What Went Down In The 2018 Midterms

One thing that keeps being noted is the absolutely blockbuster turnout across the country, and the fact that there are some places where turnout in this midterm is potentially approaching presidential year numbers. But bear in mind that a significant number of voters in pre-election polls indicated that their vote would be about the President — either in support or opposition. So, even if Trump’s name isn’t on the ballot, he is effectively on the ballot in the minds of many voters, which makes sense for why turnout might look more like a presidential year.

Polls are about to close in most of Florida, Georgia, the rest of Indiana, the rest of Kentucky, most towns in New Hampshire, South Carolina, Vermont and Virginia. In other words, we’re about to get a rush of House results. According to our forecast, there are several competitive races among them. Democrats are favored in the Florida 27th and Virginia 10th; these are probably must-wins if they’re going to flip the House. Toss-up races include the Florida 26th, Georgia 6th, Virginia 2nd and Virginia 7th. And if you scroll all the way to the bottom of the table, you’ll see that the Georgia gubernatorial race — one of this year’s most-talked-about races — also closes at this hour.

Anti-abortion groups have been pouring a lot of resources into this year’s midterms. One of their first tests will be in Indiana. Joe Donnelly is one of the last remaining Democrats who calls himself “pro-life,” but that hasn’t stopped anti-abortion groups from attacking him for voting to fund Planned Parenthood.

Social scientists are aware of a lot of cognitive biases, and we’re likely to suffer from more than a few before the night is done. One is “anchoring,” the phenomenon in which we “anchor” around early information and don’t sufficiently revise our views in light of new information. The danger here is that we develop a narrative based on the early-reporting districts and states and don’t revise it as data from the rest of the country comes in. Another is “recency bias” — that’s when we weight new information too heavily. Recency bias can crop up when you choose the last thing from a list, and it can also appear when we try to tie election-night results to events in the final days of the campaign.

Looking for signs of what’s happening with Democratic incumbent Sen. Joe Donnelly in Indiana? In Bartholomew County, which had 1 percent of the state’s vote in 2016, a bit more than one-third of its precincts have reported. Trump won it by 33 points in 2016, and Donnelly lost it by 8 points in 2012. Currently Donnelly trails there by 9 points, which is close to his 2012 result when he won statewide by 6 points.

As we watch the results from Indiana roll in, take a moment and watch what it looks like to run for the Senate as a Democrat in the state. Hint: It looks an awful lot like running as a Republican. Polling has shown Sen. Joe Donnelly doing well among red-state Senate Democrats this cycle and the posture he takes in this ad is probably part of the reason why. Liberal Democrats may not like the looks of it, but they probably like the looks of keeping Donnelly’s seat.

Janie just wrote about what the early exit polls say about immigration — and they show 48 percent of respondents saying Trump’s policies are too tough as opposed to 32 percent saying they are about right. I’d like to add an additional data point on immigration, and one that’s not subject to change.

Using the 2016 wave of a nationally representative panel survey I help oversee along with my colleague Diana Mutz, I previously reported that Americans’ immigration views actually became more liberal in the fall of 2016. And that’s actually not all that surprising. It’s a common finding that political scientists call “thermostatic response” — when policy starts moving in one direction, public opinion often adjusts by moving in the opposite direction.

And when we look at data from the latest wave of that panel survey, I found that respondents’ support for restrictionist immigration policies remains substantially below its 2012 high-water mark, and roughly on par with where it was in October 2016.

So while there’s often an assumption that presidential rhetoric is persuasive, there is extensive evidence in political science for the reverse may hold instead: the public may gravitate away from the president’s position.

Indiana and Kentucky are both counting their votes S-L-O-W-L-Y. We have 1 percent to 3 percent of precincts reporting in a few districts like the Kentucky 6th and Indiana 9th, but anyone trying to read into those results is trying too hard.

Here’s some rich irony coming out of Georgia: Brian Kemp, the GOP candidate for governor and the man in charge of overseeing the state’s elections as its secretary of state, had trouble voting today, according to a local news station that was tagging along with him. Kemp has come under fire from his Democratic opponent, Stacey Abrams, who has accused Kemp of trying to suppress the votes of Georgians.

Democrats are winning the debate on health care, according to early exit polls: 58 percent of voters say they trust the Democrats to protect health care for people with pre-existing conditions, while 34 percent trust Republicans. The findings line up with what poll results have previously told us. Again, these numbers are preliminary and are subject to change as more votes come in.

OK, there’s a little bit of a lull here on results so let’s talk about this tweet:

Republicans of a certain ilk are extremely excited to see one Willard Mitt Romney be elected the next senator from Utah. Remember Mitt? He ran for president once and then when he lost, his party did this whole post-mortem about how they needed to win over more minority voters and women, and well … things turned out pretty differently the next time they nominated a guy for president.

Romney and Trump have a tense relationship and it will be interesting to see that dynamic play out with Romney in the Senate. Many of the Republicans who continued to vocally oppose Trump are on their way out of the Senate (Jeff Flake, Bob Corker) or have gotten more friendly with the president (Lindsey Graham). Which path will Romney take? I think he’ll probably go more the Corker/Flake route, and that could be interesting to see develop over the next two years.

With the large, glaring caveat that exit polls are still fluid and we should be cautious about drawing conclusions from them this early, if the figures hold, it would suggest that Democrats can expect a good evening. But it would also suggest a reasonably good night for pollsters, as the job approval figure and generic ballot figure in these early exits are not far off from the final polling averages. This also does not mean anything about the potential accuracy of polls in state and local races at this point. (Bear in mind that in 2016, the national polls showing Hillary Clinton with a slight lead were accurate in the sense that she did win the popular vote by a slim margin, but the state polling in many key “blue wall” states was what was off by sizable margins.)

Trump has doubled down on his immigration policy in the weeks leading up to the election. Early exit poll data tells us that a plurality of voters, 48 percent, said his immigration policies are too tough, 32 percent said they’re about right, and 16 percent said they’re not tough enough. Once again, these numbers are not definitive and could change throughout the night, as we get the final election results.

Seth: Julia, as you and I have written elsewhere, the end of an election marks the beginning of a campaign to define the narrative of the election. What do you see being some of the lessons the parties choose to learn from this election? For example, if Democrats have a good night, what will Democratic activists learn from that going into the 2020 election cycle? It seems like one of the immediate interpretations many Democrats had of the 2016 election was that they had to avoid “identity politics” and maybe just nominate a white guy (because that’s not an identity — 🙄) for president. But if, say, several African-American candidates win statewide elections in the South tonight, which is a real possibility, would that shift the narrative? Would it give a different set of presidential candidates an advantage in the 2020 nomination?

Julia: I think it’s possible that it will shift the narrative about who is “electable.” One question is how many such narratives — about women, about people of color — one election can carry. Intersectionality is a sophisticated concept, and election narratives aren’t always that nuanced. This is especially true about midterms, and maybe especially about one in which Trump has dominated the story. Individual candidates may not be able to break through that story, and maybe the prospects for 2020 won’t have changed all that much after all.

On the right rail of this page — just over there → — you’ll see live-updating forecasts of the House and the Senate. We’re very excited about these — but unfortunately, I’ve run out of time to give them a proper introduction. So here are the basics:

• The projections start with the FiveThirtyEight Deluxe model as an initial projection.
• They’ll update in real time based on three types of information:
  • Votes, specifically the topline numbers in each state and district. As more votes come in, the more the model weights the vote as compared with the pre-election projection. We won’t be using precinct or county-level returns, but the model is smart about how much to weight the vote count against its priors — it will stick to the pre-election projection more in states where there’s big regional variance in the vote.
  • Projections by the ABC News Decision Desk. (What other networks sometimes describe as “calls.”) Once a race is called, it’s treated as having a 100 percent likelihood of going to the respective party.
  • Finally, the model makes inferences about states and districts where there aren’t many votes from states and projections where we do have votes and projections so far. If Democrats are overperforming in Indiana, for example, it will assume they’ll also slightly overperform in other states that look like Indiana.
• The challenge here is to strike a balance where we avoid both Type I errors (false positives) and Type II errors (false negatives). We suspect that some of our competitors will have a more aggressive approach and that this model will be on the more conservative side. But it’s more aggressive than it was in 2014 and 2016, when the model used only called states and not live votes.
• Technically speaking, we are running a probabilistic projection for each state and district that updates in real time — you can find them if you look carefully — but that’s not really what the emphasis of the project is. Instead, it’s to project the likelihood of each party winning the House and the Senate and the number of seats they’ll pick up.
• Over the course of the night, the range of possible outcomes will narrow; by design, the range will be very wide initially.
• We’re reserving the right to insert our own characterizations of a state if we know that the vote so far is not representative. For example, if the Republican is narrowly ahead in Florida but the remaining vote is from Democratic-leaning counties, we may redesignate the state as a toss-up instead of going with the algorithm’s view.

Like Nathaniel mentioned, the polls in Kentucky are closed. I’m watching the Kentucky 6th (like Geoffrey!), where Iraq War veteran Amy McGrath is running against incumbent Andy Barr. McGrath is one of many female veterans running for Congress this year. Her campaign ad, “Told Me,” went viral earlier this year — it has been viewed 1.9 million times on YouTube and shared extensively on social media.

One of the big questions I’m going to be paying attention to tonight is whether Republican drawn House maps can stem Democratic gains. Remember that tonight’s battle for the House is waged on maps that were largely drawn in 2011, when Republicans had full control of many state governments and could draw the lines. Map drawers usually account for swings in the electorate as part of the process, but tonight will be something of a first test for those lines. There hasn’t been a major swing in the electorate, in either direction, since the current maps were implemented.

Americans seem to have voted with health care on their minds. According to early exit polls, 41 percent said heath care is the top issue facing the country, followed by immigration (23 percent), the economy (21 percent) and gun policy (11 percent). Voters were asked about only these four issues.

Again, this is early data and should be consumed with caution — it could change throughout the night as more votes come in.

Tonight’s elections are expected to have high turnout, especially for a midterm. So what makes people decide to show up to vote?

Political scientists have theorized for a long time that a sense of civic duty impacts one’s decision to vote. More recently, political scientists have found that social pressure is effective in getting people to the polls. But which is more effective: shame or positive re-enforcement? Is telling people that by not voting they’re shirking their civic duty a good way to motivate them?

One study suggests this might actually be the case — sending people flyers that pointed out that they had missed the last election appeared to work better than more positive appeals. So should people tell their friends that people will know if they don’t vote? Or that they have no right to complain if they stay home? It’s possible, though it may not be great for your social life.

Polls are closed now in Kentucky. Geoffrey Skelley says he’ll be watching the race in the 6th District.

We’re seeing the first actual votes in Indiana and Kentucky! Still too few to draw any conclusions from, though.

Before actual results come in, here’s a look back at some electoral maps based on analysis my colleague Geoffrey Skelley worked on imagining if only women, men, non-college educated whites or people of color voted.

With the continued caveat that all of this might change, here’s some more early exit poll data:

• 44 percent of voters approve of Trump, and 55 percent disapprove of him.
• 41 percent support impeaching Trump, while 55 percent oppose that.
• More voters want to see Democrats in control of the house than Republicans (53 percent to 43 percent).
• Views of the Democratic party (50 percent favorable and 46 percent unfavorable) are better than that of the Republican party (43 percent favorable and 54 percent unfavorable).

Our friends at ABC News are also providing live election night coverage. Follow them here.

Polls are now closed in the first precincts of the night — in the Eastern time zone portions of Indiana (except Monroe County) and Kentucky. According to our forecast, the two most suspenseful races there are the Indiana Senate race (where Democratic Sen. Joe Donnelly basically must win for Democrats to have a shot at winning the Senate) and the Kentucky 6th District, which pits a viral Democratic star (fighter pilot Amy McGrath) against GOP Rep. Andy Barr.

And in more polling-places-being-kept open-later news, a bunch of polling places in Harris County, Texas (home to Houston), have been court-ordered to stay open an hour later. That will be good news for Beto O ‘Rourke’s team.

Our Election Bot will publish ABC projections to our live blog as results come in. Here’s how it’ll work:

As they did in 2016, the Google News Initiative is visualizing midterm searches with their Electionland app. Along with Pitch Interactive, Google has created two real time maps of search interest — one on polling places and one on voting issues like long wait times, voter intimidation and provisional ballots. As searches spike above the national average, circles appear on the chart, sized by the intensity of the flurry. Along the bottom of the app you can see a time chart of the search volume.

The project is part of a collaboration with ProPublica and hundreds of newsrooms across the country to monitor and report on voting issues. Whether searches for polling places will track with increased turnout rates remains to be seen, but it’s an interesting experiment in real-time monitoring and accountability.

Increased turnout in the midterm election is the product of many factors, but consider this: Outside groups trying to influence House and Senate races have reported spending more than $30 million on GOTV activities this cycle. That total includes special elections, but it’s still a huge jump from the reported $8.2 million spent by such groups during the 2014 election (such spending was up for congressional races in 2016, but there also was that matter of a presidential race, too).

A crucial Indiana county for Democrats — home of Indiana University’s flagship campus in Bloomington — is keeping polls open an extra hour.

So far, early exit polls tell us that Trump is a big factor in this election: 39 percent say their vote is in opposition to Trump, while 26 percent say their vote is in support of him.

Like Clare mentioned, the “year of the woman” narrative is pretty baked in to this election cycle. But as experts note, the surge has been a little lopsided — Democratic women outnumber Republican women. Of the female congressional candidates I’m watching tonight, just 23 percent are Republican.

T-minus 10 minutes until the first poll closures.

A lawsuit was just filed against Georgia Secretary of State Brian Kemp, who is also running for governor, to keep him from presiding over the election while he’s a candidate. Kemp has come under fire for a wide array of election-related issues, including accusations of voter suppression.

During the primary season, Nathaniel Rakich and I tracked endorsements by groups, both liberal and conservative. One thing we didn’t track was celebrity endorsements, but there have been a lot of them. Here are a few that have rolled in this month:

• Taylor Swift endorsed Phil Bredesen, the Democratic Senate candidate in Tennessee.
• Beyoncé endorsed Beto O’Rourke, who is challenging Republican Sen. Ted Cruz in Texas.
• Rihanna endorsed Andrew Gillum, the Democratic candidate for governor in Florida.
• Dave Chappelle made a video for Ben Jealous, who is the Democrat running for governor in Maryland.

There’s a pre-existing narrative for this Election Day that it will be another “year of the woman.” Some people have problems with the framework, and I’m sure I’ll get into them more as the night goes on, but just a heads-up that we’ve just seen Guam elect its first female governor.

Folks have been noting the enormous quantity of money spent in this election cycle and wondering what kind of effect it will have. Here’s an interesting item: Colorado’s Proposition 112, which would expand the minimum distance for oil and gas projects near occupied buildings, is strongly opposed by the energy industry. Supporters of the initiative have spent $1.6 million; opponents have spent more than $31 million. A recent poll has it marginally favored 52-48. It’s possible that overwhelming spending advantage by opponents has limited the initiative’s appeal, but it’s also remarkable that a campaign that has been outspent 20:1 may actually win tonight.

A lot of people (including at least one election official) seem to find the long lines that have been reported today exciting — a sign of voter enthusiasm. But while it certainly shows voters’ determination, this is really evidence of an under-resourced election system.

Now that our night shift crew has punched in, but while we’re still waiting for the first poll closings: Has anything happened today, or have you learned anything new today, that has caused you to update your read on the races tonight?

Before we begin to get results, I just want to say that I hope everyone enjoyed this election, because it wasn’t cheap. The Center for Responsive Politics estimates that more than $5.2 billion was spent this cycle, making it the most expensive midterm election in history. Previously no midterm surpassed $4.2 billion in spending, when adjusted for inflation.

Hillary Clinton’s 2016 campaign manager, Robby Mook, knows all too well that early exit polls can be misleading. On election night two years ago, early exit poll results suggested that Clinton was winning against Trump. Those projections changed as the night went on and more votes were tabulated. So, again, please be cautious of those early exit polls tonight and don’t take them too seriously until the final votes are in.

Putting a polling place inside a gated community, and requiring people to show ID to enter that community even though they don’t need it to vote, seems like a real problem.

The wave of scientists and science-trained people running for public office for the first time has boiled down to a relatively small number of races. There are about six congressional candidates who fit this profile and are favored to win tonight:

• Kim Schrier in the Washington 8th
• Jeff Van Drew in the New Jersey 2nd
• Sean Casten in the Illinois 6th
• Lauren Underwood in the Illinois 14th
• Chrissy Houlahan in the Pennsylvania 6th
• Josh Harder in the California 10th

I’ll be keeping an eye on what happens to them. But they’re obviously not the only candidates out there who care about evidence-based policy and the importance of science as a tool for discovery.

Part of the problem with defining science candidates is that it inevitably leaves out some people who feel like they should be counted. Take, for example, Haley Stephens, who is favored to win the Michigan 11th tonight. She’s not counted as a science candidate by either the magazine Science or political scientist Matthew Motta, but when she and I spoke this summer, she definitely framed herself that way — highlighting her long advocacy for STEM education and technology-based jobs.

A good PSA from friend-of-the-site Decision Desk HQ, for those of you who are following the election with multiple sources:

Hello! And Beep boop…

I’m Results Bot, and I will be watching the midterm election results with you tonight. I will be informing you when ABC News projects winners for House, Senate and gubernatorial races tonight, or if FiveThirtyEight determines that a race will go to a runoff.

There is one thing you should know about me: I’m an emotional bot. I will be comparing the results of each race to the Deluxe version of FiveThirtyEight’s forecast that were expected as of this morning. For example, if the Deluxe forecast had a race listed as “lean Democratic” this morning before results started to come in, but ABC News projects that a Republican will win, I will be surprised (😮); if we had forecasted the race be a likely win for one party and it goes for the other, I will be shocked (😲); and if ABC News projects a race for a party that we had listed as a solid win for the other, I will be ALARMED (😲🚨)!

Thank you for joining me and my human colleagues this evening.

At some point tonight, you’ll see the results from some random House district start to trickle in. Maybe the Republican candidate is winning big. Well, what does that mean? Here’s one (somewhat) simple method to figure that out.

We lined up every House district from the most likely to be won by a Democrat (according to our final Deluxe forecast) to the most likely to be won by a Republican, and then looked at how many seats Democrats would gain if they won a given seat and only seats with equal or greater Democratic odds. Here are the districts where the House will be decided (those rated as Likely, Lean and Toss-up in our forecast):

Of course, it’s very unlikely that races will be decided so neatly. But as results come in, you can use the table above to take stock of how the night is going.

Zooming out a bit, you can see why Democrats are favored to win a majority — if Democrats win every race that has a 60 percent or greater chance of going their way, and no others, they will pick up enough seats to win control of the House.

In case you missed it, a lot of women are running for Congress this year — 260, actually. And today we plan to keep an eye on how many win. We are also going to be watching for how LGBT candidates, veterans and candidates who self-funded their campaigns fare.

Here’s what current representation looks like (and could look like after tonight) for these groups:

• Of the 535 current members of Congress, 107 are women. According to the final Deluxe version of our forecasts, if every woman currently leading in her race won, women’s representation would increase from 107 women to 124, with most gains coming in the House.
• Seven of the 535 members of Congress openly identify as LGBT. But if all the LGBT candidates who are currenting leading in their races were to win, that number would increase to 10. We’ll be tracking these candidates using data from the Victory Fund, an organization that works to elect LGBT candidates, and one person we’ll be closely watching is Sharice Davids, a former White House Fellow and an ex-MMA fighter. Davids is running for Congress in Kansas, and would be the first Native American lesbian elected to Congress, and the first LGBT person to represent Kansas if she wins. Davids is forecasted to win 51.9 percent of the vote, so the race will be close.
• Historically, veterans have been well represented in Congress, but according to the Pew Research Center, their numbers have undergone steady decline since the early 1950s, shortly after World War II. As of 2017, 20 percent of Senators and 19 percent of House members were veterans, down from 70 percent and 48 percent in the early ’90s. So, using data from With Honor, a group that helps veterans win office, we will be watching to see whether that decline continues or shows signs of reversing.
• We are also paying special attention to self-funders because, in the age of small-dollar donations (which have fueled many high-profile races), we are interested in whether millionaire business owners and entrepreneurs willing to spend their own money are being left behind or are doing well. We’ve identified 41 congressional candidates who have given or loaned at least $400,000 to their own campaigns, according to Federal Election Commission data.

So stay tuned as we track how these groups do tonight.

To add onto what Perry said: I think Beto’s best shot at the presidency is if he loses his Senate race by, like, half a point. He’ll have come tantalizingly close, but he won’t have a day job to prevent him from actively running starting tomorrow.

Ollie, thinking about those scanners backing up made me wonder how old the voting machines in New York are. Usually, when we talk about aging vote technology we end up talking about the risk of hacking — but it goes beyond that. The older the machines are, the more likely they’re just going to run into basic maintenance and equipment failure problems, and the harder it will be to find replacement parts. Ten or 15 years is kind of the lifespan for these machines. But, according to a report by the Brennan Center, most states are working with machines that are already a decade old.

Greetings from Guam,* where it’s already 8:00 a.m. on Wednesday and Democrats have scored their first win of the night. Democrat Lou Leon Guerrero defeated Republican Ray Tenorio to become the next governor of the U.S. territory.

*I am in Guam in my imagination.