Note1 This article was written in conjunction with Ryan Matsumoto, a contributor at FiveThirtyEight, Developer Programs Engineer at Google and Stanford Alum. Follow him on Twitter @ryanmatsumoto1.
Note2 A previous version of this article had Ossoff up 1.3% with 57% odds. Recent polling data had us revise those estimates to 0.6% and 53%, respectively.
In the 2016 Iowa Democratic Primary Hillary Clinton won at least six precincts by way of a coin toss. Both Clinton and Bernie Sanders had an equal number of votes, the count said, and the law stated that ties were to be settled by the toss of a coin. Probabilistically, there’s no fairer hand; the chances that either could win was 50% (someone etched Washington a couple extra hairs).
On June 20th, voters will decide an almost equally toss-up race.
This Tuesday, voters in Georgia’s 6th Congressional district will have their say in the most expensive House race in American history. Democrat Jon Ossoff and Republican Karen Handel are facing off to fill the seat left open by President Trump’s Secretary of Health and Human Services Tom Price. If successful, Ossoff can give hope to members of “The Resistance” who are looking to produce a large Democratic Wave and make inroads to take (back) the House of Representatives in 2018.
But until we have results, we only have analysis. On the side of history, Georgia 6 is a heavily Republican district where Romney beat Obama by 23 points and Price consistently won over 60% of the vote in House races. However, Trump did much worse than past Republicans, beating Clinton here by just 1.5% last November. Open House races like this one are often seen as a referendum on the incumbent president. It wouldn’t be too far fetched for Ossoff to win this traditionally GOP district (with a “benchmarked” score of Republicans +8% — we go into detail about this below).
During the jungle primary (technically we should be calling it the “first round” of the election, but I digress…) that took place in April, 11 Republicans, 5 Democrats, and 2 Independents all competed against each other for the chance to reach the runoff election. As the only competitive Democrat, Ossoff won a strong 48.1% of the vote while Handel edged out the other Republican contenders with 19.8%. Overall, Republicans won 51% of the vote while Democrats won 49% — certainly a much closer race that the district’s history had indicated.
Although a single House seat isn’t that important in the grand scheme of governing, special elections are interesting because of what they might say about the political environment heading into the 2018 midterm elections.
So far, Democrats have failed to win any Republican seats in House special elections. However, that matters far less than the margins of victory compared to past elections. FiveThirtyEight estimates a district’s partisan leaning using a weighted average of the past two presidential elections. Using this metric, Democrats have been overperforming their expected margins in House special elections by 15.3% on average. This suggests that there is a very strong pro-Dem environment. Although we need to be careful about overinterpreting a few off-cycle elections, the average performance of a party in special elections does have some predictive value for forecasting the following midterm election.
Beyond special elections, there are several other indications that the tide has been turning against Trump and the Republicans. Trump’s net approval rating is at -16.7 while Democrats lead on the Generic Congressional ballot by 6.9 points. Both metrics have shown a significant decline since the Georgia 6 primary, suggesting that Handel could suffer more this time around because of Trump’s unpopularity.
Relying on these older measurements and predictions has some drawbacks, though. For one thing, like we say above, the national environment has become much more pro-Democrat than it was for the first round of GA06’s election. Public opinion polling offers us a much more recent (not to mention high-quality!) look at what’s transpiring down in Georgia.
According to my simple average of polls conducted over the last week (as of June 19th, 6:20PM), Ossoff has a 0.6% lead over Handel. This isn’t out of line with what we expect given the pro-Dem environment elsewhere in the country. On top of that, the primary (which residents were not paying nearly as much attention to) was already much more blue than it was previously. Indeed, a simple formula (dare I say “model?”) calculated by Nate Silver posits that Jon Ossoff’s performance in round 1 of the primary suggests a 4 point win for the Democrat, and polling affirms that (albeit only directionally).
|Pollster||Date||Handel||Ossoff||Lead||Source: Various Pollsters, FiveThirtyEight|
|Trafalgar Group||June 19||50.5||48.5||-2|
|Opinion Savvy||June 15||49.4||49.7||0.3|
|Trafalgar Group||June 13||47||50||3|
|↓More than 1 Week Old↓|
|Abt SRBI Inc||June 8||44||51||7|
But polling also tells us a more uncertain story. According to data compiled by Harry Enten of FiveThirtyEight, the margin of error for the leading candidate in a House special election polling average is +-7.5%. That means we can expect anything from a 6 point Handel win to a 9 point Ossoff win come election night, a huge range of outcomes to be sure. To be fair to pollsters, high-profile and competitive races usually enjoy better polling than races that are lopsided or have low turnout. That would give us reason to believe polls will be closer to right than wrong in Georgia 06, but I’m not ruling anything out!
Using those numbers alone for a quick statistical simulation of Ossoff’s margin of victory, we can expect Ossoff to win just 53% of the time. To put that into context, polls give Handel a better chance of winning Georgia’s Sixth Congressional District than Trump had of winning the 2016 US Presidential Election or than Theresa May’s Conservatives had of missing a parliamentary majority in 2017 United Kingdom General Election. This is to say that anything could happen; don’t bet on a horse that loses you money more than 2 out of 5 times.
Speaking of money, Jon Ossoff’s probability of victory has been hovering around 60% in prediction markets. There is something to be said for the so-called “wisdom of the crowd” that often factors into prediction markets, but I think they may be overstating Ossoff’s odds here. Conventional wisdom about Georgia 06 seems like it’s hovering in a 70-75% Ossoff world, when in reality, even our 53% estimate may be giving to much credit to data that often falls from expectations.
If polls are a cautionary tale for Ossoff supporters, the early vote may be an even louder warning sign. According to recent estimates a total of 142,000 early ballots have been cast in the runoff, an increase from 60,000 in round 1 early voting including 40,000 voters who didn’t cast a ballot at all in round 1. Big turnout is usually good news for Democrats, except this time (when the news is neutral).
In the first round of Tuesday’s election, Democrats were earning 65% of early votes in the district, a 30 point margin over Republican candidates. But on Election Day Republicans won 58% of ballots (15 percentage point margin over Democrats), ultimately winning the district by 2 percent.
This has the obvious implication that Ossoff needs a big election-day shift to win Georgia’s Sixth, but some worthy notes lie beneath the surface. Since Ossoff will likely lead early in the night (since early votes are counted first), we can expect the race to get more competitive as the evening wears on. This is typical — but suffice to say that early returns will probably indicate a race that’s more Ossoff-leaning that it turns out to be.
Still, we should take this analysis of early votes with a grain of salt. Turnout patterns are not very stable between primary and runoff elections, where enthusiasm really matters for the former and competitiveness matters a lot for the latter. What ultimately matters is who turns out both early and on election day.
Election Night Projections
So, how will we know on election night if Ossoff is on track to win? Elections are notorious for producing misleading initial results, like when the New York Times’ predictive model showed Clinton as a heavy favorite early, or when the Detroit Free Press mistakenly projected Michigan for Clinton.
The basic problem is that initial results may not be representative of the political leanings of the entire district. Handel could gain an early lead if the most conservative precincts report their results first. Alternatively, Ossoff could gain a strong lead with early votes before Handel’s strength with Election Day voters swamps him. What we want to know is whether or not the district is looking more(less) Democratic(Republican) than last time around.
The best way to gauge who is winning is to look at completed precinct results. Georgia 6 has 210 precincts, giving us many data points to consider. Using precinct results from the primary, Ryan estimated benchmarks for each precinct that Ossoff should reach on average if he wants to clear 50%. We can continually re-calculate ‘Average Ossoff Performance Relative to Benchmark’ as precinct results trickle in. For example, if this metric is +4.2 points after the first 30 precincts report, we might expect him to ultimately earn 54.2% of the vote. Since the early vote can be misleading, a precinct’s ‘Performance vs. Benchmark’ will only be tabulated once all of a precinct’s votes (both early and election day) have been reported.
To move beyond a snapshot of who is currently winning the race to who will likely win in the end, Elliott has built a model that uses election returns in reporting precincts to project returns in outstanding ones. It updates live, and you can follow along here Or, if you don’t want to have two spreadsheets open, the information will be streaming live on his Twitter.
The final margin of this race may give us some clues about shifting partisan allegiances in the Trump era. If Ossoff wins, this could be evidence that moderate Republicans who defected from Trump in 2016 are voting Democrat at the Congressional level. However, if Handel wins by a surprisingly large margin, that could be evidence that anti-Trump appeal is not going far enough, and that Republicans are returning to their pre-Trump partisan leanings (recall that the District voted for Mitt Romney over President Obama by 23%).
Ultimately, this election isn’t the sole test of whether or not the Democrats will take back the House in 2018. Given the partisan leaning of the district, Democrats don’t need to take this seat, but they do need to be competitive to stay on track to win back the House in 2018. Even if Ossoff loses by a percent or so — which is well within the realm of possibility, Democrats would be well-positioned to threaten the Republican majority.
Finally, what may be more important still is the impact that the election to Georgia’s Sixth has on healthcare politics. Jon Ossoff has continuously rallied opposition to the Republican’s Affordable Healthcare Act. If district is a referendum on anything other than Trump, it’s probably healthcare — and an Ossoff victory could be great for Democrats and moderate Republicans who want a better plan.
Want more Georgia 6 news? Follow the link below for precinct-level benchmarks and projections.
Projections for the Georgia 06 Election
If you’re visiting us on election night, Click here to view Live in Google Sheets
- May 15 Elections: PA Primaries and More Special Elections (Which Already Predict a Likely Democratic Victory in November)
- Congressional Polling Versus Special Elections: Are Democrats Really Heading for a Wave?
- What Would Happen if Thirty House Members Resigned?