Can Kamala Harris win in Incumbent Mode?
Challenger Mode has been working for her. But running as the frontrunner can be harder.
I had a little mental misfire today which — bear with me here — I think may be revealing of one of the risks Kamala Harris faces as Democrats hold their convention next week.
I always check the output from the election model before we publish an update. The reason for this is simple: sometimes we make mistakes when inputting data — say, double-entering a poll, or attributing it to the wrong state — and I’ve been doing this for long enough that I usually have a keen sense for when something “doesn’t look right” and there has been a data-entry issue. Today, there were no such surprises — Donald Trump has had a comparatively good couple of days in the polls, so I was expecting Harris’s win probability to decline and it did, to a 54 percent chance of winning the Electoral College from 56 percent on Thursday and 57 percent on Wednesday. It’s not a huge swing and within the range where it could easily be statistical noise. Still, Harris’s lead in our national polling average has declined from a peak of 3.1 points to 2.4 points now.
The timing of the convention is probably fortunate for Harris, who has been coasting on good vibes since she took over for Joe Biden. You could easily imagine the media overemphasizing tightening polls, or just getting bored of the “Harris rising” storyline, and applying more scrutiny to Harris for things like her lack of press appearances. Usually, I don’t think the tone of media coverage matters as much as other people do — in fact, I think partisans on both sides have an overly simplistic view of the press and often mistake cause for effect.1 But since the Harris campaign seems eager to ride the momentum wave, it might matter more than usual.
Conventions, however, typically produce a short-term bounce in the polls, which could serve as a booster dose of momentum for Harris and Tim Walz.
In fact, our model explicitly accounts for this by applying a “convention bounce adjustment”. It gives a little bit of a haircut to the polling of whichever party just held its convention. After the DNC, it will expect Harris to gain slightly in the polling. Specifically, it expects bounces to peak in the range of 2 or 2.5 points immediately after the conventions, and then fade to zero over the course of about three weeks. Since Harris leads by 2.4 points now, that means it will expect national polls right after the DNC to show her ahead by 4 or 5 points. That’s the point spread, in essence. If she’s still only ahead by 2 or 3 points in what should be one of her strongest polling periods — or certainly, if her numbers decline — her odds will start to fade in our forecast. If she’s up 6 points or something, they’ll improve. If it’s right in the range of 4 or 5 — declining to 2 or 3 after a few weeks like the model expects — we’ll maintain the status quo.
Now about that mental mistake I made. Just as the model will adjust Harris’s polls downward after the DNC, it was adjusting Trump’s downward after the RNC. So I was checking to see what residual effects there still were from the adjustment the model was performing to Trump’s polling. And there weren’t very many. In Michigan, for instance, the model is boosting the incumbent party’s vote share by about 0.1 points as a result of the RNC adjustment, and reducing the challenging party’s vote share by about 0.2 points. So a net of 0.3 points against Trump, which seems reasonable. Although we’re out of the immediate post-RNC period, there are still some older Michigan polls that get a little bit of weight in the average. So the RNC adjustment is almost all gone, but not quite. Good — that’s exactly how the model is intended to work.
Except, at first, I thought it was broken. Because in the model, the candidates aren’t labeled “Harris” and “Trump” or even “Democrat” and “Republican” but rather as “challenger” and “incumbent”. And when I saw the model raising the incumbent candidate’s vote share in Michigan and lowering the challenger’s, I thought that was weird. Why was the model boosting Trump’s numbers and subtracting from Harris’s after the Republican convention, when it was supposed to do the opposite?
Then I realized: it’s because I’d transposed the “incumbent” and the “challenger” in my head: Harris feels like the challenger and Trump feels like the incumbent when it’s the other way around.2 Trump feels old and Harris feels new. Trump feels like the favorite and Harris feels like the underdog, even though she no longer is in our model.
Look, maybe it’s just me. It’s been a fun but very long week of media hits and now going on the road for On the Edge. Still, I wonder if other voters aren’t making the same transposition, thinking of Harris as the challenger.
If so, that’s probably helping her, because contrary to what the conventional wisdom holds, incumbency isn’t much of an advantage in presidential elections anymore. In fact, it may well be a disadvantage, considering how poorly incumbents have performed in elections around the world — and how unpopular Biden is.
This may be a tricky dynamic for Harris to navigate. On the one hand, she’s about to enter Incumbent Mode. She’ll want to appear “presidential”, especially given any lingering sexism voters have about a woman in the White House. And at the DNC, Democrats will presumably want to tout the Biden-Harris administration’s accomplishments — and Biden himself is speaking, although on Monday, not coincidentally the day that usually gets the worst TV ratings. On the other hand, Trump has run in three elections in a row, Biden remains toxic politically, and the mood of the country is dour. So Challenger Mode has had a lot of appeal for Harris. But the novelty of it may wear off, especially if she continues to lead in polls and she’s seen as the frontrunner, which could lead to tougher media coverage.
I’m not a messaging-and-tactics guy, but strategically, Harris is probably going to have to pull off a couple of different looks: gravitas one day, fun the next day. There’s a risk of the memes getting stale, of Harris not shifting gears until she’s faced a more serious setback, like a bad performance in the first debate.
In the short run, she should be fine, and I’d still probably rather have her hand to play than Trump’s. But the last 48 hours of polling are a reminder that the road to the White House isn’t going to be a cakewalk.
That is, they think the candidate is rising in the polls because of favorable media coverage, when it’s the other way around.
To be precise, of course, Harris is the incumbent party’s candidate but not the incumbent president.
The question j have is whether Harris got her convention bump early in the unusual conditions by which she became the nominee.
It seems to me that we are in uncharted territory. Trump is, in some ways, more of an incumbent than Harris. He has a whole presidential term to defend, while Harris remains a cypher. It is only recently that the press (and others) have been calling it the Biden-Harris administration, rather than just the Biden administration. It remains to be seen if, on Election Day, the polls have any connection with the reality on the ground. (I'm not suggesting that they are skewed in any particular direction, just that they may be off more than usual.)