Frank Bruni, a contributing Opinion writer, hosted a written online conversation on Friday, Nov. 1, with Kristen Soltis Anderson, a contributing Opinion writer and Republican pollster, and Nate Silver, the author of “On the Edge: The Art of Risking Everything” and the newsletter Silver Bulletin, to discuss polling and politics in the final days of the race between Kamala Harris and Donald Trump. Their conversation has been edited for length and clarity.
Frank Bruni: Kristen, Nate, thank you for joining me. Election Day is almost here, and still it seems that nobody really knows anything. Democrats fret — that’s their nature. MAGA Republicans strut, emulating their idol. And you two? Nate, because you recently wrote a widely read guest essay for us about your gut auguring a victory by Donald Trump, I want to check on your gut now. Guts change. Has yours?
Nate Silver: Well, the whole point of that article — including the headline — was that I don’t think my gut is worth anything in this case. Many people interpreted it differently, as though I was revealing my super-duper secret real prediction. And I expected that. But your gut feeling a week before the election will mostly be an emotional response or picking up on the vibes through osmosis — Republicans are invariably more confident so that seeps through — and I don’t think either of those things will help you make a better prediction.
Kristen Soltis Anderson: Whenever people ask me who I think will win, and I steadfastly refuse to tell them — because I don’t feel confident we know how this is going to go given the available data — people get very frustrated. It is understandable, we all want certainty. People want to mentally or emotionally prepare for an outcome. They don’t like surprises. And I’m sorry to say, you should continue to mentally and emotionally prepare for a wide range of outcomes.
Silver: For what it’s worth, though, the models have been pretty momentumless lately, whereas Harris had been falling in mid-October. Prediction markets are also shifting more in line with the models, showing a 50-50-ish race. Maybe it’s dawning on people that this is an uncertain and close race.
Bruni: Nate, you bring up that dread 50-50 number. Kristen, you mention frustration. I want to talk about the frustration of this pesky “tie” word. I distrust it. I abhor it. A tie is very, very unsatisfying — and how can it really be? As narrative, the Harris-Trump face-off doesn’t play: Chapter 1, it’s a tie. Chapter 5, a tie. Chapter 10 … a tie! Is that truly possible? Please illuminate. Not just for me but also for many readers like me, please administer some tie therapy — including, if possible, your thoughts on how likely the result could be far from a tie?
Silver: There’s something like a 40 percent chance in our model that one candidate wins all seven battleground states. Since polling errors tend to be highly correlated — if Trump beats his polls in Michigan, he’ll likely also do so in Pennsylvania — even a minor polling error could lead to a relatively decisive result in the Electoral College (whether or not Trump will admit it if he loses, which he almost certainly won’t).
But people want models to be oracular and offer a very precise prediction. And sometimes their value instead is in emphasizing the uncertainty in the contest.
Anderson: Nate’s model is on target. A 60 percent chance that the race is truly fairly close and we are counting well into Wednesday and beyond, but a 20 percent chance on either side that this will be wrapped up quickly with one candidate taking the battlegrounds in short order sounds about right. And that’s what people fail to understand when we throw around the word “tie.” They think it means we are confident it will be a very close race. They think there is certainty that this is neck-and-neck. Pollsters have been trying to be better about communicating about uncertainty this year, but I fear it hasn’t broken through.
Silver: The other thing is that the conventional wisdom tends to exaggerate even small polling swings — we went from 55-45 in Harris’s direction to 55-45 for Trump, and people are treating it as though we’ve gone from likely Harris to Trump being a sure thing. The narrative about Trump having momentum was sort of vaguely, loosely true — he’s in a better position than immediately after the debate, for instance — but also grossly exaggerated. Still, Democrats aren’t used to this. In 2008, 2012, 2016 and 2020, they went into Election Day as the clear favorite — wrongly in 2016, of course.
Bruni: I know you two focus on numbers, evidence, what’s measurable, but please tap your wealth of political experience, from close observation over so many years: What one factor, among the many factors in the foreground of this race, do you think could turn out to be much more consequential than we’ve imagined? Venture a best guess, an informed hypothesis. This will be my personal tie therapy from you.
Silver: Well, it’s Trump who’s relying more on marginal, unlikely voters, while Democrats do well with the upscale suburbanites. Those voters can be hard to capture in polls.
Anderson: When the polls have missed in the last few elections, the consensus is that it is because of polls missing some kind of unlikely voter. In 2012, this was a driver of some pollsters’ overestimation of Mitt Romney’s chances. In 2016 and 2020, it was Trump’s unlikely voters who were missed.
Silver: They also might not show up, especially sort of the Theo Von-Joe Rogan younger male demographic. And Harris should have the better ground game, with Trump sort of having turned his over to Elon Musk. It’s probably the difference of at most half a percentage point, but the election could come down to that.
Anderson: When pollsters make decisions about their samples and what weights to apply, they are making an educated guess about what turnout will look like, but it is just that — an educated guess.
Bruni: Many of the people in my life keep telling me that Harris will win because Trump has revealed such dark colors so very obviously that he won’t, all said, be able to get enough voters to risk him again. I say to them: That sounds like the 2016 confidence that he’ll lose in a new outfit. But could they be right? Could it really come down to that: Trump seeming — to a decisive cohort of voters — an act of recklessness and nihilism too far?
Silver: It’s hard when Democrats have had the same message — this is the most important election of your lifetime, so vote for us, or democracy gets it — three elections in a row. You’ll have people voting on Tuesday who were 9 years old when Trump first descended the escalator at Trump Tower.
Bruni: That damned escalator — I’ve never had such issues with a device for human transport.
Silver: He was president before, and although 2020 was pretty hellacious with the pandemic and then widespread social unrest, including Jan. 6, people don’t seem to pin the blame on him — they sort of conflate the timeline and associate Biden with the lockdowns.
Anderson: When Trump left office, Gallup found him leaving with a 34 percent job approval — just abysmal. The disgust with Jan. 6 had left an awful taste in people’s mouths. But then when NBC asked people just a few weeks ago, the retrospective job approval of Trump was 48 percent.
Silver: When Biden-Harris say democracy is under threat, I absolutely agree with them. But they’re in the White House right now and Democrats control the Senate and many of the important swing-state governorships. So this is a tougher and more counterintuitive sell for Harris than it seems.
Anderson: The reality is that America was ready to turn the page from Trump in the last election, thinking that with Joe Biden they would get some kind of calm and normalcy and adults in the room. When that didn’t materialize as they’d hoped, suddenly Trump started looking more appealing again by comparison. That’s how you get someone with all of Trump’s controversies, dark rhetoric and baggage to still have 48 percent retrospective job approval.
Bruni: Because you’re talking about how voters process things, I want to say — I want your reaction to my belief — that the single greatest flaw in our analysis and predictions is assuming voters are armed with accurate information and paying a decent amount of attention. I’ve attended so many political rallies — events that theoretically attracted engaged voters — at which people had no clue about the basic facts of the candidates and the election. We talk of inadequate models: Isn’t the greatest inadequacy a search for rationality in an irrational pool?
Silver: Yeah, even the people who youthink would be relatively well-informed, like the Wall Street guys I play poker against, often see the world through stylized facts and can get captured by sentiment within their narrow social circles. (These are the guys who thought Michael Bloomberg would be an excellent presidential candidate.) And that’s to say nothing about a swing voter in Latrobe, Pa. It’s a problem for political analysis that the vote is increasingly divided along educational lines, and people who work in journalism and politics are by definition news addicts who largely went to highly selective colleges.
Anderson: This is also why, nearly every single time some kind of controversy happens that I get asked about on cable news — “will this move numbers?” — I’m almost never wrong when I say “probably not.” Most Americans, thankfully, are not on a constant I.V. drip of the latest political news. As one of my focus group respondents in my last Times group of Michigan voters put it, his friends who remain undecided voters are “like a cat chasing a laser pointer” rather than constantly ingesting political news and overthinking things.
Silver: What translates to the ground level is almost unknowable, I think, which is part of why I prefer to rely on polls rather than vibes.
Bruni: OK, Kristen, right now the part of annoying cable-television interrogator will be played by Frank Bruni. Do you believe that any developments over the past few weeks or past few days — Trump’s Madison Square Garden rally, say, or Biden’s “garbage” comment, or something else — could have a real effect on this perpetually tied election? Is there maybe a 2024 analogue to the 2016 James Comey intrusion or some Comey Lite incident? Again, I know you two are, honorably, data people, but still. Work with me there.
Anderson: The very strongly held views people have of Trump have anchored this race in place. There’s so little that will change how people feel about Trump because they’ve heard it all before and decided how they feel about it.
Silver: The Madison Square Garden rally was correlated with quite a lot of Google search traffic both for Trump and for the comedian Tony Hinchcliffe. So maybe it broke through just a little bit. But we’re talking about plausible polling movement of a tenth of a point or a quarter of a point, almost certainly not more than that. Even things like his criminal conviction and the assassination attempt against Trump didn’t move the polls very much — maybe by half a point to a point at most. Even Biden’s debate, possibly the most consequential debate of all time, only moved them by about a few points or so.
Anderson: The trouble with a race that appears close and unmoving is that nothing matters, but also everything matters. If it truly is close, and the “garbage” comments motivate 1,000 more marginal Trump voters to turn out who might have otherwise not bothered, or turn off 1,000 Puerto Rican voters in Latrobe, that’s a shift that would be imperceptible in a poll but could be decisive if this is close as it was in Florida in 2000 (where it came down to an unbelievably small number of voters).
Bruni: Nate, pollsters make a lot of choices and assumptions — both when they cast their net for respondents and when they interpret the results — about who will show up to vote. Which choices that they’ve made this election cycle do you think are the most questionable — in terms of voters’ race, gender, age, etc.?
Silver: What’s most questionable is how an extremely high percentage of the polls in the swing states — something like 75 or 80 percent of them — have shown about a 2.5-point race or narrower. Statistically, that’s nearly impossible. You’re supposed to get more outliers when you’re only sampling 800 people at a time. So other than the New York Times/Siena polls, which have consistently published results that deviate from the consensus, I’m not sure we’re getting pollsters’ real opinions anyway. They’re basically just saying ¯_(ツ)_/¯, to use a slightly dated meme.
Anderson: Well, at my firm, we just put out a poll that showed Trump up by five in Pennsylvania — you definitely can’t say we are herding! But the incentive to herd is real. I’ll give you an example. In 2022, my firm put out some polls in the closing stretch that were pretty great. We got a strong rating from FiveThirtyEight under Nate’s time as well as under the new regime. We got the Virginia governor’s race within a point in 2021, did great work in notoriously hard-to-poll Nevada and even got Pennsylvania really close. We also nailed the Georgia governor’s race, but our poll definitely underestimated the size of the Kemp-Warnock vote — voters who went for both the Republican governor, Brian Kemp, and the Democratic senator, Raphael Warnock — and so we showed Herschel Walker ahead. Now, when people want to criticize the Pennsylvania poll, they point to the Walker whiff, rather than any of the other great results we’ve produced.
The reality is even if you’re a good pollster, every so often you get a dud. It’s the nature of what we do. Of our three blue-wall polls, two looked like everyone else’s, and our Pennsylvania one didn’t. I can look under the hood and go, you know, is the gender gap really that small in Pennsylvania? I don’t know if that feels right. But my gut is not a process. Feelings are not a process. You have to have a process and stick to that process and let it fall as it may. If you mess with the process after the fact, that’s herding.
Bruni: Kristen, I can be the purest pessimist you’ve ever met, and even I don’t think there’s any chance Harris loses Pennsylvania by five points. But that poll result makes me want to ask both of you about the vice-presidential picks. When they’re made, we all give our analyses: Helpful! Unhelpful! Never, ever matters! We’ve now had months of JD Vance (God help me, and God help the Childless Cat Ladies of America) and of Tim Walz. With all that evidence, do you think these selections are likely to have any effect, and, Pennsylvania-wise, should Harris have chosen Josh Shapiro?
Silver: As a founding member of She Shoulda Picked Shapiro, I think it’s relatively clear now that she made a mistake. Pennsylvania seems to be lagging a little behind the other blue-wall states. Meanwhile, Walz was mediocre in the debate, and he’s been mediocre and nervous in his public appearances.
Anderson: We tested a hypothetical Harris-Shapiro ballot. She does a few points better on the margin. But it’s all a hypothetical. Take it with the requisite grain of salt.
Silver: The Vance pick was worse, though, and made at a time when the Trump campaign got really overconfident — that was back when Biden was still running and they couldn’t seem to fathom that Democrats would summon the power to replace him, you’d just had Trump shot at in Pennsylvania, and they thought they couldn’t possibly lose. He remains one of the most unpopular running mates in modern history.
Anderson: If Harris wins, people will point to the Vance pick as part and parcel of the boys-versus-girls dynamic that has become the narrative of this election. But I think the Vance pick is more important for what it foreshadows about the future of the Republican Party and 2028.
Bruni: Apart from veep selection, if both Harris and Trump had the ability to turn back the clock to, say, July 22 (the day after Biden dropped out), what one statement/decision/development/etc. would you advise them to take back, in terms of maximizing their chances for victory?
Silver: I can’t for the life of me understand why Harris hasn’t found a way to distance herself from Biden. Maybe gently, but honestly, not-so-gently would work fine, too. When she was asked on “The View” what she would have done differently than Biden, and couldn’t answer — anyone who prepped her for that interview should be fired and never work in politics again. She inherited lots of the Biden staff, which was maybe understandable at the time given that she had to build a campaign overnight and that she wanted to keep the party unified, but she should have taken the opportunity to get a fresh start.
Anderson: As for Trump, if there’s any looking back, I think defining Harris earlier by her 2019-20 era political positions will be the answer. Immediately after the switcheroo at the top of the Democratic ticket, the David McCormick Senate campaign in Pennsylvania dropped an ad that was just a series of Harris’s furthest left views from that era. But it took a little longer for the Trump team to settle on what its message about Harris would be. Too far left? Too much like Biden? And so on. She was stunningly unknown for a sitting vice president.
Bruni: Let’s do an extended lighting round. Top-of-mind, quick answers. On a scale of 1 (it’s hurting Harris badly) to 5 (it’s helping her greatly), how do you think the atypically rushed, compressed nature of her presidential campaign affects her prospects of being elected president?
Silver: 2. A lot of people thought it would help her and she could ride the momentum straight through to November. But even a three-and-a-half-month campaign is still roughly something like 107 news cycles. The scrutiny did come, and she didn’t really pivot effectively. But she’s also learning how to fly the plane as she builds it.
Anderson: I think it helped more than hurt, so I’d say a 4. Maybe if she’d been on the trail longer, she could have gotten better at the extemporaneous interviews and had time to give voters a firmer sense of who she is in her core — not her list of policies, not her biography, but what her beliefs are about the role of government and the role of America in the world. But ultimately, she benefited from a quick ascension, the rapid unification of her party and then the short runway to the finish line.
Bruni: On a scale of 1 (hurting Trump badly) to 5 (helping him greatly), how significant an effect is Elon Musk in all his bouncing, money-dispensing and curious glory having on Trump’s prospects?
Silver: I’d say a 3. I think Musk has changed the vibes more than the underlying reality. As for the money, both the candidates have more of it than they need, and I’m not sure I’d want Musk running my ground game. That said, Harris hasn’t really found a way to message against it, like by running on a more populist theme that emphasizes Trump’s association with plutocrats and his pay-to-play tendencies.
Anderson: I also give a 3, which is an unsure, because for me the biggest question is whether his turnout machine actually works for Trump or not. The political parties are usually the ground game for their candidates, built up over many cycles and ready to run in election years. This year, Republicans have used more of their party apparatus for things like legal battles, while outside groups like Musk’s are tasked more with turnout.
Bruni: Musk, Jeff Bezos, Joe Rogan, Liz Cheney, Bad Bunny, Beyoncé — we’ve had a spectacular series of cameos in this election. Leaving out Trump’s and Harris’s running mates, name a person whose intersection with this election maybe, just maybe, could have a crucial positive or negative difference? One name.
Silver: I suppose if Rogan were to turn around and unexpectedly endorse Harris, maybe he’d be the one. Otherwise, I don’t think these endorsements matter. Voters are pretty smart at inferring who a given celebrity would endorse anyway, so it would need to be something unexpected. And there aren’t any politicians who are popular enough for it to matter. Maybe Mitt Romney — not that he’s very popular — could at least net Harris some small number of Nikki Haley voters.
Anderson: The Bad Bunny one is the one that is the biggest X factor to me. I agree endorsements don’t mean much. But I find most of my fellow political pundit types don’t have the slightest clue how big a deal Bad Bunny is, which is my first clue that his involvement here at the very end could be undervalued in The Discourse.
Bruni: If you had to pick a surprise winner in one of the Senate races, who would it be?
Silver: The trendy pick is Dan Osborn in Nebraska, but the last two polls have shown the Montana Senate race tightening up again, so maybe Jon Tester has been declared dead prematurely, a little bit like Heidi Heitkamp in 2012. But we’re in much more partisan times now. So I’d need some pretty good odds to bet on him, 4:1 or something.
Bruni: Go, Tester! Maybe, to seal the deal, we have Bad Bunny hop over to Montana.
Anderson: I’ll go out on a massive limb and simply say that in 2014, the public polls dramatically underestimated Larry Hogan’s chances of winning his race for governor in Maryland. The public polls currently show him down by a significant margin against Angela Alsobrooks; to be clear — I’m not predicting he’ll win. But if this year gets really strange and we have the Revenge of the Split-Ticket Voter, who knows?
Bruni: If Harris loses, who is your odds-on favorite to be the 2028 Democratic Party presidential nominee?
Silver: The thing is, Democrats had strong midterms in both 2018 and 2022, so they really do have a deep bench. I suppose I’ll just tick off some of the more obvious names: Shapiro, Gretchen Whitmer, Wes Moore or maybe one or both of the Georgia senators.
Anderson: If Harris loses, I imagine there will be a lot of angst about whether Biden should have dropped out earlier and allowed a process to play out, and a lot of pining for Shapiro.
Bruni: Moore has had some very rough patches; I disagree there. Whitmer would be seen in many ways as the exact adjustment from Harris that the party needs, and I’d also love — this is just emotion — to see Democrats nominate a woman for the third time in four cycles. Third time’s the charm, right? (But I’m not at all giving up on the second time!)
Silver: I worry that Democrats will take away the wrong lesson if Harris loses and become almost superstitious about having to nominate a guy. Obviously, we’d want to look at the cross tabs for hints of misogyny and so forth, but I don’t think Harris’s gender is one of the 10 most important factors hurting her. One point I’ve tried to emphasize in my newsletter is that this has been a pretty tough ask of Harris — taking over in midstream, dealing with the effects of high inflation (yes, better now) and high immigration at a time when voters have been extremely sensitive to those things in elections all around the world. And she has to live down both her association with the unpopular Biden and the unpopular left-wing positions she took in 2019.
Bruni: If Trump loses, who is your odds-on favorite to be the 2028 Republican Party presidential nominee?
Silver: Donald Trump.
Bruni: Dear God, no!
Anderson: I think Vance. Donald Trump will be 82 next time. You have to understand the extent to which Republican voters see things like Vance’s debate against Walz, or his combative interviews with news media figures, and go: Yes, that’s what I’m talking about.
Bruni: I think someone like Glenn Youngkin or Brian Kemp. Republicans would have to say: The Trump route was a dead end. Time to adjust.
Silver: I think it might take Republicans a cycle or two to navigate out of the wilderness. Especially if Trump claims the election was stolen, which I assume that he will, the usual autoimmune response might not kick in.
Bruni: Imagine you have a beach vacation flight booked for Nov. 7 (and I hope you do!) but can’t leave unless the election is resolved. What are the odds you get to board your plane?
Silver: Higher than people might assume — 75 percent? Assuming we’re going based on what the networks say — again, I can’t emphasize enough that I don’t expect Trump to willingly concede. (The Electoral College tallying process was tightened up by Congress, and the courts generally didn’t play ball with Trump in 2020, so I don’t think there are legal routes — but I haven’t forgotten about Jan. 6.)
As I mentioned above, though, there’s a 40 percent chance that one candidate wins all seven swing states. And another 25 percent chance or something that one candidate wins six of seven. In most states, the vote counting process should be a little bit faster, especially with fewer mail ballots than in a Covid election. Even elections that were relatively close, like 2004 and 2016, were called by the early morning.
Anderson: How are we defining “resolved”? There are going to be legal challenges no matter how resounding a victory it seems. I’ve been telling my friends, I’m just trying to get to Thanksgiving.
Bruni: Lastly, it’s 12:01 a.m. on election night. What are you drinking?
Silver: I probably won’t be yet, unfortunately — or probably just a coffee — because while we might get a call by 3, midnight is really pushing it. And if that does happen, that’ll mean it’s an unexpected landslide in one direction or another, so I’ll have to start in on my “why polls were so wrong again” story.
Anderson: Midnight? Red Bull. That’s when the number parsing and post-election “what the heck happened” analysis just starts to get fun. I’ll be on air as well. I’ll save the old-fashioned for the flight to the beach vacation.
Bruni: I have a nice bottle of white Burgundy chilled for this. It’s either my celebration, my consolation or my get-me-through-another-few-long-days. And it’s wine, not bourbon, so I can stay steady enough to keep typing if still on deadline! Good luck to the two of you and to all our readers. Here we go!
The post Is the ‘50-50-ish Race’ Driving You Crazy? You Are Not Alone. appeared first on New York Times.
