Sabatos Crystal Ball

Special Circumstances

Georgia’s Sixth District and the dangers of overinterpreting special elections

Kyle Kondik, Managing Editor, Sabato's Crystal Ball April 13th, 2017


Whatever happens in the first round of voting in the special election in Georgia’s Sixth Congressional District on Tuesday, it seems like a safe bet that the result will get a fair amount of national attention because of what it may tell us about the 2018 midterm. But before getting into what those lessons may be, let’s remember that this is a special election — and thus it features special circumstances.

Here are a few:

So, to sum it up, the GA-6 special is indeed special: It uses an election format that hardly any other 2018 races will use; it features only one prominent Democrat who has used his unique position to harness an immense fundraising base while a giant Republican field fights for scraps; it is taking place in a district that changed dramatically at the presidential level from 2012 to 2016 in the Democrats’ favor; and it is an open seat.

This is all a way of saying that those who project the GA-6 outcome, whatever it is, onto the still-distant 2018 midterms do so at their own peril. History tells us that these special elections can be a harbinger of the future, although there are plenty of examples illustrating when special elections provide misleading or mixed signals of what is to come.

Greg Giroux, a political reporter for Bloomberg and one of the nation’s leading experts on electoral politics, has helpfully compiled every House special election held since the start of John F. Kennedy’s presidency in 1961. There have been 247 special elections in the five and a half decades since then, including Tuesday’s special in KS-4. Roughly a fifth of those elections — 47 of 247 — saw a party change, and generally speaking, the party changes broke against the White House. Of those 47 party changes, 35 of them involved the party that did not control the White House winning the seat from the president’s party, while the other 12 flipped in favor of the president’s party.

This makes some sense, given what history tells us about midterm elections. The president’s party typically loses ground in midterms, and a special election essentially amounts to a mini-midterm election: Turnout is significantly lower than a presidential election and the party that doesn’t hold the White House can be more motivated to vote.

It’s easy to find times when a special election party change seemed suggestive of an upcoming partisan wave in favor of the party that did not hold the White House. Here are some examples:

But in other instances, special elections can provide misleading or mixed signals:

So as we assess GA-6, it’s certainly possible that Ossoff could win outright on Tuesday and that Democrats could end up having a disappointing midterm anyway. Or he could lose decisively in a June runoff and Democrats could come back and have a massive wave election next year. It’s ultimately just one election held amidst unusual and, dare we say it, special circumstances.

We’re calling GA-6 a Toss-up, a designation we applied to the race roughly two weeks ago after the National Republican Congressional Committee sounded the alarm bell and started aggressively spending money in the district. That’s in addition to the millions the Congressional Leadership Fund, a Super PAC that is close to Speaker Ryan, has also spent in the district. Since then, Ossoff’s huge fundraising has come to light, as have early voting statistics that seem to indicate heavy Democratic interest in the race (although Republicans, who have more candidate choices and thus perhaps waited longer to vote, are catching up).

So there’s a lot of uncertainty about the outcome: Polling, typically spotty in House races, generally shows Ossoff in the low 40s. If that’s all he gets in the first round of voting, and the combined Republican vote is over 50%, one would assume that Ossoff’s general election opponent would start with the upper hand: After all, the first round results are better than any poll — they are actual voting results that can be a preview of the runoff on June 20, if there is one. However, if Ossoff’s vote and the scattered votes for the four other Democratic candidates add up to a total approaching 50% (say, 45% or more), it may indicate that the runoff should be quite competitive. Obviously, a first-round win by Ossoff would be noteworthy because he would have exceeded Clinton’s 46.8% 2016 share significantly — and blown recent previous Democratic House performance in the district out of the water. Another factor: As of now, Ossoff and Democrats have not been attacking the Republicans because it’s anyone’s guess how the first round will play out, while outside GOP groups have been hammering Ossoff, hoping to drive down his numbers (and while Ossoff has been running lots of positive ads on his own behalf). Ossoff and national Democrats may be preparing to drop the hammer on whichever Republican emerges from the first round, again assuming Ossoff does not win outright on Tuesday. In other words, the dynamic changes on Tuesday in advance of a possible runoff: The GOP survivor goes from running against his or her fellow partisans to running against Ossoff, while Ossoff can shift into attack mode because he would have a clear opponent.

Perhaps the more useful way to interpret the results in GA-6, whatever they may be, is to put them in context of the other special elections that have happened so far in state legislative races as well as the KS-4 special.

In 10 special elections so far — nine state legislative races and the KS-4 U.S. House special — the Democratic candidate has improved on Clinton’s 2016 margin in eight of them. Across all 10 races, the average net improvement has been 11 points. Now, this is a small sample size, and the results vary dramatically — ranging from a Democrat doing a net 22 points worse in a deeply Democratic state Senate district in Connecticut all the way to one who did a net 34 points better in an Iowa special House election in a Democratic-leaning district. (Daily Kos Elections and Huffington Post are keeping track of these races.) If this trend continues throughout 2017 as the data accumulate, this could be suggestive of broader GOP problems and intensified Democratic enthusiasm.

Indeed, back during the last midterm cycle, the always-perceptive Sean Trende of RealClearPolitics (and a contributor to our new book on the 2016 election, Trumped) identified consistent Democratic underperformance in elections conducted in 2013. That year, the early tea leaves correctly suggested the midterm outcome — a weak Democratic performance in 2014 defined by poor Democratic turnout in many places. The 2017 tea leaves may not be such an accurate predictor, but at least looking at all the races as a group helps iron out the unusual circumstances of a given race — and, as noted, GA-6 features a lot of unique characteristics — and tells us more about the broader environment.

If that’s the way to project special election performance forward to the midterm, then GA-6 is but one data point that may or may not confirm national trends. So by all means, pay attention — but not just to a single race like GA-6. Remember Lester Freamon’s advice from The Wire: “All the pieces matter.”

And that includes many pieces that go beyond special elections: national House generic ballot polling, the president’s approval rating, candidate recruiting, and retirements. It may be on those latter two factors where special elections exert some influence — if the accumulated results suggest overperformance by one side, that party could have an easier time finding candidates, while incumbents from the party on the wrong side of the results could give stronger consideration to spending more time with their families.

Housekeeping: Other races and ratings changes

Table 1: House ratings changes

*Note: We announced the changes in MN-1 and GA-6 on March 27 and March 30, respectively, on Twitter, but we had not previously mentioned them in our Crystal Ball newsletter. To stay up to date on our ratings changes, please bookmark our Ratings Change page.