(I got knocked for a loop by Covid at the end of August, and I am still dealing with some post-Covid fatigue. So, apologies if this is all a bit scattered, but a few thoughts on the state of the race. Hopefully what follows will help you manage your sanity, if not your anxiety, over these final weeks of the campaign).
Harris has a small national lead and the battleground states - seven this year, including North Carolina, are polling at razor thin margins. There have been lots of complaints about polling in recent years. Some are valid, some less so. The reality in a close race is that polls can only tell us so much. Indeed, we "know" the race is close because of polling. But whether Harris ends up winning the national vote by four points or two points, that's all within the margin of normal polling error and the same is obviously true of the battleground states that, because of our absurd system, ultimately decide who wins the presidency.
In the meantime, there's no good reason to give more weight to one of the major poll aggregators over another. Nate Silver shows Harris with a relatively slender 2.2 percentage point national lead for Harris and his model gives Trump slightly better than a 50% chance of winning the electoral college, as of yesterday. As of this writing, The Economist aggregator is at 4.5, 538 aggregator is 3.0 (and gives Kamala a 61% chance of winning the election) and The Hill is 3.7. If you average all of these, you get something like Harris at plus 3.3. Based on recent election outcomes, the assumption is that Harris needs about a three point national margin (again, because of our preposterous system for electing presidents), to have the clearance she needs to win enough battleground states to capture 270+ electoral votes.
As a reminder, in 2020, Biden won the national vote by 4.5%, but eked out victories by less than a percentage point in Wisconsin, Georgia and Arizona and barely over that in PA. Had the under-one-percent states gone to Trump, the electoral college would have been tied at 269-269, which would have flipped the election to the House and resulted in a Trump victory.
But that assumption about the national vote margin Harris needs all amounts to a supposition, a correlation based on very recent events. It's not a scientific law. It could be that Harris ends up winning the popular vote by two points, and also squeaking by, say, in Wisconsin, Michigan, PA and NC, which will get her over the top. We just don't know, and we can't know.
It's been much commented upon that Trump outperformed the polls in both 2016 and 2020 and that this is the result of a "shy" Trump voter who doesn't want to talk to pollsters and is not otherwise motivated to vote, but will come out for Trump. Will that happen again in 2024? Hang on a sec.
To back up, because the difference in 2016 resulted in a Trump win, whereas in 2020 it did not, much more attention has focused on the polling misses in the former than the latter. But the polls missed by substantially more in 2020 than in 2016. 538's final aggregator in 2020 had Biden winning the popular vote by 8.4%, meaning there was a nearly four percentage point miss. In 2016, 538's final aggregator had Clinton winning the popular vote by 3.6 percentage points. She won by 2.1%. In other words, the popular vote miss was twice as bad in 2020.
Does that mean polls are underestimating Trump's vote share in 2024? Maybe, maybe not. Pollsters are constantly updating their models and assumptions. Most of them really do want to get it right, for very obvious reasons. In 2012, pollsters underestimated Obama's final vote share by about two percentage points. Since 2022, Democrats have tended to outperform polls, both in the 2022 elections and in special and other elections since. Is there a post-Roe effect that will swing the electorate toward Harris? Maybe. We know there will be some polling error. We don't know in which direction. Maybe Harris’ 3 percentage point lead in the polls is underestimating her support by two percentage points. If that’s true, obviously the final outcome looks very different than the toss-up the race appears to be.
There's been perhaps more liberal criticism of mainstream media political coverage this year than ever before. That's partly a reflection of the challenges of covering Trump, a pathological liar who has taken advantage over the past decade of the norms of "balanced" journalism, combined with the stakes of this election. Margaret Sullivan and Jay Rosen are two of the sharpest observers of the problems with the conventions of political journalism in the Trump era. I'm sympathetic to much of this criticism, but a couple of recent pieces of data have reminded me, or at least suggest, that such criticism can overstate the impact of media coverage.
To illustrate, here’s a recent polling nugget from CBS.
I could show you others, but by and large after nine years, Americans know who Trump is. And overwhelmingly, on a personal level, they don't like him. In other words, the premise of the criticism is that if only people knew X, they would be less likely to vote for him. I don’t believe that is true to the extent that I once did. To be clear, there is no question that Americans are misinformed about many aspects of our politics, including where Trump stands on issues and the impact of his proposed policies on their lives.
But it can both be true that many Americans believe things that are patently untrue - like the preposterous and nauseating blood libel about Haitian immigrants in Springfield eating pets that Trump/Vance are quadrupling down on on the one hand, and on the other, that a better information environment might not have the impact on our elections that we might wish for. By and large, people vote for instinctive or psychological reasons, not because of a carefully considered balancing of facts and issue positions.
We're not super computers, assigning discrete weights to the economy, immigration, abortion, Gaza, Ukraine, etc and then calculating which candidate more closely matches our weighted issue preferences. We have values, we have our instincts about how the world should work, we have snippets of information and we try to reconcile all those things into a narrative that helps us explain why we're going to vote for x and not y.
If voters tell pollsters something that is untrue to justify their vote for one candidate for another, it doesn't follow that if they knew better, they'd change their mind. The reality is that people are more apt to reconcile their views on issues to their underlying partisan leanings and instincts than vice versa. That's not true for everybody, or to the same degree. But it's *truer* than the belief that the average voter dispassionately evaluates party positions on key issues and votes accordingly. That fact adds to the frustrations and challenges of trying to figure out how to appeal to those swing voters (or undecideds, or split ticket voters - choose your terminological adventure!) who do decide elections. As polarized as we are nowadays, and there are many fewer split ticket or swingy voters than there used to be, they still exist. Joe Biden won Pennsylvania by just over one percent in 2020. In 2022, Democrat Josh Shapiro won the governorship in a blow out, beating the wacky Doug Mastriano by nearly fifteen points). In other words, there were a non-trivial number of Trump/Shapiro voters in PA.
Likewise, in North Carolina this year, Democrat Josh Stein is likely to beat Republican Mark Robinson handily, in a state that has leaned Republican and that Trump won by just over a percentage point in 2020. That's in large measure because Robinson is an unhinged extremist. Of course, examples, exist on the other side. Ron DeSantis barely squeaked out the governor's race in Florida in 2018, beating Andrew Gillum by less than .5%. In 2022, he won reelection, in an absolute blowout, beating former Governor Charlie Crist by nearly 20 points. In other words, there were plenty of Biden/DeSantis voters in 2020/2022.
All of which is to say that split ticket and genuine undecided voters still exist. Many say they need more information, though this should not be taken literally. They may say they want to know more about a candidate's position on X, but ultimately they're more likely to decide whom to vote for based on harder-to-quantify feelings of trust, or who the people they listen to in their lives are going to vote for, etc.
All of this, to repeat, creates basic challenges for campaigning, messaging and resource deployment. It also means that any one explanation for why a campaign is doing better or worse is likely to be an inadequate one. Over 160 million people are likely to vote this year. We can’t know for sure who is going to show up. If pollsters knew exactly who was going to show up, they could reasonably estimate how 90% are going to vote. That remaining ten percent - they’ll split their votes, in proportions we can only guess at.
None of this is meant to induce despair. As I said to my buddy JH yesterday, you’d rather be in Harris’ position than Trump’s at this moment. But there is no good reason to assume that this is anything other than a coin flip election. Which makes this a good time to recount Michelle Obama’s sage advice at the DNC in August:
So no matter how good we feel tonight or tomorrow or the next day, this is going to be an uphill battle. So folks, we cannot be our own worst enemies. No. See, because the minute something goes wrong, the minute a lie takes hold, folks, we cannot start wringing our hands. We cannot get a Goldilocks complex about whether everything is just right. And we cannot indulge our anxieties about whether this country will elect someone like Kamala instead of doing everything we can to get someone like Kamala elected.
And as MO also repeated: “Do something!”
Thank you Jonathan. Your careful and scrupulous analysis is much appreciated. Hope you are mending up from COVID okay.