These upsets are in many ways indicative of broader trends in politics and the American electorate.
David Axelrod’s grumblings on CNN on election night notwithstanding, the polling for the midterm elections was largely on point. Out of 506 races (including all House, Senate, and gubernatorial races), the polls* correctly predicted the winner in over 460** of them. In aggregate, they were right on the outcomes of both the House and the Senate, at least directionally if not in the exact number of seats gained or lost.
Of course, that doesn’t mean the polls were not without their failures. They missed their mark in several key races across the country, in some cases drastically so, resulting in some surprising upsets. Six of these upsets were especially remarkable because of some combination of how unexpected and consequential they were. All but two of those six were categorized as “Likely D” or “Likely R,” meaning that the favored candidate had a greater than three in four chance of winning, and all of them have wide-ranging implications for how the country will be governed and who controls the levers of power.
Though they constitute only a small sample size, these upsets are in many ways indicative of broader trends in politics and the American electorate. The three Democratic victories were all in once-safe Republican urban or suburban congressional districts—areas that are increasingly turning away from Donald Trump and the Republicans. The three Republican victories—two in the Senate along with the Florida gubernatorial race—don’t tell quite as coherent of a story, but do show that Democrats still have a lot to worry about come 2020. In particular, they demonstrate that taking back the Senate will be a heavy lift and that winning Florida in the next presidential election will likely be an uphill battle.
One quick clarification: this list does not include races in which the outcome might be surprising based on conventional wisdom but was largely in-line with the polling/forecasting. For example, with Democrat Lauren Kelly defeating Republican Kris Kobach in the Kansas gubernatorial race, conventional wisdom tells us that a statewide race in Kansas should be a slam dunk for a Republican, but the polling actually indicated that it was a toss-up.
Results are current as of 11/8/2018 and subject to change as more votes are counted. Other than the Florida Senate, all races have officially been called.
Ron DeSantis defeating Andrew Gillum, Florida Governor
- Pre-election odds: 7 in 9 for Gillum – Likely D
- Predicted margin: D+4.1%
- Actual margin: R+0.6%
- Polling error: 4.7%
For the second election in a row, Florida defied its blue-tinged polling numbers to deliver Republicans a statewide victory. Of all the Democrats’ losses in this election, this one might sting the most. Tallahassee Mayor Andrew Gillum was a heavy favorite coming into the night, and his loss not only means that Republicans will retain complete control of Florida’s state government, but it could be a bad omen for 2020. Gillum was one of the party’s most progressive candidates running in a statewide election, and his candidacy was in a sense a litmus test for how such a liberal platform would fare in a purple state. He was also slated to become Florida’s first African American governor.
Instead, Florida elected Gillum’s polar opposite. A former U.S. Representative for Florida’s 6th congressional district, Ron DeSantis is a Republican in the true Trumpian mold. During the campaign, DeSantis called the president directly to ask for his endorsement, and he famously aired a lighthearted yet somehow still incredibly ominous campaign ad showing him building a toy wall with his daughter. And much like Trump, DeSantis’s populist appeal has a not-so-subtle racial element to it. He has a history of cavorting with journalists with white supremacist leanings and received support from various white nationalist groups. DeSantis also said on Fox News that Florida shouldn’t “monkey this up” by electing his black opponent. Then there were the infamous racist robocalls; even though DeSantis seemingly had nothing to do with them, they say a lot about his base.
Mike Braun defeating Joe Donnelly, Indiana Senate
- Pre-election odds: 5 in 8 for Donnelly – Lean D
- Predicted margin: D+1.9%
- Actual margin: R+7.3% (with only 94% reporting)
- Polling error: 9.2%
When the first results started coming in, this race figured to be a close one. In the end, it was anything but. By the time the dust settled, incumbent Democrat Joe Donnelly had lost to Republican challenger Mike Braun by a wide margin, making it the first of four (assuming Florida’s result doesn’t change after a recount) Democratic incumbent Senators to lose their reelection bids. And with a polling error of over 9 percent, this was one of the biggest misses of the night.
Donnelly was first elected to the Senate in 2012. His unlikely victory in deeply red Indiana can largely be attributed to the fact the then-incumbent Republican Richard Lugar, a relative moderate who had held the seat since 1977, lost in the primary to the more radical Tea Party-backed Richard Murdock. Braun is a prominent Indiana businessman who formerly served in the state’s House of Representatives.
Rick Scott defeating Bill Nelson, Florida Senate*
*As of this writing, the race hasn’t been officially called, but Scott holds a 0.2 percent lead (less than 22,000 votes). There is likely to be a recount.
- Pre-election odds: 5 in 7 for Nelson – Lean D
- Predicted margin: D+3.4%
- Actual margin: R+0.2%
- Polling error: 3.6%
Florida, Florida, Florida. You can never make this easy, can you? Once again, it looks like the Sunshine State is headed toward a recount. Though former Republican Governor Rick Scott hasn’t technically won yet, in all likelihood, this one is staying in the red column, so it makes the list for now.
Of the four Senate seats the Democrats lost in this election, this is the biggest surprise and certainly the hardest to swallow. Whereas Democrats in North Dakota, Missouri, and Indiana seemed to be serving on borrowed time, Florida is a state that the party is expecting to get more blue over time. But like the gubernatorial race, this indicates that things could be headed in the opposite direction.
In his 18 years in the Senate, Bill Nelson made a name for himself as a moderate while still managing to put his stamp on more liberal initiatives such as same-sex marriage, environmental regulation, and expanding Medicaid. He also went to space!
As a two-term governor, Scott is perhaps best known for adeptly handling the response to Hurricane Irma. However, he backed significant pro-gun legislation such as the infamous stand-your-ground law and implemented a hard-line immigration policy. He also denies the existence of anthropogenic climate change, but then again, he’s not a scientist. Rick Scott has not been to space.
Kendra Horn defeating Steve Russell, Oklahoma’s 5th congressional district
- Pre-election odds: 14 in 15 for Russell – Likely R
- Predicted margin: R+10.0%
- Actual margin: D+1.4%
- Polling error: 11.4%
That a Democrat could win in Oklahoma kind of speaks for itself. But what’s more remarkable is that this is the first time a Democrat won in Oklahoma’s 5th in 44 years. If there ever were a sign of a blue wave, this would be it. And while Kendra Horn is no Elizabeth Warren, she’s not exactly a pure centrist either. “She campaigned on a broadly left-wing platform: gun control, more funding for education, tighter campaign finance controls and opposition to private prisons,” the Huffington Post wrote.
Horn defeated Steve Russell, a Republican incumbent that voted with Trump, over 93 percent of the time.
Max Rose defeating Dan Donovan, New York’s 11th congressional district
- Pre-election odds: 4 in 5 for Donovan – Likely R
- Predicted margin: R+5.2%
- Actual margin: D+6.0%
- Polling error: 11.2%
In a way, it was only a matter of time before Republicans lost their grip on New York’s 11th, which encompasses all of New York City’s Staten Island and parts of south Brooklyn. A sign of a deepening trend across the U.S., this was the last all-urban district (as defined by CityLab) held by the GOP in the entire country.
Even though Trump carried this district by nearly 10 points in 2016, Democrat Max Rose defeated Republican incumbent and Staten Island native Dan Donovan handily. Rose’s win is a particularly stark repudiation of Trump’s agenda given that Donovan voted with the president 87 percent of the time.
Just 32 years old, Rose is a moderate with a military pedigree, which seems to have become a common profile among Democratic victors in traditionally Republican strongholds. Think Mikie Sherrill with a shaved head.
Joe Cunningham defeating Katie Arrington, South Carolina’s 1st congressional district
- Pre-election odds: 11 in 12 for Arrington – Likely R
- Predicted margin: R+8.0%
- Actual margin: D+1.4%
- Polling error: 9.4%
Joe Cunningham’s victory in South Carolina’s 1st is as round a rejection of Donald Trump as they come. The seat is currently held by former South Carolina Governor Mark Sanford. Though Sanford initially endorsed Trump in the 2016 election, he positioned himself early on in Trump’s presidency as one of his few Republican critics in the House. He censured the president for the Muslim travel ban and even told Politico in February 2017 that Trump “represents the antithesis, or the undoing, of everything I thought I knew about politics, preparation and life.”
Trump, as you can imagine, didn’t take too kindly to that. So he threw his weight behind upstart challenger Katie Arrington, a representative in the South Carolina State House and avowed Trump supporter. Arrington beat Sanford in the primary only to lose to her Democratic opponent. Cunningham, a 35-year-old construction law attorney, yoga studio owner, and certified Handsome Young Man™, has a surprisingly progressive platform for a Southern Democrat. He supports gun control, equal pay for women, and fighting climate change, and is against offshore drilling.
* Forecasts are based on FiveThirtyEight’s Deluxe model. In addition to raw polling data, the deluxe model also incorporates expert ratings such as those from Cook Political Report and “fundamentals” such as partisan lean and fundraising numbers.
** Election results are based on New York Times reporting. This number does not include races that were rated as toss-ups by FiveThirtyEight, and there are still several races that have yet to be called.