Welcome to FiveThirtyEight’s weekly politics chat. The transcript below has been lightly edited.
micah (Micah Cohen, politics editor): Everyone ready for a helluva debate about politics?
natesilver (Nate Silver, editor in chief): Hell yeah.
nrakich (Nathaniel Rakich, elections analyst): Yup.
clare.malone (Clare Malone, senior political writer): I’m gonna be mediocre.
natesilver: I’m ravenously hungry and was up until 5 a.m. working on the statistical model that will spit out our House forecasts. Other than that, I’m gonna be in tip-top shape for this chat.
micah: Here’s our question: Who are the important swing voters in 2018?
The nominal spark for this question was this article by friend-of-the-site Nate Cohn. Here’s the top:
“The battleground in the fight for control of the House is starting to come into focus with 99 days to go until the November election. It’s not exactly the battleground that analysts expected.
It’s not dominated by well-educated, suburban districts that voted for Hillary Clinton. Instead, the battleground is broad, and it includes a long list of working-class and rural districts that voted for Donald J. Trump in 2016.”
micah: But I want to talk about voters more than districts, though obviously we can get into both.
So … opening bids?
nrakich: Well, I think Nate (Cohn — I have to be specific when there are three Nates on the table) is right that the battleground is pretty broad this year, covering all sorts of districts and voters.
clare.malone: Why are you all named Nate?
A topic for another chat
micah: For realz.
It’s annoying.
nrakich: Speaking broadly, I think the main question is whether 2018’s swing voters are 2012’s swing voters or 2016’s swing voters.
In other words, the white working-class voters who voted for Obama but also Trump? Or the affluent former GOP suburbanites who are maybe now Democrats in the Trump era?
My guess is they’re both on the table in 2018.
clare.malone: One point that I think was interesting was Nate COHN’s about candidate quality:
“The Democrats have succeeded in recruiting well-funded and strong candidates in many of the battlegrounds, which has tended to lessen the advantage of incumbency even in the districts where Republicans are running for re-election.”
micah: Also, not to quibble and/or self-horn-toot, but not every analyst — cough … us … cough — expected the 2018 battleground to be in well-educated suburban districts. We’ve been saying “broad map” for a while.
natesilver: I mean … I suppose I have a hard time reducing it down to one group. You might say: (i) Who are the most important swing voters? (ii) Who are the most important voters in the Democratic base? (iii) Who are the most important voters in the Republican base?
Midterm elections are, to a large extent, about clawing back territory from the other party, rather than necessarily forging new ground.
So I lean toward thinking it’s those Obama-Trump voters who matter in category (i), which means a tilt toward the white, working class.
micah: What’s our evidence for that?
clare.malone: Cohn pointed to Democrats doing well in special elections in Obama-Trump areas (Conor Lamb, for example) but having a harder time in places that were more traditionally Republican until the last election (i.e., ’burbs filled with those college-educated white voters).
nrakich: Yeah, a few months ago, I looked at Democratic special-election overperformance by state in state and federal special elections. Democrats overperformed by the most in red states, including newly red ones in the Midwest like Pennsylvania, Wisconsin and Iowa, where there was plenty of room to revert to the mean. Here’s the table from that piece:
How Democrats have fared in special elections
Difference between Democratic vote margins in each state’s special elections vs. those districts’ partisan leans,* since Jan. 20, 2017
In districts with special elections … | ||||
---|---|---|---|---|
State | Num. of special elections | Avg. district partisan lean | Avg. dem. margin | Dem. swing |
Kentucky | 2 | -55.2 | +1.7 | +56.9 |
Oklahoma | 8 | -38.7 | -6.6 | +32.1 |
Tennessee | 2 | -47.0 | -14.9 | +32.0 |
Alabama | 1 | -29.1 | +1.6 | +30.7 |
Pennsylvania | 2 | +14.2 | +43.2 | +28.9 |
Wisconsin | 2 | -28.6 | -1.4 | +27.2 |
Iowa | 4 | -17.7 | +8.4 | +26.2 |
Kansas | 1 | -29.3 | -6.2 | +23.1 |
Missouri | 9 | -30.1 | -9.1 | +21.0 |
California | 1 | +69.3 | +87.2 | +17.9 |
New Hampshire | 11 | -2.1 | +15.6 | +17.7 |
South Carolina | 7 | +2.7 | +20.2 | +17.6 |
Montana | 1 | -21.3 | -5.6 | +15.7 |
Louisiana | 3 | -16.9 | -1.3 | +15.6 |
New York | 3 | +44.5 | +57.7 | +13.2 |
Minnesota | 3 | -17.6 | -7.4 | +10.2 |
Michigan | 2 | +22.9 | +30.5 | +7.7 |
Delaware | 1 | +12.2 | +17.4 | +5.2 |
Utah | 1 | -35.2 | -32.4 | +2.7 |
Maine | 1 | -17.9 | -15.2 | +2.7 |
Massachusetts | 4 | +15.8 | +14.6 | -1.2 |
Georgia | 11 | -11.5 | -14.1 | -2.6 |
Connecticut | 5 | +7.5 | +2.4 | -5.0 |
Washington | 5 | -11.7 | -18.1 | -6.4 |
Rhode Island | 1 | +32.2 | +25.0 | -7.2 |
Florida | 4 | -1.7 | -10.3 | -8.5 |
Again, those include state-level races, but they’re consistent with what we’re seeing in federal specials.
natesilver: Well, Micah, we’ll have the midterm forecast up at some point later this month, at which point I’ll be able to cite a lot more evidence!
nrakich: Plus, Micah, in that article you linked to a moment ago, we compared Democratic overperformance in special elections to demographic factors like education levels and income. The relationship was rough, but we found that Democrats overperformed more in working-class areas.

natesilver: I’ve become somewhat convinced, though, that the Sun Belt path — e.g., picking up a ton of districts in states like Georgia and Texas — is a little overrated for Democrats.
For one thing, those aren’t the districts where they’ve performed all that well in special elections, as Nathaniel noted. But for another, they’re a bit redder than you’d gather from looking at presidential elections — both states are very red in state elections, for instance.
clare.malone: OK, so WHY are Democrats doing better, relatively, in these places with non-college voters and/or lower-income voters?
Is it because they have more history there?
Is it because candidate message about inequality is resonating?
Is it because Trump is unpopular?
natesilver: I mean, it’s like — what baseline are we using? Hillary Clinton already really overperformed in rich, suburban areas, so it’s hard for Democrats to overperform relative to her in those places.
nrakich: Yeah, Nate, that’s an important point.
natesilver: They might roughly match her performance there, though, and then overperform relative to Clinton in a lot of places where Obama did better or where voters will give Democrats a longer look for Congress than they would for the presidency.
nrakich: Democrats in special elections have generally exceeded Obama’s marks in districts where Trump overperformed more than they have exceeded Clinton’s marks in districts where she overperformed.
natesilver: Yeah.
nrakich: That could be because Clinton vs. Trump was just about the best possible matchup for Democrats in affluent suburban areas. But in working-class districts, it’s easy to imagine a populist Democrat doing a LOT better than Obama did in 2012.
micah: Better than Obama? Not Clinton?
nrakich: Right, Micah.
micah: So — just to put a bow on this — these Obama-Trump voters voted for Obama, then voted for Trump, and now they’re voting for Democrats in larger numbers than they did originally?
That’s remarkable!
nrakich: Based on special election results, yes. For example, in a high-profile special election in Wisconsin’s 1st Senate District in June, the Democratic candidate won by 3 percentage points. That was an improvement not only on Clinton’s 18-point loss in the district but also Obama’s 5-point loss.
But, granted, special elections are “special.”
clare.malone: So to bring it back to the race and gender thing: Are Democrats running more white men in these places and they make for better candidates in these particular areas?
nrakich: That’s interesting. I don’t know.
natesilver: I’m not sure it’s necessarily about the candidates so much as places that don’t have all that much loyalty to the Republican Party.
clare.malone: I don’t know … I wouldn’t discount race and gender in voters’ instincts, Nates.
nrakich: Do white men make for better candidates? There’s plenty of evidence that female candidates do well when they run (they just don’t run as often).
On the other hand, white men are likelier to already be elected officials, and elected officials do perform better than political novices, generally speaking.
And in places like Minnesota’s 8th Congressional District, which went 54-38 for Trump, a lot of the downballot elected officials are Democrats because the area is so ancestrally Democratic. So Democrats have a better pool of candidates to recruit from in those districts than in affluent suburbia/Sun Belt ones.
natesilver: I mean, I think the Democrats’ gains are likely to be concentrated in (i) the Midwest, (ii) California, (iii) New York/New Jersey. Those three places, specifically.
clare.malone: So moving off gender to Nate’s point about lack of loyalty to the Republican Party. Are the Democrats making a more concerted effort to return to some of their original party branding — i.e., populist, of the people? Are they trying to shake identity politics in certain places?
I’m just trying to add a little hypothesis meat to the bones of these numbers that we’re seeing. What are they doing in these places that’s resonating?
micah: hypo-meat
thesis-meat
clare.malone: FLESHing it out.
nrakich: Yeah, Clare, maybe it’s more natural for Democrats in those traditionally Democratic areas to go back to the party’s working-class roots. Look at Richard Ojeda in West Virginia’s 3rd Congressional District. He’s running as an unabashed Trump-voting, Coal Country Democrat.
natesilver: I guess I just think that there’s almost always a backlash against the president’s party at the midterms, and this president is more unpopular than average. So the Democrats have a lot of wind at their backs from “the fundamentals.” They don’t necessarily have to have terribly clever messaging. It does help that they’ve nominated viable candidates almost everywhere — that’s a big deal. And that Republicans have retired in a lot of places, as we’ve written.
nrakich: But, Nate, that doesn’t explain why there are variations from place to place in how hard the political wind seems to be blowing.
natesilver: I think most of those variations can be explained by creating a better baseline of what our expectations are than “just” 2016.
Looking at longer-term voting patterns, voting in statewide elections, in past elections to Congress. And maybe at which areas have demographics that would lend themselves to higher or lower midterm turnout.
nrakich: Oh, totally. In fact, I think 2012 is probably a better baseline than 2016. So you just think that, if we find the proper baseline, Democrats would be doing uniformly better over that baseline?
natesilver: Maybe, yeah.
I mean, in our House model, we’re also going to incorporate state legislative elections into our version of PVI, which is a good measure of overall partisanship.
micah: Wouldn’t stuff like elasticity — how many swing voters, as opposed to committed partisans, a place has — make a difference?
natesilver: That too — there don’t tend to be as many swing voters in the South.
micah: In how strongly places/voters react to the politics winds.
nrakich: I was just going to mention that, Micah!
micah: Too slow.
nrakich: We should really do elasticity for congressional districts.
natesilver: To be clear: I’m NOT saying there won’t be variation at the race-by-race level.
But I don’t think the region-by-region variation should be surprising.
micah: I’m confused. So you think there will be region-by-region variation?
natesilver: There could be, Micah. But we won’t necessarily know it in advance. The patterns we’re seeing so far shouldn’t surprise people all that much, IMO. I certainly don’t think people should be surprised that the wave (if there is one) won’t necessarily be concentrated in rich-ass suburbs.
clare.malone: Can I look forward to 2020 for a second? Let’s assume that the Democrats are doing better on a district-by-district level with Obama-Trump voters. What does that mean for their strategy for who they nominate in 2020?
Do they pay greater attention to those red(ish) districts they’ve been winning, or is it a whole new ballgame and they’re back to square one?
nrakich: By reddish, do you mean the districts with lots of Obama-Trump voters, Clare?
clare.malone: Yes, Nathaniel.
micah: I think it’s a whole new ballgame, but because of the way voters are distributed and the Electoral College, Democrats need to go after Obama-Trump voters to some extent. Right?
nrakich: I don’t think it’s a whole new ballgame.
micah: ⚾
🏀
🗑
clare.malone: Friction. I like it.
nrakich: If Democrats do disproportionately well in Trumpy districts, that plus the pro-Republican Electoral College bias you mention means Democrats should go in hard on a Midwest/populist type.
President Tammy Baldwin, anyone?
natesilver: Keep in mind that Democrats can make a ton of gains in the House just by sweeping through the remaining GOP seats in California, New York and New Jersey. That doesn’t help at all in the Electoral College.
micah: I’m a bit surprised we’ve settled on Obama-Trump voters as the swing voters in 2018 (to the extent there is a the group) — I thought we’d have some more contrarian takes!!
nrakich: To be clear, I’m not saying they’re the ONLY swing voters.
micah: Yes you are. Well, I’ll edit this so it reads like you are.
nrakich: But I think there’s better reason to look to them than suburban folks.
natesilver: I don’t know — until fairly recently the conventional wisdom was that the Romney-Clinton voters were more important, that districts like Ossssofffffff’s Georgia 6 would be key
nrakich: Yeah, but we at five thirty eight dot com have been pushing back against that conventional wisdom for months.
micah: That’s true — credit to you all for that.
nrakich: One thing that has also changed since a few months ago is we’ve started to see more polls, and polls are getting more predictive now. As Cohn notes, polls have generally showed Democrats doing well in pro-Trump places and Republicans holding their own in pro-Clinton ones. We saw a pretty eye-opening poll of the Wisconsin governor’s race last week in which Scott Walker, the Republican incumbent, was losing by 13 points. But in a new poll of the Nevada Senate race, the Republican incumbent, Dean Heller, and his Democratic challenger, Jacky Rosen, are statistically tied. You wouldn’t normally expect that in a year when Democrats have a 7-point lead in the generic ballot.
clare.malone: Well, we did talk for a long time about Reluctant Trump Voters™
micah: Yeah, what about them?
clare.malone: We were watching them to see if they’d stray from the president, and they’ve mostly been pretty loyal.
natesilver: Yeah, but will they turn out?
micah:❓
clare.malone: Maybe this is where the contrarian turn in our convo comes.
natesilver: So far, Democrats have gotten very good results in special elections — which consist of, you know, actual voters. And they also look pretty good, as Nathaniel said, in district-by-district polls, which are mostly conducted among likely voters rather than registered voters. Those could be signs of a turnout advantage.
Or at least that Democrats won’t necessarily have the turnout disadvantage that they usually do in midterms.
micah: OK, so this is where I want to end: If Obama-Trump voters are a/the crucial bloc of swing voters, what do we know about them?
What are they like?
What messages will they be receptive to?
Where are they?
clare.malone: Well, I think you can look for them in the Upper Midwest and Rust Belt regions — Northeast Ohio, where Iowa and Wisconsin meet. From NPR:


Places that had a history of high union membership and thus strong Democratic Party ties but that were perhaps moved by the social conservatism of Reagan while still embracing populist economic policies.
Or something.
nrakich: According to a survey last year, they don’t like Obama or the Affordable Care Act, they support Trump’s policies on issues like immigration and trade, and they have, ah, “racial resentment.”
But they also support infrastructure investments and a higher minimum wage. That’s a message Democrats can run on.
natesilver: It’s not my job to tell parties what messages to use, nor do I have any special insight about messaging anyway. With that said, it’s noteworthy that Trump’s lowest approval ratings came while Republicans were trying to pass major policy initiatives such as the tax cut (successful) and their Obamacare repeal (not successful).
So I think a message framed around maintaining a check on Trump and the excesses of the Republican Congress would make sense. That’s classic midterm strategy, since midterms are all about balancing.
nrakich: Less the check on Trump thing, right? Maybe on policy.
But the Russia stuff hasn’t moved the dial as much.
I think running against the GOP Congress is a better strategy. You’ll never go wrong saying you hate Congress!
natesilver: The Russia stuff might move the dial among the Democratic base. And it likely contributes to Trump’s overall unpopularity.
Keep in mind, by the way, that most voters expected Clinton to win. So they may have elected Republicans to Congress expecting them to be a check on her. Since they got Trump instead, the balancing-type arguments could be especially powerful.
micah: But there’s some obvious tension here between (i) appealing to Obama-Trump voters, (ii) the messaging and priorities of the #resistance and (iii) Romney-Clinton voters, right?
natesilver: No.
nrakich: Yes, but it’s a midterm, not a presidential year. There doesn’t have to be one message.
clare.malone: It’s GOING to be a tense primary for Democrats in 2020. We’ll be talking about #lanes again, I think
micah: To be clear, we’re saying the 2018 battleground includes more rural, working-class, non-college-educated places, not that it excludes well-educated suburbs?
Or are we?
nrakich: Right. It’s clear that the battleground includes all sorts of districts.
Both parties are acting like, and analysts are unanimous in thinking that, both the well-educated 45th Congressional District in California and the blue-collar 2nd Congressional District in Maine are contested, for instance.
natesilver: Yeah, it’s a variety pack. But like I said earlier, I think geographically it mostly gets concentrated in the Midwest, California and New York/New Jersey.
There are a few relative layups for Democrats outside of those regions, like in Barbara Comstock’s district in Virginia.
I also would apply a relatively broad definition of “Midwest” to include places like West Virginia and Kentucky and not necessarily just the Rust Belt.
clare.malone: “Anything that touches Ohio.”
nrakich: So basically everywhere except the South.
natesilver: And there just aren’t a whole heck of a lot of seats in the Interior West.
nrakich: Right.
micah: We have an official FiveThirtyEight definition of the Midwest — it’s probabilistic.

OK, everyone give me your best kicker for final thoughts.
natesilver: The best thing you can do is to nominate viable candidates in every district where the wind might blow your way, and Democrats have done a pretty good job of that.
nrakich: NO DISTRICT IS SAFE!!!!!
micah: Clare.
…
…
🥁
🥁
clare.malone: We made you read this 3,000-word chat, but we didn’t even definitively answer the question in the headline.