Outside the Beltway

Crystal Ball Moves Trumpward

It all comes down to Wisconsin, Pennsylvania, and Michigan.

James Joyner · Friday, June 14, 2024 · 20 comments

Two weeks after former President Donald Trump was convicted on 34 felony counts, with constant reporting of his increasing incoherence, one might think he would be dropping in the polls and becoming less likely to get re-elected. Yet, the race remains remarkably stable.

Here are the trendlines in the FiveThirtyEight polling aggregate:

If you prefer RealClearPolitics:

That one shows more volatility, but it is mostly a function of a compressed scale magnifying tiny fluctuation.

University of Virginia political scientist Larry Sabato’s long-running Crystal Ball is even weirder:

Today we’re making a half-dozen changes to our Electoral College ratings, all of them benefiting the Republicans. These moves don’t significantly change our overall outlook, which is that we don’t really see a clear favorite in the presidential race, but they do better align our ratings with that overall outlook.

Here’s their revised map:

To the extent anything is changing in a race that seems extremely unchangeable, one would think it would be in the opposite direction. The news has been all bad for Trump, with no comparable developments on Biden’s side.

Their underlying assumptions strike me as reasonable, mostly because they match my own.

The 2016 and the 2020 results act as something of a Rosetta Stone for deciphering 2024. That does not mean there won’t be shifts from those elections—of course there will be. But when confronted with polling results that differ wildly from what we saw in both of those elections—like, for instance, polls showing a tied race in our home state of Virginia after Joe Biden won it by 10 points in 2020, while polls in other places do not consistently show such a strong shift against the president—we tend to defer to the actual past results. There doesn’t have to be, and likely won’t be, a perfectly “uniform” swing from 2020’s results to 2024. But we do think some basic patterns will endure—Virginia voting more Democratic than the nation is one of them.

Given that Trump is the constant in 2016, 2020, and 2024, it seems perfectly reasonable to assume that states that voted for or against him both times will make it a trifecta. The toss-ups are, therefore, the states that went one way in 2016 and another in 2020.

Relatedly, it’s still too early to be using polls to make dramatic claims about how states will vote. Polls are often described as a “snapshot in time,” and while they tend to be used as a prospective measure (projecting forward to the election), they actually are retrospective instruments, as they measure attitudes that existed whenever the polls were fielded. To be clear, most of the voting public is immovable, but the key voters that will decide the election are movable, and they may shift in and out of voting for one of the major party nominees, a third party option, or skipping the vote altogether. So there’s some volatility here. Our general assumption is that Biden is going to perform at least a little better in November than polls are showing now, much like Donald Trump generally performed better in November of both of his election years than what late spring polling suggested. Biden probably has a little bit more base consolidation to do than Trump—we may actually be seeing some of that in the wake of Trump’s conviction on business record falsification charges in New York a couple of weeks ago. To be clear, that doesn’t make Biden a favorite in our eyes—again, we just don’t see a favorite.

I would add that Biden’s base is larger and that he has a functioning party apparatus and Trump intentionally destroyed his.

All that said, we also recognize the clear big-picture trends. Trump has been polling better than he typically polled in both 2016 and 2020, and that has been the case for many months. Biden’s approval rating is in a dangerous zone—the high 30s—and he has been in that weak place consistently since November, according to the FiveThirtyEight average. Biden is not going to be at net-positive approval by Election Day—fortunately for him, he does not need to be, but one would probably expect to see some level of improvement if he is going to win reelection. The danger for Biden is that voters may just be done with him: There is some nostalgia in polls for the pre-2020, pre-Covid, and pre-inflation period that coincided with Trump’s presidency. That doesn’t necessarily mean the public is clamoring for Trump, who remains unpopular; it’s just that they may prefer him to Biden, or may just be thinking more about what they don’t like about Biden (the incumbent) than Trump (the challenger). One thing that Biden has going for him is that Trump does not seem to have trimmed the sails on his own rhetoric at all—Trump continues to laud the Jan. 6, 2021 rioters who tried to disrupt the 2020 electoral vote count as persecuted patriots, for instance, a position we just can’t imagine helps him with the middle of the electorate trying to decide between two flawed major party candidates.

I remain baffled by public opinion on Trump. He’s much more intensely disliked than Biden and yet people seem to be resigned to his massive shortcomings. Regardless, to the extent polling remains useful at all, it’s undeniable that Trump has outperformed Biden throughout the long cycle.

Regardless, given that the national polls are so stable, why seven adjustments? And why all in the Trump direction?

Georgia, by margin the Democrats’ narrowest presidential victory in 2020 (although Arizona’s margin was only very slightly higher), goes from Toss-up to Leans Republican. This nudges the GOP total to 251 votes at least leaning to them, 19 short of the magic number of 270. One could argue that all three of Arizona, Nevada, and Georgia could or should be Leans Republican. We think that’s premature, particularly Nevada, where polls often overstate Republicans and Democrats have a proven ground game operation. It’s splitting hairs to some degree, but Georgia was probably the most surprising of Biden’s 2020 victories, and it may be a little bit more Republican at its core than the other two are: Democrats hold all of the Senate seats from these three states, but Democrats were shut out in statewide executive office elections in both 2018 and 2022 in Georgia, which was not the case in Arizona and Nevada. There also has been publicly reported pessimism from Democrats about Georgia for many months. Georgia is the only one of the seven states decided by 3 points or less in 2020 that does not have a high-profile Senate or gubernatorial election this year: Arizona, Michigan, Nevada, Pennsylvania, and Wisconsin have Senate elections and North Carolina has a gubernatorial race. There has been some suggestion that this could hurt Democratic turnout in Georgia, but that is not really factoring into our thinking: Presidential races, to us, are what drive turnout. Let’s remember that “Leans” does not mean “Safe.” Georgia still remains one of the key swing states even as we see a little bit of a GOP edge there for November.

I, too, would put Georgia in Trump’s column. Like most of the Deep South, it was a Democratic stronghold for generations after the Civil War. It was one of the few states to go Republican in the disastrous 1964 election, went for Wallace in 1968, and for Nixon in 1972. Native son Jimmy Carter won in 1976 and 1980 and Southerner Bill Clinton won in 1992 (by a slim 43.5-42.9% margin). Trump carried the state 50.8-45.6 in 2016 and just barely lost (49.3-49.5) in 2020. If Biden carries it again, he’ll sweep the swing states.

Pennsylvania, the native state of President Biden and one of the states that helped nudge Donald Trump over the finish line in 2016 before flipping back blue in 2020, goes from Leans Democratic to Toss-up. This reduces, from 260 to 241, the number of electoral votes at least leaning to Democrats in our ratings. We also are making a concurrent move in the Senate race there, moving Sen. Bob Casey (D-PA) from Likely Democratic to Leans Democratic. We think Casey will likely do better than Biden, but not overwhelmingly better, so it makes sense to make these moves in tandem (we’ll say more about this race and the overall Senate picture in a future issue). Pennsylvania moves to Toss-up but we are keeping Michigan as Leans Democratic; Michigan is generally more Democratic than Pennsylvania in presidential elections (that has been the case in each of the last seven elections). It is true that Michigan has a notable Arab-American population that likely will shift against Biden to perhaps a large degree in protest of how Biden has handled the situation in Gaza, although this is not a huge segment of the population: The Democratic margins in three core towns in the Detroit area with sizable Arab-American populations (Dearborn, East Dearborn, and Hamtramck) accounted for about a fifth of Biden’s overall 2020 statewide winning margin (about 30,000 of a roughly 150,000 raw vote statewide win). Any slippage is of course important, but we also think the magnitude of the impact of that likely slippage was probably overstated in the coverage of the protest vote against Biden in the Michigan primary back in late February. One feature Pennsylvania has that Michigan doesn’t is that the former is partially covered by the Appalachian region, which has been moving against the Democrats for at least a couple of decades at this point. Many of the counties in the Keystone State where Trump’s margin improved from 2016 to 2020 were in the Appalachian western/northern parts of the state in areas further away from Pittsburgh, and that very well may continue in 2024. We do see the potential for Biden to improve in the Philadelphia suburbs (although Philadelphia itself got a little less blue in 2020 and could move a little further in the same direction in 2024). The bottom line here is that we still expect Michigan to be a bit bluer than the other pieces of the so-called “Blue Wall,” Pennsylvania and Wisconsin. Hence the difference in ratings (more on the importance of these states momentarily).

Pennsylvania is an enigma to me. It went Republican in every election from 1860 to 1932, with the exception 1912, when it went for Teddy Roosevelt (who was of course actually a Republican running as a third party candidate). Since then it voted Democrat three elections in a row, Republican three elections in a row, Democrat three elections in a row, Republican, Democrat, Republican three elections in a row, Democrat six elections in a row, Trump, and then Biden. Damned if I can make any sense of that.

The single electoral vote in Maine’s 2nd Congressional District goes from Leans Republican to Likely Republican. Trump won this single electoral vote by about 6 points under the current lines, and it is the kind of white working-class area that is hard to see swinging strongly to Biden in this election.

Fair enough.

The states of Alaska, Iowa, and Ohio move from Likely Republican to Safe Republican. They would only be gettable for Democrats in a perfectly optimal situation in a presidential race, and even if Democrats hold the White House, it doesn’t seem like an optimal situation, politically, is forthcoming for them.

Alaska is about as safely Republican as it gets. And Iowa and Ohio have been much less swingy of late. Trump won both states easlly in both 2016 and 2020; he’ll do so again.

But, despite making multiple moves in the Trumpward direction, the key takeaway is that the toss-up states remain tossups, making the outcome unpredictable.

So the difference now is that whereas the Democrats were closer to the magic number of 270 before, Republicans are now a bit closer. But neither side is at or over 270.

If Republicans do in fact win all of the electoral votes at least leaning to them in our ratings, 251, the 19 electoral votes in Pennsylvania would put them at exactly 270. Taylor Budowich, the CEO of a pro-Trump Super PAC, recently wrote in a strategy memo (as reported by CNN and other media outlets) that winning Pennsylvania “is the ball game.”

That’s not an outlandish argument—in fact, we’d probably go a little further: If Trump wins any of Michigan, Pennsylvania, or Wisconsin, he very likely will win the election.

That’s a scary thought but likely right. They’re all states Trump won by the narrowest of margins in 2016 and lost in 2020. Michigan was 47.5-47.3 in 2016 and 47.8-50.6 in 2020; Pennsylvania 48.2-47.5/48.8-50.0; and Wisconsin 47.2-46.5/48.8-49.5. Notably, though, all three voted Democrat in every contest since 1992 (1988 in Wisconsin’s case) except 2016. So, if forced to bet, one would think 2016 the outlier.

Yes, one can come up with scenarios where Biden hangs on in which he carries just two of the three—Michigan and Pennsylvania—and replaces Wisconsin with Arizona, for instance. But we don’t see such a scenario as all that plausible—the historically competitive Industrial North has been a crucial part of recent winning Democratic presidential coalitions (the last Democratic winner to lose even one of these three states was Jimmy Carter in 1976). Keep in mind that if Biden loses Arizona, Georgia, and Nevada from his 2020 map, he could still get 270 electoral votes on the nose if he holds everything else he won four years ago, including the single electoral vote in the Omaha-based 2nd Congressional District of Nebraska. Under the current lines, Biden carried that district by 6 points; in a world where Biden is holding Michigan, Pennsylvania, and Wisconsin, we’d have to imagine he is also holding NE-2, which was several points bluer than those states four years ago and has a fairly high four-year college attainment rate (we wouldn’t expect such a place to strongly shift back red while the more competitive Industrial North is standing still). So if Biden is losing NE-2, he’s probably in bigger trouble elsewhere (the same would be true if Biden lost some other places that are on the periphery of the competitive map but were bluer than the nation in 2020, like Minnesota, New Hampshire, and Virginia).

If Biden loses any of Minnesota, New Hampshire, and Virginia it’ll be a rout. But I just don’t see that happening.

They conclude:

We are entering a busy period of the presidential election calendar. The first debate (if it happens) is scheduled for Thursday, June 27, much earlier than usual. The Republican National Convention kicks off in Milwaukee on July 15; Trump’s sentencing in the New York case is scheduled for July 11, and Trump’s eventual VP pick will likely come around this time as well. The Democratic National Convention starts on Aug. 19. The time period after the conventions will be a good time to reassess. We suspect the election will still seem like a Toss-up at that point, but let’s wait and see.

It’s just crazy to me that there’s going to be a debate in less than two weeks. Have we ever had a presidential debate before the nominating conventions? Given the degree to which the candidates’ age and speaking ability are under scrutiny, it could actually make a big impact on the race.