The Data Guy Who Got the Midterms Right

Tom Bonier of TargetSmart on how Republican polls were able to skew media predictions.

A polling station during the midterm elections, Detroit, Nov. 8, 2022.
A polling station during the midterm elections, Detroit, Nov. 8, 2022. Photo: Matthew Hatcher/Getty Images

How is it that Democrats were able to buck historical trends and avoid the kind of midterms massacre traditionally suffered by the party in power? And why did so many election forecasts miss the mark? Ryan Grim talks with Tom Bonier of Democratic data firm TargetSmart about what the media got wrong in 2022.

[Deconstructed theme song.]

Ryan Grim: Alright, on today’s episode of Deconstructed, we’re going to try to figure out why it was that Democrats were able to defy political gravity in the midterms, hold on to the Senate, and nearly hold on to the House. And to do that, we’re going to be helped by Tom Bonier, the head of the Democratic data firm TargetSmart, who, along with Simon Rosenberg, was the loudest voice shouting that the media was getting it all wrong.

He joined the podcast last summer to lay his theory out, and he’s back today to explain what we can learn from what he was seeing, and what the media was missing. Tom, welcome back to Deconstructed.

Tom Bonier: It’s great to be back.

RG: So, Tom, can you remind listeners what TargetSmart is and why it’s able to get such extraordinary access to voting and other data?

TB: Yeah. I’d love to. At our core, we’re a voter data company. So we collect data on every registered voter around the country. So it’s, at its core, voter registration lists.

And then what we do is we append other information. So voter models on partisanship, or turnout likelihood, issue preferences, things like that. And we work with Democratic campaigns and progressive organizations.

RG: And so the most important question to me is: What actually happened in this election? Who came out to vote for Democrats, and why? Because the answer to that will shape what kind of agenda Democrats are going to pursue, which is what matters to actual people whose lives are shaped by what the government does or doesn’t do. That’s really the point of everything we’re doing here.

But I want to get to that question after we figured out what went so wrong with the media first. And so you and Simon were generally making three fairly basic points. Let me lay them out and tell me if you think I’m missing anything, or mischaracterizing: one, basically, was that in the special elections with real live human beings doing real live voting, Democrats were consistently outperforming; two, was that Democrats were dominating the early and absentee vote; and three, that some of the polling averages and the conventional wisdom was being skewed and shaped by Republican campaigns who were dumping tons of internal polls into public, making it look a little bit rosier for them than it actually was. Were there other key arguments you were making that I missed? And how would you characterize it?

TB: Yeah, I think that’s largely it.

I mean, in the end, there were sort of two different worlds or two different competing theories: the one theory was held pretty widely, especially in the media. And then there was one that was really focused on all the data that you mentioned. The media theory seemed to be more calibrated around precedent — and just the general default notion going into this election, I think a lot of people went on the record of predicting a red wave. And so any sign that maybe suggested that that could be happening, I think they were over-calibrating, too, perhaps. Whereas, in the end, we were looking at the hard data that you mentioned, and the only other thing I’d maybe throw into the mix would be fundraising data.

RG: Mhmm. Mhmm.

TB: When you look at that, that can be a pretty good sign of intensity or engagement. And just seeing that Democratic campaigns were just doing better in terms of raising money, especially from small-dollar donors.

RG: Mhmm.

TB: When you look at that, you look at all those signs together — and again, there was nothing that Simon or I were doing that was magic, and we weren’t going out and making an exact forecast or predicting what was going to happen. What we were doing is looking at the data, sharing the data, and just saying: Well, look, if the prevailing narrative is this is going to be an inevitable red wave, it’s not emerging in any of the data we’re looking at. And, in fact, the only data where it was coming from was the polling. And, as you say, it was really just a subsegment of the polling, and it wasn’t the polling writ large.

RG: And you guys got absolutely savaged and mocked by reporters and pundits for what you were saying.

TB: We did. Yeah.

RG: And I want to play a quick clip from Nate Silver on his FiveThirtyEight podcast that came out not long before the elections. Zach, do you have that clip? Clip one.

Galen Druke: Oh, no.

Nate Silver: Oh, no. Not this partisan, pollster bullshit.

GD: I mean, you had to expect that.

NS: Not this bullshit, dude.

GD: You had to expect it.

NS: I’ve never seen so much hopium smoke done. [Laughs.] What’s the guy’s name like Simon Rosenberg or something?

GD: You’re talking to someone on Twitter who was like: Look, I counted all of the polls in FiveThirtyEight’s polling average, and a good portion of them are now Republican partisan pollsters, which is correct. And I’m sure that people who are looking at the details enough to see which polls are actually showing up in our averages —

NS: Yeah.

GD: — will notice that this year, in particular, there are a lot more partisan pollsters and those partisan posters tend to be Republican and tend to have better numbers for Republicans. So I understand discouraging people from we support DARE, we don’t want people using any kind of drugs including hopium.

NS: [Laughs.]

GD: However, let’s explain to folks what’s going on with a little generosity here.

NS: Uh, with little generosity!

GD: Yeah, with a little generosity.

NS: It’s six days before the election — I don’t have time for generosity.

GD: I always have time for generosity.

NS: So first of all, our model knows a few things. It’s pretty smart.

GD: It’s been around the block.

NS: It’s been around the block.

GD: This model has been stomping the pavement for a long time.

RG: So they actually floated the idea of maybe being generous to an alternative explanation. And then they immediately dismissed the idea of being generous to it.

Did you happen to hear that episode in real-time? And what was it like to do this for a living and have pundits, talking about hopium? And: Who is this guy, Simon Rosenberg? What was it like to be on the receiving end of that stuff?

TB: I didn’t hear that in real-time, not something I listen to on a regular basis. I did hear that sort of thing pretty commonly, in fact, you know, in even less kind depictions, even though that was fairly mocking, by itself. [Laughs.] And, you know, to be fair, I would hear this not just from — and I know Simon was hearing the same — this wasn’t just from the prognosticators and the media, the Nates of the world, it wasn’t just from reporters, this was also from Democratic strategists and pollsters.

RG: Mhmm.

TB: Someone messaged me and said: I want some of what you and Simon are smoking a few days before the election. [Laughs.] The notion being — and this was a Democratic strategist, someone who I respect a lot, but this was a fairly commonly held perception that there was just something amiss with what we’re putting out there.

But again, it’s kind of funny, because when you look at where we’ve ended up, now the prevailing wisdom is: Well, the polls were right. And some of them were, no doubt. I think where we’ve settled on this is that the quote high-quality polls were right, but that distinction wasn’t being made by FiveThirtyEight or RealClearPolitics, or the people who were assessing these polls.

And so, to your question of how it felt, just speaking honestly, not great. [Laughs.]

RG: Mhmm.

TB: Not great, because there’s a lot of pressure there. And certainly, as the CEO of a data company, I think there’s extra pressure. We’re trusted to provide the best possible analysis that’s based on data, and driven by data. And when every voice you hear, especially from voices who you respect substantially, are kind of holding you out at arm’s length and acting like you just might be crazy, you have to rethink things. So I would find myself, especially in the last week, waking up every morning and feeling that sense of dread of like: Am I looking at this in a wish-casting way? Is this hopium — which is a term I despise?

RG: Mhmm. [Chuckles.]

TB: And I would have to talk myself through the data every morning, literally go through and look at Kansas and NY-19, Alaska at the special elections, look at the voter registration data, and just lay out the evidence and make a methodical case to myself of: Well, what are we seeing? What do we know? What don’t we know? What’s the case to be made for a red wave? What’s the case to be made for something else?

And the evidence was overwhelming — I mean, almost entirely on the side of — no red wave. And so then I would get back to a feeling of confidence. But it was literally a daily exercise.

RG: Right. And so to stick with the FiveThirtyEight crew for a minute: In that episode, they go on to try to explain why they think they’re able to balance out the Republican polls that were being dumped into the public space and tilting the numbers. And they end with kind of a disturbing suggestion of what Democrats ought to do.

And, Zach, do you have clip two handy?

NS: Well, one thing we have is what’s called a house effects adjustment. If a poll consistently skews toward Democrats or Republicans, then the model can adjust for that. And it would understand that a Trafalgar poll showing Oz up by five points is not the same as if Quinnipiac says that or something, right?

GD: Mhmm.

NS: If it’s an explicitly partisan polling minute conducted for a Republican candidate or organization, then the house effects adjustment is even quicker to kick in. So in principle, the fact that you have a lot of partisan polls, a model can, for the most part, account for that.

I’d also say that there’s no reason that Democrats couldn’t put out their own polls. To some extent, it reflects some degree of confidence because these pollsters will get a lot shit if they’re wrong, right? You know, Trafalgar, etc, if Democrats have a good night, that would really hurt their credibility. So they are taking some risks. And Democrats, if they believe the real race numbers are here, they can publish their own polls, potentially, and actually, take that credibility risk.

RG: If Democrats are really confident about the fact that we’re getting this wrong, they can dump their internal polls and skew it back in their direction. What do you think that would do to the way that people understand polling? Like, should it be a thing that people are jockeying for?

TB: That’s one I did here in real-time, not that particular interview, but then Nate tweeted that out, I assume, right around the same time —

RG: Mhmm.

TB: — and I reacted to that immediately, because it’s such a deeply flawed, bad idea. His notion, he said, that their average should act like a betting market. And if one side is betting heavily on a certain outcome, that should tell you something. Well, betting markets should not be how we’re internalizing and understanding these races. And the notion that, well, if Democrats thought they were doing better, they would put out more polls hopefully has just been proven wrong once and for all, but I think it does really illuminate this notion of people saying: Well, the polls were right.

A great example of this is the Washington Senate race because that was one where a lot of Democrats saw these polls coming out in the last few weeks that showed Sen. Patty Murray in a tight race. And I think that colored this perception of well, gosh, that’s a red wave —

RG: Mhmm.

TB: — because we’ve all lived through wave elections, and you have surprises, you have candidates who lose who you weren’t expecting to lose. And look at the FiveThirtyEight average: Their closing average showed her up 4.7 points. And if the polls worked, why did that average show her up 4.7 and she won by 15.

RG: Mhmm.

TB: And the fact of the matter is, while all the polls in their average, over the last month, were all Republican polls, there weren’t any public polls in there beyond — there weren’t what they call the high-quality polls, there wasn’t a New York Times-Siena [College] poll or any of these others. So the notion that you all you can just throw them all in and quote trust the average, which was what we were told time and again, or the notion that well, if Democrats truly believed she was going to win by more, they could have or should have put their own polls in, something that is analytical shouldn’t rely on partisans throwing in more polls to change the average. That makes no sense at all. But again, that was the prevailing wisdom before the election. And some of that still seems to be lingering after the election, which is more than a little bit troubling to me.

RG: And so the other contested field of battle was over the early voting and absentee voting requests — what you can draw from that? And I think they commented a little bit on that in that episode, too. Zach, do you have clip three and then four?

NS: If it weren’t for those special elections, I would kind of totally buy the doom case. But there’s something about how that reflected some real state of the world and then things have really shifted that much since August.

GD: That smells like some potential hopium for our Democratic listeners.

NS: Well, I mean, but it’s trying —

GD: We don’t have to rehash the whole —

NS: But that seems a better argument to me than like the: Hey, man, just trust the early voting numbers.

RG: So there they’re even doing the [laughs] stoner voice. They’re talking about the early voting numbers. And then it turns out that their listeners, I think, were following what you were putting out very closely because they say that the number one question that they were getting from listeners was actually about these early voting numbers. And so they address it really quickly. And then I want to get your response to it.

Zach, can you play their responses, clip four?

GD: OK, so this is truly, I think, our most popular question when we get down to the final weeks of an election, so let’s get it out of the way. Brian asks: Is there any way to actually use historical early turnout data to model these election results as they relate to heavy increases in early voting?

NS: Because it’s a moving target. I mean, we’ve had different parties over the years have emphasized the early vote to greater or lesser degrees. And that’s a problem, right? It’s not like some regular number that you can just kind of target toward. It’s different in every state, and different for every race; obviously, in 2020, Democrats put much more emphasis on mail voting and to some extent, voting early, but not necessarily voting early in person, because they were voting by mail. And there’s just nothing to anchor to.

RG: And so is that true? Do we have anything to anchor early vote numbers to? Why were you able to extract such a kind of high-grade hopium from those numbers that they don’t see as worth much?

TB: [Laughs.] High-grade hopium.

RG: High-grade hopium. It was the good stuff.

TB: Right. I look at it this way: Over 50 million people voted before Election Day, which is going to be a little bit less than half of all the ballots cast. If you can look at a database of 50 million people, and we know who they are, we know which elections they voted in, and we have a good sense of their partisanship; we know race, ethnicity, age, education levels. If you can look at 50 million people who have voted in an election, and you can’t draw some conclusions, you’re not a very good data analyst.

And I think there were a lot of people who ignored the early vote because it was counter to the narrative that they were expecting or that they had predicted before Election Day. And to be fair, yeah, there’s a lot of bad early voting analysis out there. And it is a little bit complicated, but so are all of these things. So are dealing with polls. The fact of the matter is, it was a good window into Democratic intensity and enthusiasm.

And if there were sort of two broad generic questions, if I can oversimplify, in this election, if you’re assuming it’s a red wave, it’s a question of: Are Democrats going to come out at a level high enough that can maybe mute that red wave? Or is this just going to be a wash, and Republicans are going to come out in such large numbers and Democrats are going to stay home, that it will look like a typical wave election? We know what wave elections look like from a turnout perspective.

So yeah, the early vote definitely is skewed overwhelmingly Democratic, because the president convinced his supporters that mail voting is fraud and took away one of their biggest tactical advantages —

RG: Mhmm.

TB: — from prior election cycles. Sure. We know that. But to your question, what do you anchor it against? We have the 2020 election, we have prior elections, we had a sense from polling data in terms of how we expected the early vote to shift, because we knew a lot of people who voted early in 2020 would go back to Election Day. But we also know that would be skewed younger, which is a more Democratic electorate.

So, in a state, if we’re looking at the early vote, and we see that it’s more Democratic than it was at the same point in time as 2020, that’s a good sign for Democrats. And the places where we saw that happening line up with the most positive Democratic results in the end, like Pennsylvania, like Michigan, Arizona.

I mean, the thing that makes me laugh is we think about the amount of press that was given to, you know, a sample size of 100 respondents, 70 respondents in a The New York Times poll that talked about independent suburban women trending against Democrats.

RG: Oh, that crazy influential poll at the very end —

TB: Right!

RG: Right. That saw some 30-point shift in Pennsylvania’s Pittsburgh suburbs or something, and everyone —

TB: Right, I think that might have been the Wall Street Journal poll. And, yeah, that shaped so much of the narrative in the last handful of days. The New York Times poll did the same. I mean, God, there was a poll in Pennsylvania after the debate that said it was a poll on people who weren’t voting for Fetterman based on the debate performance. I did the math on it: It was literally 12 poll respondents, and this was getting press. So we’re comfortable with talking about dozens of poll respondents. But when it comes to 50 million votes, that’s hopium.

RG: Right.

TB: I mean, that’s such a bizarre thing for anyone to say, let alone someone who’s supposed to be an election analyst.

RG: Not even dozens. Like, in that case, a dozen people.

TB: Right.

RG: And when you put it that way, when you say: Well, hey, look, there are 50 million people who have voted here, maybe there are some lessons that can be drawn from that, that seems pretty clear. So what’s your guess as to why did the media get this wrong? And not to dunk on the media, but how could they do it differently next time around so that we don’t see a result like this where you kind of manufacture a narrative, and then the narrative doesn’t come through, and then people are losing faith in what they’re hearing and what they’re seeing?

TB: Yeah, I think a big part of what they got wrong was sort of an overcorrection to their priors. They had been through the experience of 2020, where the polls had a Democratic bias; 2016, where they had a Democratic bias. So they were looking at the polls, and they were just assuming, I think, in how they’re internalizing these polls, that they would have a Democratic bias again. So they go back now and say: Well, the polls were right — again, if you limit to the polls, after the fact, that we now know were right, sure.

RG: Right. [Laughs.] The polls that were right were right.

TB: Right! Which is pretty much what they’re saying. And then like, yeah, there is a fairly neat distinction that the sort of better-known, somewhat apolitical media polls were more accurate. Yeah, but they weren’t making that distinction in real-time. And, certainly, these averages weren’t doing a very good job of dealing with that. So I think that’s the other issue.

And just in general, they were overly reliant on polls. I mean, looking forward, what do you do in the future? Well, now we have a pretty wide range of potential outcomes. So when we think of 2024, I can see now these analysts who are focused exclusively on polls and aren’t using the other data, the voter registration data, the special election data, the fundraising data, the early vote data, if you’re just calibrating on polls, sure the polls were better, but now we have a Republican-biased election where almost all the polls, every poll, basically outside of Florida, was biased towards Republicans. So that’s going to have to factor in the equation now.

I foresee a lot of people saying: Well, look, if the poll bias looks like it did in 2022, then things could be three or four points better for Democrats, which is about what the average pull bias was. But if they look like they did in 2020, it could be three or four points better for Republicans. Suddenly, you’re dealing with an eight-point range more and more —

RG: Mhmm. [Laughs.]

TB: — in a lot of the states. That’s not going to give you a lot of confidence. And so they really do need to start bringing in a lot of these other richer datasets that, in the end, as we learned this year, are actually more predictive than the polls, or at least need to be used in the context of the polls. Because that’s one important thing that we’re able to do is look at these polls and say: Well, look, this is a poll that’s based on the notion that young voters aren’t going to show up. We saw that in Pennsylvania, too, with a lot of these bad polls that were showing Oz pulling into the lead, which as we know now, Fetterman is going to win by about five points. It just didn’t happen. That wasn’t a reality. It was just driven by bad polls.

Well, if the media looked at those polls and saw, yeah, this is based on a turnout scenario, that’s not likely to materialize. I don’t think we need the media to draw any of these broader conclusions, but they need to report the context of these polls.

RG: Mhmm.

TB: And then the general public can draw a conclusion as to whether or not that’s likely, I don’t think many people would have looked at this election and said: Yeah, after the Dobbs decision, there aren’t going to be gender gaps, which is what a lot of these polls showed.

RG: Right. Right.

TB: Or young people are just gonna stay home when we have seen them come out in places like Kansas.

RG: Right.

TB: That just was never going to happen.

[Musical interlude.]

RG: So now there’s a new debate, a new data debate about the election that has really high stakes, which I was alluding to at the beginning. Immediately out of the gate, people saw this uptick in the youth vote, and Democrats argued that this showed the benefits of the student-debt cancellation and also standing up for abortion rights after Roe was overturned. I’d add to that, that the climate bill was passed, plus the pardons around weed and also the strong job market and the wage gains at the bottom, which didn’t get much attention. I think all of that helped with young people.

And the conclusion there would be that Democrats ought to do more things that young people want, because if they do, they’ll come out and support them. Now, I’m biased toward that explanation, because young people are more progressive and I want Democrats to do progressive things.

But you have a different camp of centrists, who are now saying: No, there wasn’t actually any significant increase in youth turnout. And actually, the election was won by persuading independents and Republicans to switch sides and vote Democratic, which means Democrats should be kind of cautious, reasonable centrists against these extreme MAGA types, and that’s all they need to do to win.

And I think you saw this, I wrote this week about David Shor’s argument in that direction. And you had a little back and forth with him on Twitter about this. Based on the numbers we have so far, what do we know about youth turnout and what it says about this debate?

TB: Well, you know, and we have to be skeptical about any argument that completely eliminates one side or the other.

RG: Mhmm.

TB: We know winning campaigns, generally, are almost always going to be some combination of mobilization and persuasion. And then we can argue about the rough mix. But, you know, as you say, there have been a lot of hot takes after the election that suggests Dems stayed home, younger voters didn’t come out, to the extent that anyone thought that these issues were going to mobilize the youth vote, it didn’t materialize. And those have gotten a lot of uptake. [Laughs.] They’ve spread like wildfire — again because I think they matched people’s priors. And they can say: Well, I was wrong about my forecast for the election. But in the end, the general underlying dynamic was consistent with the way I view the world.

And there’s no evidence of it. The problem is it’s not based on evidence. So the good news is we’ll have answers to all these questions on the voter files that we collect. We have data — again, for each individual, we don’t know for whom someone voted, but we know whether or not they cast a ballot. So we can see what turnout actually was. We have that for five states now. It’s not the whole country. I’m not going to draw up the broadest conclusions yet. But it’s a pretty good sample. So we’re talking about Georgia, Nevada, Colorado, Washington State, and — oh! — Oregon. And in those states we have basically complete voter history, we can look at the youth vote, and we see a fairly consistent trend at this point.

And if you take the last two midterms, you’ve got 2014, which is the best parallel. That’s the last midterm where Democrats were in control in the White House, Republicans were the out party. The discussion before this election was: Will it be 2010 or 2014? 2010 was a huge Republican wave; 2014 was a smaller Republican wave. 2018 was not in the conversation. When we look at the youth vote now, what we’re seeing consistently, is that the youth vote this year blew 2014 out of the water, way higher than 2014 across those five states; it comes close to 2018 in all of those states, by varying degrees; and there’s some instances — I just looked at Washington State today where we just got that vote history in — where the youth turnout, at least Democratic youth turnout was matching 2018. And when we look at the 3rd congressional district, which was one of the biggest surprise Democratic pickups with a 34-year-old Democratic candidate running, youth turnout actually exceeded 2018.

And, I mean, the context of that: 2018 set a record for midterm turnout overall —

RG: Right.

TB: — and a record for youth turnout. So to come anywhere close to that, in this election, where younger voters, frankly, when you think back to the earlier part of this cycle, pre-Dobbs, pre any of the legislative accomplishments that you mentioned, younger voters were not very engaged in this election. They were giving President Biden his lowest approval ratings, which I’ve never seen before — generally, its inverse. But they were just not very connected to Democrats and to the political process in general.

And then you saw that change. You saw that change with Dobbs; you saw it change with student debt forgiveness; with the legislative wins. And so the fact that here’s where we end up, younger voters actually came very close to matching these record levels of 2018 turnout in the states where we have the data so far. I don’t think we’re gonna see that in every single state. But the fact that we’re five for five is —

RG: Right.

TB: — suggesting that what we can say is of the evidence that we have at the moment, of the states where we have the vote history, youth turnout was very high.

RG: And we probably won’t get to see the Republican methodology behind their internal polls. Because you were talking early about how Washington State was so wildly off, because it relied so heavily on partisan Republican polls. So we probably won’t see how many young people they guessed would come out to vote in Washington State, but I’m pretty comfortable guessing that Republicans felt like they would get a 2014-level youth turnout, like if they’re doing their models —

TB: Yep.

RG: — their pricing, we don’t see it. It’s not — they hate Brandon. Young people are staying home. And so they’re going to model it as if it’s 2014. That would almost perhaps explain 100 percent of their miss if you actually sell 2018-level youth turnout rather than 2014. Am I exaggerating that? Or do you think a huge amount of their Washington State miss and then, therefore, the media’s Washington State miss was actually underestimating youth turnout?

TB: Yeah, my gut is you’re right. And you’re also right that generally, one of the common threads with these Republican polls is a lack of transparency so we won’t really have the ability to judge their samples for the most part, though I am planning on going back and looking at Washington to see if we can find one. And to the extent that their youth sample comes anywhere close generally what we found in these instances is that they’re oversampling Young Republicans. To the extent that there’s a response bias, it’s that they’re getting a more conservative youth sample because young Republicans are more survey responsive than Young Democrats.

And so yeah, I think that was the general notion which I can’t just blame Republican pollsters, to be clear, because, again, that was the general commonly held media narrative that I think a lot of consultants and strategists on both sides of the aisle, embrace to varying degrees the notion that again, is this election going to look like 2010 or 2014? It was: pick your flavor of Republican wave —

RG: [Laughs.]

TB: — it wasn’t the consideration that, wait a second, this is a different election, Dobbs had a huge impact on that. And then Dobbs by itself is an issue of taking away a woman’s right to choose in general, is a huge factor. But then you have the fact that it connected this narrative of Republican extremism more broadly, which I think younger voters were able to organize around as well. I think it all sort of contributed to a surprise for a lot of people in terms of youth turnout. Again, if you paid attention to the data, though, if you paid attention to Kansas— I talk about this again and again and again, but I don’t think it can be repeated enough — women under the age of 25, in Kansas, when you look at what percent of registered voters, women under the age of 25 voted in that early August primary election, a higher percent of them voted than the percent of all men registered voters who voted in that election.

RG: Hmm.

TB: If you look at that, how can you draw the conclusion that young voters aren’t going to be engaged in this election?

RG: And the imbalance between the two parties among young voters is fascinating. I’ve seen people talk about how the Republican youth turnout was either flat or down relative to say, 2014, or past cycles. What do we know about that? And how unusual is that? Because usually, I would think that kind of turnout by age would move kind of across the cohort — it would be up for Democrats, Republicans, and independents. And then you might see people switch their votes or vote differently. But they’d all turn out at the same rate, but it does seem like progressive Democrats — I say progressive because a lot of these young people are not actually registered Democrats, they turned out and Republicans didn’t at the same clip. Is that right? And how unusual is that?

TB: In the states where we have vote history, that’s absolutely right, which, again, actually suggests that when we look at the youth turnout overall and compare it to 2018, and we’re seeing that in 2022, this year, the youth turnout was lagging slightly behind 2018, when you account for that partisan differential, and the fact that yes, younger Republicans, for the most part, were actually turning out of the lower rate, Democrats were actually coming much closer to matching 2018, if not exceeding 2018 in the states where we have vote history, is suggesting that yeah, that partisan differential was important. It makes sense, in a way. I think a lot of these younger Republicans just weren’t motivated by their party. They weren’t motivated by the messaging or lack thereof. Some of them actually have a stronger Trump connection. And we’ve seen how [for] Republicans, Trump is a double-edged sword for them. In some ways, him being on the ballot has motivated more Republicans to turn out. But his general extremism and what he’s done to the party has also not only motivated more Democrats and progressives to turn out, it has turned a lot of people who were formerly independent into Democrats, it swayed people against the party. And so, you know, I think you’d certainly see that with younger voters.

RG: It’s like his aura, and his whole vibe and presence is enough to fire Democrats up and get them to the polls, but it seems like he has to be literally on the ballot for some Republicans to turn out — like, just MAGA without Trump doesn’t do it for them. And so I’m curious about how that plays into Georgia — because Georgia went to a runoff, it’ll be the race that decides whether Democrats have 50 or 51, which is consequential down the road, consequential for legislative activity, it makes Manchin much less powerful, but it doesn’t decide control of the Senate. And also Trump’s not on the ballot. We don’t have any numbers yet, right? But what’s your sense of how that plays into the runoff, which comes in early December?

TB: Yeah. So not only is Trump not on the ballot, but now Kemp isn’t on the ballot, which for Walker, I think becomes more of a challenge because Kemp acted as a bit of a buffer, someone who is viewed as standing up to Trump. And I think that helped Walker to some extent, and now he doesn’t have that at the top of the ticket. I think it’s easier to paint him as — or at least connect him with — that Trump brand of extremism, and the voters of Republican voters who are turned off by that but would vote in the November election, I think a lot of them stay home.

I do think it’s an open question in terms of: Does the youth vote come back out for Democrats? Because again, you wrote about this much more eloquently than I’ve been able to lay out the argument, but as you mentioned a minute ago, you wrote about this specifically in Georgia, the youth vote was quite good in the November election and so [it’s an] open question as to whether or not to come back out. I do think we’ll see them engaged at a pretty high level, but because the Senate control is not on the line, I think that’s an interesting question.

And obviously, Kemp’s voter suppression bill that is in effect made this very compressed election calendar, a four-week election calendar, it takes the early vote and takes vote by mail, basically, entirely off the table, which, as we know, has been you know, hugely Democratic going back to 2020 — and then a much more condensed early in-person period, which hasn’t been as Democratic, but still Dems are winning that early in-person. We’ll see what sort of impact that has on the election. Younger voters were skewing much more to election day. So I think there’s a reason to be hopeful there. But that’s a big open question.

RG: Are there any numbers yet? And we’re talking on the day before Thanksgiving — are there any numbers on requests for ballots or anything? Or is it zero data? What do we know about Georgia at this point?

TB: Nothing yet. So our target early site that we launched publicly, for the general election nationwide, It’ll be out there again, likely by Monday after Thanksgiving. This weekend going into next week will be the beginning of the early vote, though even that’s been litigated as Republicans try to condense it, eliminate it in a lot of cases, it’s going to be uneven in terms of where it’s actually out there. So, you know, this is one state where I’d say the early vote actually will be less predictive here, relative to where it was in the November election. But no, at this point, we don’t have any data because people won’t be able to vote by mail. So ballot requests aren’t even a thing at the moment; it’s going to be overwhelmingly skewed to early in-person and mostly Election Day voting, meaning most of what’s going to happen will just be left to that one day.

RG: And it’s so crazy. And for people who aren’t following it, they’re supposed to start voting on Saturday after Thanksgiving but was it Kemp or Raffensberger said no, you can’t start voting Saturday after Thanksgiving. And then a court said: No, you actually can start voting. And so it seems like mostly Democratic areas are going to allow voting on Saturday.

Republicans are picking up pieces of the working class, including the Black and Brown working class, over the last couple of cycles to the point where they’re about 50-50 when it comes to working-class voters at this point. But they haven’t caught up to the idea that they can win — if you let as many people vote as possible that Republicans can actually do better. Like, they just don’t seem to have internalized that at this point.

And so they wind up in these crazy places where Democrats are taking advantage of all of this early voting. And in this situation, you might literally have voting available in Democratic areas, but not in Republican areas, because of Republican decisions.

TB: Right? Well, it’s so weird, because, you know, in the immediate aftermath of this election, you saw prominent Republicans, including Mitch McConnell, and Lindsey Graham, and Republicans who lost, like Blake Masters and Arizona, and there was this fairly universal thread among their post-election comments, which was: We got destroyed in the early vote, vote by mail, early in-person, but just Democrats did a much better job of it. And we need to get better at it. And I’m thinking: Y’all were lapping Democrats, just four years ago, when you look at Florida and Arizona, and these places where Republicans used to vote by mail, especially to their extreme benefit, and it was Democrats who are generally relying on one day of voting in order to make up those margins.

And then, as we all know, Trump flipped that around just by divine edict, in 2020, saying that this was all fraudulent. And I thought: Well, look, Republicans are actually finally waking up after this election, and they’re realizing how tactically dumb this is. But you’re right, Georgia is their first opportunity to flip that script. And they’re not only not apparently making an effort to do so, they’re actively making it harder, or, as you noted, impossible in a lot of these places where it just won’t be an option. It seems like a big tactical error.

RG: That’s just so incredibly self-destructive. It’s just amazing. Amazing. It’s just amazing to see — just, guys, if you vote, that’s the whole thing here. Like it’s an election, you have to vote.

TB: Well, just given that these run-offs, they’re unusual. People are not used to voting in the beginning of December, and then the election, and so, you know, you would tend to think that would be a rationale for making it easier and giving your campaign more bites at the apple, so to speak: if we can’t turn them out today, we’ll turn them out tomorrow, if we can’t turn them out tomorrow, we’ll turn them out the next day. And you can go back, and back, and back and also you have a diminishing universe of voters with whom you need to communicate because we get data in real-time, as do Republicans, on who’s cast a ballot that day. And so you can strike them from your get-out-the-vote universe, you can strike them from your persuasion universe, and now you’re focusing on a smaller universe of voters, your money’s going that much further, and the fact that Republicans seem to be stuck in this rut — I mean, you talk about it being self-destructive, I think that’s one of the broad themes of the Republican Party since Trump took over is that they find new ways to self-destruct, and to not learn their lessons — in fact, to calibrate against the lessons that they should be learning by continuing to nominate extremist candidates, and continuing to make it harder for everyone to vote, including their own supporters, and then tactically, pushing all of their operations and all of their get-out-the-vote work into one single day.

RG: Yeah, and if you do that, and then there’s a freak snowstorm in Reno, it rains in Las Vegas, and you have voting machine problems in Maricopa County, that disproportionately ends up hurting you. But I’m not a Republican consultant. It’s not my job to tell them what to do. [Laughs.] No, keep doing it! If you’re a Democrat, you love the decisions that they’re making here.

But, Tom, thanks so much for joining me again. I really, really appreciate it.

TB: Thank you, Ron.

[End credits music.]

RG: That was Tom Bonier and that’s our show.

Deconstructed is a production of First Look Media and The Intercept. Our producer is Zach Young. Laura Flynn is our supervising producer. The show was mixed by William Stanton. Our theme music was composed by Bart Warshaw. Roger Hodge is The Intercept’s editor in chief.

And I’m Ryan Grim, D.C. bureau chief of The Intercept. We’ll have a special episode hosted by Akela Lacy posting this Tuesday, in which she interviews Larry Krasner, the lightning rod district attorney from Philadelphia. That day will be Giving Tuesday, and if you can spare something for our journalism at The Intercept, you can do that at

If you haven’t already, please subscribe to the show so you can hear it every week. And please go and leave us a rating or a review — it helps people find the show. If you want to give us additional feedback, email us at Thanks so much!

And see you soon.

Join The Conversation