Sabatos Crystal Ball

How’d we do?

Longstanding picks of a Democratic House, significant Democratic gubernatorial gains, and GOP Senate pan out

Larry J. Sabato and Kyle Kondik, Sabato's Crystal Ball November 7th, 2018


It took a lot of Krazy Glue, but we think we pieced the Crystal Ball back together, reassembling after 2016 shattered us and just about every other prediction group.

As of this writing, early Wednesday afternoon, and with many uncalled House races remaining, the real-time seat projections from both the New York Times and FiveThirtyEight were suggesting that the Democrats would win a 229-206 majority in the House, for a net gain of 34 seats, exactly the seat change we picked in our final selections.

Democrats built their new majority in part by persuading voters in many Republican-held districts carried by Hillary Clinton in the 2016 presidential election to elect Democratic House members. Of the 25 Clinton-won GOP districts, Democrats have won at least 14 and very likely will win several more. But Democrats also will win a similar number of districts won by Donald Trump, including upsets against Reps. Dan Donovan (R, NY-11) on Staten Island and Steve Russell (R, OK-5) in Oklahoma City. Another surprising Democratic win came in Charleston after Rep. Mark Sanford (R, SC-1) lost his primary. But for the most part, the seats the Democrats flipped were ones that we projected to flip.

If indeed the final tally in the House is 229 Democrats and 206 Republicans, that sets up another competitive battle for the House in 2020, when Republicans (under this scenario) would need to net a dozen seats to win back the majority. A good thing for Democrats is that many of the suburban seats they picked up in this election — CO-6, MN-3, VA-10, and others — probably will be relatively easy to hold with Donald Trump on the ballot, and the Democrats did not max out their potential seat gains. Meanwhile, Republicans will focus on unseating several Trump-district Democratic incumbents they were unable to credibly challenge this year and clawing back some of their losses in less Clinton-friendly suburban areas as they plot their own path to restore their shattered majority.

We spent the whole cycle arguing that the lopsided Senate map, one of the worst that any party has had to defend in the history of Senate popular elections, made the Republicans considerable favorites to hold the upper chamber and potentially even add seats, and that’s what happened. While Arizona remains uncalled (and Florida probably is headed to a recount, although the GOP appears to be in the driver’s seat), it seems like the GOP could be headed for a three-seat net gain. If Republicans do get to three, it would give them 54 Senate seats, meaning Democrats would need to win at least four seats to flip the Senate in 2020, depending on which side wins the presidency. As we noted prior to the election in a lookahead to the 2020 Senate map, it will be hard for Democrats to scrape together such a gain next time. That’s particularly because Tuesday night’s results, where three dark-red state Democrats in Indiana, Missouri, and North Dakota were wiped out and two others in Montana and West Virginia only won narrowly, are a very bad sign for the lone remaining Democrat in a heavily GOP state, Sen. Doug Jones (D-AL), who will face the voters in November 2020. (A million things can happen before 2020 — we get it.)

A couple of supposedly close Senate races that the Crystal Ball never rated as Toss-ups, New Jersey (D) and Tennessee (R), ended up not being close. Meanwhile, we warned readers on Monday to watch out for a more impressive result than expected from Beto O’Rourke (D) in Texas, and he turned in a strong showing, losing by about three points. His performance had to have helped the Democrats net two House seats and contributed to other Democrats giving the GOP a scare in several more. Texas may not yet be reaching swing state status, but the dormant Democratic Party there is regenerating in the Trump era.

There will be one more Senate race, a Nov. 27 runoff between Sen. Cindy Hyde-Smith (R) and former Clinton administration Secretary of Agriculture Mike Espy (D) following Tuesday night’s all-party election. The combined two-party vote in the first round of voting was approximately 58% Republican and 42% Democratic; our Likely Republican rating there remains operative.

Meanwhile, Democrats appeared to net seven governorships, an impressive tally although not quite as big as Democrats would have hoped. Still, flipping Illinois, Michigan, and Wisconsin gave the Democrats some marquee victories. Republicans are heartened to have held two big prizes, Florida and Ohio, although they also apparently found themselves on the wrong end of yet another very close gubernatorial election in Connecticut for the third consecutive election. That pales in comparison, though, to Democratic agony in Florida: The Democrats have lost three straight Sunshine State gubernatorial races by about a point each time. As of this writing, it appears as though Georgia Secretary of State Brian Kemp (R) might avoid a runoff in his race, the one contest we left as a Toss-up in our ratings. Republicans also held their own losses to half a dozen by winning Alaska, but because the Last Frontier has an independent governor, that does not impact the Democrats’ seven-seat net gain.

It may be several weeks before the final outcome is known, at least in some of the closest House races.

But very soon it will be on to 2020. Expect Democratic presidential candidates, and there will be many of them, to start jumping into the race before the year is through. If they’re as smart as many appear to be, they’ll step back first and study the 2018 results. Some candidacies will be encouraged, but others will not. Future Crystal Balls will elaborate!

Thanks again to the hundreds who helped us see through the fog over the last two years. We deeply appreciate it, and your feedback will always be welcome.

Final picks for 2018

Democrats in House; Republicans in Senate; Democrats big in gubernatorial races; be on guard for upsets

Larry J. Sabato and Kyle Kondik, Sabato's Crystal Ball November 5th, 2018




— Sorry, friends, but you are going to have to actually read this one.

— Our full list of ratings changes is available here.

Our best guesses for Tuesday

The 2018 midterm has long been a study in contradictory signs. There is, for Republicans, the benefit of running at a time of relative peace and prosperity. Unpopular wars and economic recessions have spelled doom for the president’s party in many past midterm elections. But then there is also the weak approval rating of President Trump, who thanks to his deliberately polarizing style has kept the GOP base in line but strongly alienated Democrats and, perhaps more importantly, independent, swing voters. Democrats have held a steady lead in the high single digits on the national House generic ballot polling, a lead suggestive of a potential House flip but not one large enough to indicate that such a flip is an absolute lock.

There is the shifting political landscape that emerged nationally in 2016, with some traditionally Democratic blue collar small cities and rural areas across the North moving toward Trump and the Republicans, and some traditionally Republican suburbs dominated by voters with high formal educational attainment breaking sharply away from Trump and the GOP.  Those latter areas make up a significant share of the competitive House districts, many of which seem poised to deliver for Democrats on Tuesday, although some Trumpy, traditionally Democratic turf is part of the Democratic House calculus too.

There are the competing maps in the battles for major statewide offices. In the Senate, Democrats are defending 26 seats while Republicans are only defending nine. In the gubernatorial races, the Republicans are defending 26 seats while Democrats are only defending nine. We’re expecting vastly different overall results in the Senate and gubernatorial contests.

Our expectations for this election have been consistent for the past several months. We favor the Democrats in the House and Republicans in the Senate, and we expect the Democrats to pick up a significant share of governorships. Our picks[1] follow.

We’ve also included an “upset watch” listing in each category to flag some potential surprises that we didn’t pick in our ratings but that might emerge on Election Night.


Table 1: Final Crystal Ball House ratings

Note: Districts shaded by color of current party control (red for Republicans, blue for Democrats.

Our ratings changes leave 229 seats at least leaning to the Democrats and 206 at least leaning to the Republicans, so we are expecting the Democrats to pick up more than 30 seats (our precise ratings now show Democrats netting 34 seats in the House, 11 more than the 23 they need). We have long cautioned against assuming the House was a done deal for the Democrats, and we don’t think readers should be stunned if things go haywire for Democrats tomorrow night. That said, it may be just as likely — or even more likely — that we’re understating the Democrats in the House. Many of our sources on both sides seemed to think the Democratic tally would be more like +35 to 40 (or potentially even higher) when we checked in with them over the weekend.

Those who think the Republicans can or will keep the House will think we are being overly aggressive in some of our ratings. For instance, we now have Democrats favored in all four of their takeover opportunities in New Jersey. If Democrats come through, they will hold all but one of the House seats in the Garden State, sending the GOP to a low in the state’s congressional delegation not seen in more than a century. Democrats netting three seats in Virginia might also seem high to some, along with Democrats netting four combined seats in heartland states Iowa and Kansas.

On the flip side, those who think the Democrats will win the House comfortably will quibble with some of the seats that we’re picking the Republicans to hold. We don’t have Democrats winning any new seats in Georgia, where Democrats are hoping to net a couple of suburban seats, and we have Democrats netting only a single seat in Florida despite the party having a few other credible targets there. If Democrats netted three seats in Pennsylvania, as our ratings indicate, that might be a mild disappointment for them, too. We have something of a split decision in California, with Democrats picking up four seats, a little short of their ideal scenario. Those looking for the Democratic number to go higher might look to these places to exceed our expectations.

Upset watch: Speaking of California and Florida, Reps. David Valadao (R, CA-21) and Carlos Curbelo (R, FL-26) have long seemed like two of the most secure Republicans in districts Hillary Clinton won, and yet both may have been ill-served by the president’s hard emphasis on immigration down the stretch. Curbelo losing actually would not be much of a surprise; some Democrats expect it. Valadao losing would be more surprising. Reps. Don Young (R, AK-AL), Duncan Hunter (R, CA-50), Steve King (R, IA-4), Greg Gianforte (R, MT-AL), and Chris Collins (R, NY-27) are all Republicans in usually safe seats who are being pushed for myriad reasons; it may be asking too much for all of them to win, but that is what our ratings suggest.

Also, watch Reps. Mike Kelly (R, PA-16) and Scott Perry (R, PA-10), who both face credible opponents in redrawn, Trump-won seats. It wouldn’t surprise us at all if one lost. Two Republicans who won very close special elections earlier this cycle, Reps. Karen Handel (R, GA-6) and Troy Balderson (R, OH-12), are also right on the edge of losing.

While his race has attracted zero attention, Rep. Collin Peterson (D, MN-7) represents the most Republican district held by any Democrat — Trump won it by about 30 points. It would be shocking if Peterson lost, but it would make sense in the larger scheme of things. Meanwhile, also in Minnesota, most are assuming Republicans pick up the open MN-8, a Trump-won seat covering the Iron Range. And yet the district’s Democratic DNA is deep enough that a Democratic hold remains a possibility. And an open Democratic-held seat, NH-1, is always competitive and has not been quite as easy of a hold as Democrats might have hoped.

For a list of all 435 Crystal Ball House ratings, please take a look at the chart at the bottom of our ratings page.


Map 1: Final Crystal Ball Senate ratings

Because of the bad map Democrats faced this year, the GOP picking up seats always seemed like a possibility, even a strong possibility. Our final ratings reaffirm this potential; we have 52 Senate seats at least leaning to the Republicans, and 48 at least leaning to the Democrats. If that happened, the GOP would net a seat.

The potential GOP gain would come from places that make sense: We have them favored in three of the five strongly Republican states that have Democratic senators running for reelection: Indiana, Missouri, and North Dakota. Meanwhile the two Republican-held seats where we now favor Democrats, Arizona and Nevada, are much more competitive states at the presidential level and thus are susceptible to Democratic takeovers in a challenging environment for Republicans.

The reasonable range of outcomes in the Senate still seems fairly wide, with a bigger GOP gain possible, or no gain at all or even a Democratic gain. The Democrats still essentially have no path to the majority without winning one of these three states: North Dakota, Tennessee, and Texas, and the Republicans retain what appear to be edges in all three.

Upset watch: An overall upset pick to watch would be the Senate majority itself: Democrats winning everything where they are currently favored, plus Indiana and Missouri, and then one of North Dakota, Tennessee, or Texas. It’s not likely but it is possible: Just this morning, NBC News/Marist showed Sen. Claire McCaskill (D-MO) up by three points, and the same pollster had Sen. Joe Donnelly (D-IN) up a couple of points last week.

And then there is the Lone Star State. We have been flooded with messages from credible contacts in Texas, from both sides of the aisle, warning us not to discount the possibility of an upset by Rep. Beto O’Rourke (D, TX-16) against Sen. Ted Cruz (R-TX). The energy on O’Rourke’s side, they say, is palpable. This all may be reminiscent of the grassroots energy that helped power Trump himself to victory in 2016. Of course, analysis by anecdote isn’t always the right formula; while measuring crowd size might’ve helped navigate the last presidential race, it can deceive, too, like back in 1972 when the reporters following George McGovern (D) to big rallies smelled a massive upset brewing against President Richard Nixon (R). McGovern lost in a landslide. So we don’t know if the buzz is real, but we’ve heard enough of it that we’re paying attention.

On the other hand, Sen. Jon Tester (D-MT) has been a shaky Leans Democratic in our ratings the whole cycle. Would anyone be shocked if he lost? He’s never won a majority and if Donnelly and/or McCaskill end up losing earlier in the evening, it would make some sense that Tester would be in trouble too. Also, Democratic outside groups have put a little bit of money into appointed Sen. Tina Smith’s (D-MN) bid for a first electoral victory and a lot of money in Sen. Bob Menendez’s (D-NJ) bid for a third term. And of course, Republicans could end up winning any of Arizona, Florida, or Nevada, but any one of those happening wouldn’t really be an upset in races we’ve listed as Toss-ups for about a year or more until now. If so, it will likely be the actual Election Day vote (as opposed to the early vote) that would save the Republicans in these races, much like how in 2016 Republicans did very well on Election Day across the country.


Map 2: Final Crystal Ball gubernatorial ratings

For all the focus on the House and the Senate, the real story of the night may be in the gubernatorial races, where we see the Democrats poised to make big gains.

Right now, the Republicans hold 33 governorships, the Democrats just 16, and an independent, Bill Walker holds Alaska. Our ratings suggest the Democrats could net 10 governorships, while the GOP could lose nine (we favor Republicans to pick up Alaska, which throws off the net change statistic a little bit). That does not include Georgia, where we are maintaining a unique “Toss-up/Leans Runoff” rating in anticipation of a possible runoff on Dec. 4 if neither major party candidate gets a majority. If the runoff happens, just think about how much money former state House Minority Leader Stacey Abrams (D) might raise from the Democrats’ hyper-active small donor network. This is something that concerns Republicans if there’s a runoff.

More than half of the Democratic pickups could come in the Midwest. While we think the GOP could claw back one or two of these states — Iowa, Kansas, and Wisconsin are the picks we’re the least confident in – we thought the data and the year’s overall trends pointed to the Democrats in each of these states individually. Besides the national environment, there may just be a fatigue with eight years of conservative GOP rule in places like Iowa, Kansas, Michigan, Ohio, and Wisconsin, particularly in a time of conservative governance in Washington. The public is idiosyncratic and often wants what it doesn’t have; the same dynamic helped Trump win many states in the Midwest after eight years of a liberal Democratic president.

Upset watch: The Republicans have a real shot to pick off Connecticut or Oregon, two blue states agitating for fresh leadership. The Democrats could very well spring an upset in red states Oklahoma and South Dakota. The gubernatorial races follow traditional political patterns less than the federal races. And keep an eye on Alaska, which has tightened considerably since Walker left the race, leaving a matchup between former state Sen. Mike Dunleavy (R) and former Sen. Mark Begich (D) that the latter definitely has a chance to win.


We know, dear readers, that many of these picks — though hopefully not too many — will be off. But we pick all the races because we believe you deserve our best guess as to what will happen in each contest.

One last thing. A couple of months ago, we featured a handful of political science forecasting models that attempt to predict the net change in the House, and a couple of them also try to project the Senate. More details on the models are available here, and their findings are in Table 2:

Table 2: Political science model predictions of 2018 congressional race

While we did not deliberately fit our seat-by-seat projections to mirror these models, our picks do line up fairly well with them, although we’re slightly closer to the smaller forecasts for Democrats in the House than the larger ones. And note that the two models that are more bullish on Democrats in the House still also forecast a modest GOP gain in the Senate, which is also what we’re expecting. A poll-based model from Ipsos on our jointly-run Political Atlas site shows a similar House gain to what we’re suggesting as well if one assumes that the model’s Toss-ups break about evenly.

We’ll be back sometime on Wednesday with a quick reaction to tomorrow night’s results.

A disclaimer

[1]The Crystal Ball picks represent a collaborative, consensus effort between Editor-in-Chief Larry J. Sabato and Managing Editor Kyle Kondik with the help of many special advisers from both parties who have been with us for years (you all know who you are, and we enormously appreciate your help once again). There are two exceptions: Sabato deferred to Kondik on the pick in KY-6, because Rep. Andy Barr (R, KY-6) is Sabato’s former student, and Kondik deferred to Sabato on the pick for Ohio governor because Kondik used to work for former Ohio Attorney General Richard Cordray (D).

Five days to go

Where we’re leaning in the House, Senate, and governors

Kyle Kondik, Managing Editor, Sabato's Crystal Ball November 1st, 2018



— Our final picks are coming Monday. In the meantime, our longstanding overall assessment — Democrats favored in House, Republicans bigger favorites in Senate — remains in place.

— Four ratings changes in the House.

Table 1: Crystal Ball House ratings changes

Member/District Old Rating New Rating
David Valadao (R, CA-21) Likely Republican Leans Republican
CA-49 Open (Issa, R) Leans Democratic Likely Democratic
Steve King (R, IA-4) Likely Republican Leans Republican
NV-4 Open (Kihuen, D) Leans Democratic Likely Democratic

Table 2: Crystal Ball House ratings

Map 1: Crystal Ball Senate ratings

Map 2: Crystal Ball gubernatorial ratings

Where things stand less than a week out

We have great news for everyone. In just six days, the 2020 presidential campaign will begin!

(No, we’re really not kidding.)

But before then, we have the small matter of the 2018 midterm to settle. We’re still agonizing over our final picks in the closest House, Senate, and gubernatorial races. We will announce our final picks on Monday. That said, the direction of this midterm does not seem like it has changed much in the final days of the campaign. The Democrats remain in the lead for the House majority. The Republicans are even bigger favorites to retain control of the Senate. And Democrats will net governorships.

Our best guesses right now:

House: Right now, we have 212 House seats at least leaning to the Democrats, 202 at least leaning to the Republicans, and 21 Toss-ups. While we’re still gathering information about the Toss-ups, we do have a sense as to where we’re leaning in the races. As of this moment, we’d probably pick the Democrats in 12 of the Toss-ups and Republicans in nine of them. That would amount to a Democratic House gain of 29 seats. So let’s say, for now, we’re thinking an overall Democratic gain of somewhere around 30 seats, give or take. That’s more than the 23 net seats the Democrats need, but not so many more that one could rule out the Democrats sputtering out short of the majority.

One sign that points to the possibility of the GOP maintaining control is an old political science theory called “surge and decline,” which John R. Petrocik and Daron R. Shaw explore in a Crystal Ball piece this week. Historically, one of the factors that sets up the presidential party to lose House seats in the midterm is that they often overperform in the presidential year, winning a number of seats with the help of presidential coattails that they subsequently have trouble defending in the midterm. But Trump and the Republicans didn’t actually enjoy any surge in 2016: They lost six net House seats. If Republicans end up holding the House, keep that theory in mind as part of a post-election explanation.

Some real talk here: We’re going to pick all the House races, but we’re going to get many of them wrong. Many of these races are very close and could tip either way. Our overall goal is to get as close as possible to the overall net change.

Senate: Including the 65 Senate seats not on the ballot as obviously “safe” for the current incumbent party, our Senate ratings show 50 seats at least leaning Republican, 45 at least leaning Democratic, and five Toss-ups. Our current sense, subject to change, is that the Toss-ups might split three to two in either direction. If that happens, and our other ratings hold up, the Republicans would net one-to-two Senate seats.

Governors: Republicans currently hold 33 governorships, Democrats hold 16, and there’s one independent, Bill Walker of Alaska, who recently ended his bid for a second term. Including the 14 governorships not on the ballot as safe for the current incumbent party, our ratings show 22 governorships at least leaning Republican, 18 at least leaning Democratic, and 10 Toss-ups. Split the Toss-ups five to five, and Democrats would have 23 governorships, or a net gain of seven, and Republicans would have 27, a net loss of six (we currently have them favored in Alaska, which would be a gain for them but not a loss for Democrats).

Again, more to come Monday.

House ratings changes

We have a few House ratings tweaks this week before we make our picks on Monday.

The open CA-49 has always looked like the best of the Democratic pickup opportunities in the Golden State, given that it is an open seat and was the closest in 2016: retiring Rep. Darrell Issa (R, CA-49) only won by about half a point. Polls have shown attorney Mike Levin (D) up double digits on Diane Harkey (R), a member of the California State Board of Equalization. Additionally, Democrats seem to be in better shape to hold at least one of the two competitive, open seats they are defending in Nevada, NV-4. Both move from Leans Democratic to Likely Democratic.

Elsewhere in California, Rep. David Valadao (R, CA-21) has long appeared to be in a strong position to win reelection thanks to a strong local appeal even though Hillary Clinton won his Central Valley-based seat by about 15 points. In June, the top-two primary voting numbers were extremely encouraging for Valadao, 63%-37%; those results can be somewhat predictive for the fall. Yet outside groups from both sides are engaged in the district at the end, and even though Valadao remains clearly among the best-positioned Clinton-district Republicans, we’re moving the race from Likely Republican to Leans Republican.

Finally, we’re moving controversial Rep. Steve King (R, IA-4) from Likely Republican to Leans Republican. While the president won this district by 27 points in 2018, King, whose interactions with hateful white supremacists are becoming more frequent and more visible, received a recent rebuke from National Republican Congressional Committee Chairman Steve Stivers (R, OH-15), who tweeted “Congressman Steve King’s recent comments, actions, and retweets are completely inappropriate. We must stand up against white supremacy and hate in all forms, and I strongly condemn this behavior.” King does not seem to be taking his reelection campaign all that seriously, and J.D. Scholten (D), a paralegal and former professional baseball player, appears to be significantly pushing King.

We now have three Republicans in our Leans Republican column in what should be safe GOP seats: King, as well as Reps. Duncan Hunter (R, CA-50) and Chris Collins (R, NY-27), both of whom are running for reelection while under indictment. It is a real possibility that Democrats could score an upset in one or more of these seats, even though we doubt we will pick any of these upsets to actually happen in advance.