What’s Going to Happen

Yesterday, I made the case that the most likely result is 228 Democrats in the House, giving them a notable but not gigantic majority. The math here is straightforward. House elections are basically referendums on the president filtered through a set of 435 distinct match-ups in a wide variety of districts. That map favors the Republicans—a combination of geography and gerrymandering allows the GOP to lose the House popular vote by a few points without losing the chamber. But public opinion favors the Democrats: Trump has a below-average-but-not-unheard-of approval rating, so we should expect this to be a worse-than-average midterm for the GOP. And when you do the math, it’s easy to get somewhere in the high 220s or low-to-mid 230s for the Democrats.

There’s a lot of error around that estimate. If the polls we’ve been seeing all cycle are systematically overestimating the Democrats, then Republicans could still hold the House, though their seat margin would almost certainly be smaller than it is now. At the same time, a systematic polling error could benefit the Democrats—meaning that a true landslide victory where Democrats end up with a 250 plus seat count is also plausible. Some of the late polls have looked good for Democrats, but we won’t have a good way to know if that was signal or noise until we get results.

Different results will lead to different conclusions, but Democrats are almost certainly going to win the House popular vote—which at some level counts as the public pushing back on the Trump administration. But that’s not how we decide who gets legislative power in this country. Seat count matters, and on some level a W is a W for both parties. If Democrats take the House, they had a good night (possibly a great night depending on the final seat margin) and if they don’t (or if they just barely win) then it was a good night for Republicans. I got further into the weeds on the politics and policy around this here, but I think these topline interpretations serve as helpful shorthands.

SENATE 2018

On the Senate side, we can rely on the forecast from SwingSeat, THE WEEKLY STANDARD’s stats-driven Senate model. The topline prediction of the model is that the GOP ends up with 52 seats, with room for them to get into the mid-to-high 50s or barely lose the chamber. If we get last minute polling we might update this again, but I doubt the topline will change much between now and when polls close. In terms of probability, the GOP has an 85 percent chance of holding the Senate. You can get a more detailed view of the probabilities by looking at this histogram:

This histogram summarizes the topline results. The graphic is pretty simple—each bar represents a result (e.g. the GOP ends up with 49 seats, the GOP ends up with 52 seats, etc.) and the height of each bar corresponds to the likelihood of that scenario actually happening. The red bars are scenarios where Republicans win and the blue bars are scenarios where Democrats take the chamber.

There’s a lot of mass between 50 and 55 here—meaning that the most likely outcomes are in that range. But other outcomes are on the table.

The model also produces state-by-state projections, which are summarized here.

Each row is a race. The second column is our official model’s projected win probabilities (based on polls and other “fundamentals” such as incumbency, past election results, presidential approval, and more), and the third column is an older version of the model that just looks at polls. The full model is SwingSeat’s official projection, but it’s interesting to compare it to a version that just looks at polls.

I’m not a person who thinks that models are perfect oracles that can’t be argued with, but I generally like where these probabilities are. Arizona, Nevada and Indiana are truly very close. West Virginia, Montana, Texas, Florida and Tennessee each feature a strong favorite, but none are immune from an upset. In fact, candidates who are 4-in-5 favorites to win end up losing, well, one-in-five times. So we should expect some real upsets—it’s just hard for the model (or us) to pick those upsets with certainty ahead of time.

The model really likes Kevin Cramer’s chances in North Dakota (which makes sense given the eye-popping margins he’s posted since the Kavanaugh hearings). If I had to pick one race that keeps me up at night, it’d be that one: We don’t have a ton of recent data, and I do wonder if there’s been undetected movement there. There are other races where my gut is different from the model and I’m tempted to nitpick (e.g. I would knock Cruz’s win probability down a bit) but one of the goals of the model is to check your gut. And I generally think these probabilities are good.

The model’s most likely guess is that Republicans hold the safe states plus Texas, Tennessee, North Dakota and Missouri plus one upset to get to 52 seats. If I had to guess where the GOP might score an upset, I’d go with Indiana. But, again, it’s hard to pick upsets.

That being said, this is a probabilistic forecast—SwingSeat *expects* some upsets and tries to price that into the topline. So we shouldn’t think of it as a set of hard binary predictions—it’s an empirical forecast that gives you the gradations of risk and advantage for different candidates.

These results will be harder to interpret than the House results. In the House, we have obvious baselines—winning the House popular vote or winning a majority in the chamber are generally signs that a party has done well. But only a third of the Senate is up, so it’s harder to set those baselines.

That being said, if Democrats take the Senate or hold the GOP to 50 seats, that would count as good night for them. The 2020 Senate Map looks much better for the Democrats on balance (the GOP will be defending more territory, some of which is swing-y or leans Democratic), so if Republicans really want to feel good about holding the chamber in the medium term, they should try to at least get into the low to mid 50s. A result in the 51-52 seat range is decent for the GOP but doesn’t provide them with a ton of padding for 2020.

GOVERNORS 2018

Governor elections are a little bit weird—national politics isn’t as important on the state level as it is on the federal level, so you can get some wonky results (e.g. a Democratic governor in Montana and a Republican governor in Massachusetts. That being said, I think the most likely result is that Democrats net six to eight governorships, some of which are in big states.

The logic behind this is simple: Republicans won a lot of governor elections in 2010 and 2014, so the GOP is now overexposed in a year when national headwinds are against them. On a state-by-state basis, that probably looks something like the Republicans picking up Alaska, while Democrats safely pick up Michigan, Illinois, Maine, and New Mexico. After that, there’s a bevy of toss-ups. Democrats could win in Florida, Georgia, Wisconsin, South Dakota, Iowa, Ohio, Kansas, and Nevada. I tend to think that Democrats will only win some of these (Florida, Wisconsin, Ohio, and Nevada seem like the likeliest of the bunch, but I could envision other scenarios). If you add those odds up and account for a potential GOP upset in Connecticut or Oregon, you get a net Democratic gain of about six to eight.

The spread of possible results here is wide. A terrible night for the GOP could end with Democrats netting more than 10 governorships. That sort of night would put Democrats in charge in larger states where they’d have a bigger influence on the next cycle of redistricting. Alternatively, I could foresee a great GOP night involving a net loss of only a couple states (e.g. Democrats take New Mexico, Illinois, Michigan, and Maine, but lose Alaska).

The right interpretation of these results is going to be a little bit tricky because the standards we use in congressional races don’t apply here. If your party gets a majority of governorships . . . nothing happens. And a Democratic or Republican win in a red or blue state often doesn’t signal an impending realignment. It can be a one-off, weird result fueled by state and local concerns. So I would be watching a few things: the policy views of the new governors, the bigger states (because of redistricting), and how these elections shape the farm team.

The policy views of various candidates will be interesting here. Some Democratic candidates are very liberal (e.g. Gillum, Evers) and some of the Republicans are very conservative (e.g. Walker) or extremely Trump-y (e.g. DeSantis). A lot of this country’s laws are state and local, so these governors are going to have a lot of influence on the actual day-to-day lives of people.

It’s also worth watching who gets which governorships. Holding onto Wyoming isn’t as big of a deal as holding onto a state like Michigan, because Wyoming only has one congressional district. Congressional redistricting is just around the corner, and many of the governors elected tomorrow will have a big influence in that process.

And finally governors (like Senators and House members) are the farm team for future administrations. I’m not just talking about who will/won’t run for president here: key Cabinet positions, ambassadorships, and other top positions often go to governors who show that they’re smart and competent. Sometime in the next two to six years, someone not named Trump (whether they’re a Republican or a Democrat) is guaranteed to become president. And some of the governors elected tomorrow will almost assuredly join his or her team.