had. In fact, in 2020, Biden likely would’ve had to win the popular vote by around 4 points to prevail in the Electoral College. He ended up winning by 4.5 points, making 2020 a somewhat close call in terms of avoiding a split verdict.
So, should we expect 2020 levels of Electoral College bias against Democrats in 2024? I think the answer is probably not, based on polling and what we know about the ways that the party coalitions seem to be changing.
I’ll walk through this with a very simplified example, based on what national polling is telling us about the potential for racial depolarization. First, assume that this fall, relative to 2020, Biden retains roughly the same level of support with white voters, both college and non-college educated. Or, if you prefer, assume that his gains with the former generally cancel out his losses among the latter.
Next, let’s say that 5% of non-white voters who voted for Biden in 2020 switch to Trump for 2024. This would mean that Biden does about 10 points worse among non-whites nationally. Let’s also assume that this 10-point decline is uniform across racial groups, and that it occurs among both non-whites with a college degree, as well as those without one. I’d argue that this scenario, simple as it might be, is somewhere in the realm of possibility, and offers a starting point for thinking about one possibility for this fall.
Assuming that the electorate of the 2024 election is roughly 70% white and 30% non-white at the national level, which is close to where it stood in 2020, Biden would lose about 3 points in the popular vote if he did 10 points worse among non-white voters. So, he would only win the popular vote by about 1.5 points, a little less than what Clinton had won by in 2016. Yet, that 3-point swing against Biden in the popualr vote would manifest very differently state by state.
Just a quick note: I’m going to assume that Biden’s 10-point loss among non-white voters is entirely to Trump’s benefit, with no gains to third parties. I wouldn’t say that’s the most likely outcome this fall (we have plenty of reason to believe that third-party voting will be high in 2024, as it had been in 2016), but assuming that all of those non-white voters shift from Biden to Trump helps reveal the full potential magnitude of the defection from Biden.
Let’s now group our seven swing states into two straightforward categories. The first category is the diverse Sunbelt states of Arizona, Georgia, Nevada, and North Carolina, all of which have electorates that are roughly 70% white and 30% non-white, similar to the country as a whole. As such, the 3-point swing to Trump that occurs nationally shows up as a 3-point gain for him in each of these states. This means that besides North Carolina, which Trump won in 2020, the other three would flip from blue to red, including Nevada, which Biden won by just 2.4 points in 2020.
Our other group is the northern swing states: Michigan, Pennsylvania, and Wisconsin. All three have electorates that are roughly 80% white and 20% non-white. The 3-point national shift away from Biden only shows up in these states as a 2-point swing in Trump’s favor. That would mean two of the three states flip from blue to red, with a winning margin for Trump of 0.8 points in Pennsylvania and 1.4 points in Wisconsin. Michigan would remain Democratic by a margin of slightly under a point.
Again, this is an overly simple hypothetical, but it helps show how a 10-point loss among non-white voters, a reasonable estimate based on recent polling, would almost certainly be fatal for Biden in the Electoral College. But, compared to Clinton in 2016, Biden in this scenario did manage to win Michigan, despite doing worse in the popular vote than she did. He also lost Pennsylvania by nearly the same margin that Clinton did, once again, despite doing worse than her in the popular vote.
As an aside, it does seem that the Wisconsin outcome in this hypothetical, which would suggest a Trump win of 1.4 points, is slightly off. The state’s electorate is actually a bit whiter than the 80% white and 20% non-white figure that I used for the three northern swing states, which would mean a smaller Trump win, perhaps closer to 1 point, or even slightly less. I mention this because in my scenario, Wisconsin would potentially be the critical state that determines who wins the Electoral College (for Biden, he would have to also be winning Pennsylvania, while Trump’s AZ/GA/NC/NV combo puts him just two electoral votes short of 270). It wouldn’t exactly be a novelty for Wisconsin to be the decisive state, and despite only having 10 electoral votes, it’s a likely candidate to be 2024’s tipping-point state yet again.
So, given how close Biden comes in this scenario to winning the three northern swing states, we could imagine that if Biden had only experienced a decline of 5 rather than 10 points among non-white voters, he might have just barely won MI, PA, and WI, and thus the election. This particular scenario suggests that if Biden were to win the popular vote by around 3 points (conditional on matching his 2020 levels of support among white voters), he could win exactly 270 electoral votes. This would stand in contrast to 2020, when Biden probably needed a 4-point popular vote win to clinch the Electoral College.
Just to give a quick example of how Biden’s weakness with non-white voters could show up in an even more dramatic fashion than the four diverse Sunbelt states mentioned, consider California. It’s the second-most diverse state in the country (Hawaii is first), and polls right now are showing Biden ahead in California by only 20 or so points. That’s down from Clinton’s 30-point win there in 2016, and Biden’s own 29-point win in 2020. If Biden indeed does about 10 points worse in California than he had four years ago, that alone would shave nearly a full point off Biden’s national popular vote margin (California casts around 11% of the national popular vote). Of course, a major swing against Biden in the nation’s largest state doesn’t change the Electoral College calculus in the slightest, even though it would have a noticeable effect on the popular vote.
In short, there’s reason to believe that the disadvantage Democrats face in the Electoral College will be smaller this year than in 2020, a reflection of the fact that Biden’s potential struggle with non-white voters won’t hurt him quite as much in the northern swing states as it does elsewhere. It seems very plausible that the Electoral College bias could end up somewhere between its 2016 and 2020 level. Whether you consider this good or bad news, I’ll leave up to you, but if nothing else, it’s a reminder of how fickle the Electoral College can be.
One final question that’s worth thinking about, and which has implications not only for electoral politics, but beyond: will Biden’s erosion among non-white voters be uniform, as I assumed above, or will it differ from one racial group to another? The non-white vote is far from monolithic, and it’s possible that Biden will lose support across multiple racial groups, but to very different degrees. For instance, erosion with Hispanic voters will quickly show up in Nevada and Arizona, but not to a major extent in any other core swing state. On the other hand, if Biden’s decline among non-whites is more pronounced with African-American voters than with other groups, it will matter profoundly in Georgia, and to a lesser (but still important) extent in Michigan and Pennsylvania.
My example above has plenty of caveats, but it hopefully makes it easier to see how the potential 2024 coalitions may affect the gap between the popular vote and the Electoral College. Biden could possibly lose a little over a point in the popular vote (subject to a bunch of assumptions) before his position in the Electoral College is jeopardized. It also suggests that based on polling right now, Biden’s best chance of winning the Electoral College might run through Michigan, Pennsylvania, and Wisconsin, which would result in a narrow 270-268 Biden victory, assuming that Trump won all of the Sunbelt swing states.
As I say pretty much every time, thanks for reading! I never would have guessed that I’d find an audience on Substack, so for those of you who’ve stuck with me, it really does mean a lot. I hope you’re finding these posts interesting, and most likely, I’ll be back in early May, when I plan to look at all of the swing states again, review the latest polling, and share any new state ratings that I might have. Until then, take care, and happy Easter weekend for those who celebrate!
Thanks for the update. This will be a white-knuckle election cycle, and I appreciate the attempts to examine the big picture, not just the current polling.
I would say that a 10-point (net) swing among voters of color seems quite large. I am well aware that we're seeing a shift but even a 6-point (net) swing would be on the larger side of what we've seen over the past 3 elections. That doesn't mean 10 points is impossible, just that I would be surprised if it was that large.