Skip to main content
ABC News
What The Special Election In Georgia Can — And Can’t — Tell Us About The Midterms

Twelve years ago, Paul Hackett nearly shocked the political world. Hackett, a Democrat, lost a special House election in Ohio’s 2nd Congressional District by only a little more than 3 percentage points. The district had voted for George W. Bush by 28 points just a few months prior in 2004. But Hackett’s near-win turned out be a harbinger: A little bit more than a year later, Democrats stormed back to win 30 Republican-held seats and gain control of the House of Representatives.

Democrats are hoping Jon Ossoff can set off a similar wave. Ossoff is running in a special election to represent Georgia’s 6th Congressional District, traditionally a GOP-leaning seat. There’s a primary on April 18 in which all candidates from all parties face off, and polling shows Ossoff leading the field (it’s basically Ossoff vs. a slate of Republicans splitting up the conservative vote). If no candidate gets above 50 percent — and Ossoff is likely the only candidate who has a chance of doing that — the top two finishers, regardless of party, will face off in a June 20 runoff.

The reason this race is interesting is because Georgia 6 is the type of district Democrats need to win to take back the House majority. An Ossoff win, in other words, could be a really good sign for them. It would not, however, be proof positive of a forthcoming Democratic wave.

Winning only blue districts won’t get Democrats the House — they’ll have to pick up some red seats, as well. To gain a one-seat majority, Democrats must pick up 24 seats that were won by Republicans in 2016. But only 14 GOP-held seats are more Democratic-leaning than the nation as a whole, according to a weighted average of the last two presidential election results1 in each district. Looking at the margin, the 24th seat (i.e., the seat that would give Democrats the majority) is 4 percentage points more Republican than the nation. Georgia 6 is 9.5 points more GOP leaning.

So at first glance, there are a lot of districts that would seem like easier pick-up opportunities for Democrats before you get to Georgia 6 — there are 47 Republican-held House seats that lean more Democratic. If Ossoff loses, Democrats might say, “Hey, this was a really red district; we don’t need to march this far into GOP territory to take back the House.”

That’s true strictly in terms of the political lean of each district, but Ossoff is running for an open seat. Many of those other 47 Republican-held districts will have incumbents running in them. Although the advantage incumbents have in House elections isn’t as large as it was 30 or 40 years ago, it is still substantial. A model I ran on the 2016 results suggests that, once you take into account the weighted presidential vote of a district, the incumbent advantage is somewhere between 10 and 132 percentage points. That’s why Georgia 6 — despite its 9.5-point GOP lean — still makes for a decent national bellwether.

But pay attention to the margin in Georgia 6 as much as to who wins. A tight Ossoff win would be consistent with an anti-Trump national mood, but it would also be consistent with a political environment in which many vulnerable Republican incumbents could survive.3

Even taking into account the margin in Georgia 6, we have to be careful: One special election does not a big sample size make. For one, there have been plenty of special elections that deviated wildly from the recent political trend in that district or state. (I’m including Senate races to increase the sample size.)4

CYCLE DISTRICT DEM MARGIN IN SPECIAL ELECTION DEM LEAN IN LAST TWO PRESIDENTIAL ELECTIONS DIFF.
2002 Pennsylvania 9 -8 -33 +25
2002 Oklahoma 1 -9 -26 +17
2002 Arkansas 3 -13 -20 +7
2002 Florida 1 -38 -40 +2
2002 Mass. 9 +32 +31 +1
2002 Virginia 4 -4 -1 -3
2002 California 32 +55 +71 -16
2002 South Carolina 2 -48 -20 -28
2006 California 5 +59 +25 +33
2006 Ohio 2 -3 -27 +23
2006 California 50 -5 -9 +5
2006 California 48 -17 -16 0
2010 Pennsylvania 12 +8 -5 +13
2010 Illinois 5 +45 +39 +6
2010 New York 20 0 -5 +5
2010 Florida 19 +27 +26 +1
2010 California 32 +29 +30 -1
2010 New York 23 -3 -2 -1
2010 Hawaii 1 +19 +28 -9
2010 California 10 +10 +24 -14
2010 Mass. Senate -5 +21 -26
2010 Georgia 9 -87 -57 -30
2014 South Carolina 1 -9 -22 +13
2014 Mass. 5 +34 +28 +6
2014 Florida 13 -2 -3 +1
2014 New Jersey Senate +11 +14 -3
2014 Missouri 8 -40 -36 -4
2014 Illinois 2 +49 +58 -9
2014 Mass. Senate +10 +19 -9
2014 Louisiana 5 -38 -29 -10
2014 Alabama 1 -42 -29 -13
2014 Florida 19 -38 -25 -13
Special congressional election results are usually inconsistent leading up to the midterm election

Margins are rounded and determined with the best available data for each special election. The Democratic lean in the last two presidential elections weights the Democratic vote margin of the most recent presidential election 75 percent and the
Democratic vote margin of the next most recent presidential election 25 percent.

Source: Daily Kos Elections, Cook Political Report, Swing State Project

If Ossoff wins, Democrats may end up pointing to Hackett’s near victory in 2005, but Republicans could just as easily point to Pennsylvania’s 12th District result in 2010. In that election, Democrats ended up holding on to a seat that was 5 percentage points more Republican leaning than the nation as a whole.5 Nevertheless, later in 2010, Republicans gained 63 seats and won control of the House, winning the aggregate House vote by 7 percentage points.

And Pennsylvania 12 in 2010 wasn’t a freak occurrence. Relying on an individual special election result to predict the political environment in the following midterms is a good way to look foolish.

DEMOCRAT VOTE MARGINS
CYCLE DISTRICT IN SPECIAL ELECTION RELATIVE TO NAT’L DEM LEAN IN NEXT NATIONAL HOUSE MIDTERM DIFF
2002 Pennsylvania 9 +25 -5 -30
2002 Oklahoma 1 +17 -5 -22
2002 Arkansas 3 +7 -5 -12
2002 Florida 1 +2 -5 -7
2002 Mass. 9 +1 -5 -6
2002 Virginia 4 -3 -5 -2
2002 California 32 -16 -5 +11
2002 South Carolina 2 -28 -5 +23
2006 California 5 +33 +8 -25
2006 Ohio 2 +23 +8 -15
2006 California 50 +5 +8 +3
2006 California 48 0 +8 +8
2010 Pennsylvania 12 +13 -7 -20
2010 Illinois 5 +6 -7 -13
2010 New York 20 +5 -7 -12
2010 Florida 19 +1 -7 -8
2010 California 32 -1 -7 -6
2010 New York 23 -1 -7 -6
2010 Hawaii 1 -9 -7 +2
2010 California 10 -14 -7 +7
2010 Mass. Senate -26 -7 +19
2010 Georgia 9 -30 -7 +23
2014 South Carolina 1 +13 -6 -19
2014 Mass. 5 +6 -6 -12
2014 Florida 13 +1 -6 -7
2014 New Jersey Senate -3 -6 -3
2014 Missouri 8 -4 -6 -2
2014 Illinois 2 -9 -6 +3
2014 Mass. Senate -9 -6 +3
2014 Louisiana 5 -10 -6 +4
2014 Alabama 1 -13 -6 +7
2014 Florida 19 -13 -6 +7
Special congressional election results often aren’t predictive of the midterm result

Source: DAILY KOS ELECTIONS, COOK POLITICAL REPORT, SWING STATE PROJECT

As you can see in the table above, special House and Senate elections in the two years leading up to a midterm can go any which way. In any of the previous four cycles before a midterm, there’s at least one example of a candidate doing poorly in a special election — relative to the previous weighted presidential vote — only to have their party do well in the midterms.6

So even if Ossoff wins — even if he wins comfortably — it won’t be safe to assume a Democratic wave is building in 2018.

Instead, if you really want to use special election results to look ahead to 2018, you need to look at a bunch of them. While any individual special election may mislead, the average special House and Senate result compared to the past presidential vote provides a decent indication of the national environment heading into the following midterm election. In the last two midterm cycles, the Republican outperformed expectations7 in the average special House and Senate election, and the GOP went on to make gains both years. In special elections from 2013 to 2014, Republican candidates outperformed expectations by 4 percentage points in the average special election and went on to win the House popular vote in 2014 by 6 percentage points. In the 2010 cycle, the GOP beat expectations by 6 points in special elections and won the House vote by 7 points.

In the 2006 cycle, it was the opposite. Democrats outperformed their previous presidential vote by a significant margin (15 percentage points) in the average special election and went on to take the House easily in 2006 (by 8 points). The 2002 cycle, however, was nearly a draw. The average special election result fell right in line with expectations, and Republicans won the House vote by 5 points. Still, a push suggested the party in control of the White House wasn’t going to suffer the usual midterm drubbing.

The point is that Democrats and Republicans should be keeping an eye on the other four House special elections coming up over the next few months — not just Georgia 6. If Democratic candidates in those districts are outperforming the previous presidential vote compared with the nation by 9 to 10 percentage, too, it could be a sign that whatever happens in Georgia 6 is not a fluke. A Democratic national House win by 9 to 10 percentage points would probably be large enough to take back the House in 2018. Of course, an Ossoff loss combined with Democratic underperformance in those other races would indicate that Democrats are in trouble.

Footnotes

  1. 2016 is weighted to 75 percent, while 2012 is weighted to 25 percent.

  2. It’s larger when you include races where one of the two parties didn’t field a candidate in the general election or runoff.

  3. Of course, many Republican incumbents may also retire if the national mood is strong enough for Ossoff to be elected.

  4. Some notes on the table below: For Georgia 9 in the 2010 cycle and Louisiana 5 in the 2014 cycle, I took the sum percentage of the Democratic and Republican candidates running in the first-round runoff in which all candidates run against each other regardless of party affiliation. For Hawaii 1 in the 2010 cycle, I take the sum percentage of the two Democratic candidates running in the general election. For New York 23 in the 2010 cycle, I took the sum percentage of the Conservative and Republican candidates to calculate the Republican candidate total. For all elections in the 2002 cycle, the analysis relies on two-way or major party presidential vote in each district because that was all that was available.

  5. Again, according to a weighted average of the previous two presidential elections.

  6. That includes California 48 in the 2006 cycle, which is rounded to 0 in the table.

  7. The two previous presidential results, weighted.

Harry Enten was a senior political writer and analyst for FiveThirtyEight.

Comments