2017 Special Elections Look Like a Wave

Pundits and campaign professionals have wrestled with whether 2018 will be a wave election or not, using metrics like President Trump’s job rating, candidate recruitment, fundraising and the four Congressional special elections in the last six months. A deeper more-local look at 2017 elections, though, show an already-building wave. Over 29 contested special elections at the state level, Democrats have substantially over-performed both Hillary Clinton’s 2016 performance and Barack Obama’s 2012 performance. If that continues apace, it will be enough to flip almost a thousand state legislative races, give Democrats control of more state houses than Republicans, and most likely have similar impacts on up-ticket races like Congress/Governor/Senate.

While congressional special elections have generated a great deal of coverage, they have not been a fair test of where each party is going. They are a small sample size (four), and each special election has taken place in a district that 1) was strategically chosen by the Trump administration, with the presumed belief that they would yield comfortable GOP holds and 2) is full of the type of voters who had already elected someone enthusiastic about joining the Trump Administration. State legislative elections are a far more representative experiment of how voters are behaving:

1. There have been 29 competitive special state legislative elections since Trump’s inauguration, far better to read a pattern from than the congressional specials (contested defined as both major parties getting at least 20% of the vote)

2. Those 29 seats are distributed more or less randomly by party and geography as the 7000+ legislators around the country pass away, retire, take another job, or vacate office due to scandal.

In those 29 elections, Democrats have substantially overperformed past results. To be more exact, since Trump’s inauguration Democrats have run an average of 12.1 points ahead of Hillary Clinton’s margin and 6.0 points ahead of Barack Obama’s 2012 margin. That includes Democrats flipping 8 of 20 Republican-held districts while holding all 9 of their own open seats.

The chart below shows Democratic performance in special elections vs. the 2016 Presidential election. The further a dot is above the line, the better a Democrat did relative to Clinton. A dot below the line means a Democrat underperformed the Presidential election.

Almost without exception, Democrats are punching well above their usual weight in these races. If this pattern held across every state legislative election in the country in the next three and a half years, Democrats would:

· Add about 850 legislative seats, going from their current ~1100-seat deficit nationally to a ~700-seat advantage

· Take control of 20 new state chambers, moving from total control in only 12 states to total control in 25 (1 would be split, 23 would be GOP-controlled, and 1 is non-partisan Nebraska)

Additionally, many Beltway pundits continue to debate whether Democrats should target so-called blue-collar Obama-Trump type districts or more white-collar, suburban Romney-Clinton districts. The answer so far on the legislative level, is “Yes”; Democrats need not acquiesce to that false choice. Just like FiveThirtyEight, we find that Obama’s 2012 performance and Clinton’s 2016 performance in a district are equally predictive of 2017 results. We found the best backwards predictor of this year’s special elections was:

· Average the district’s Presidential margin in 2012 and 2016

· Add 9.1 points — the average between how much Democrats are over-performing Clinton (12.1%) and Obama (6.0%) in special elections so far

An example: a hypothetical district where Trump won by 10 and Obama and Romney tied. The seat averaged a +5 Republican advantage in the last two Presidential elections, which based on special elections so far would predict a 4.1% Democratic win in 2017 (5-point GOP advantage + 9.1 Dem). We then applied this formula to more than 7,000 legislative seats in 49 states, leaving out Nebraska’s non-partisan unicameral body. In states where DailyKos has not calculated both 2012 and 2016, we used the one election they had calculated. The exceptions where Alabama and Mississippi where they calculated neither and we assumed no seats flipped.

Writ large across the nation over the next four years (the time it takes for every seat to hold an election), this would result in the following shifts:

Because both 2012 and 2016 have been equally important predictors, a lean Obama district that swung heavily to Trump is just as ripe an opportunity as a strongly Romney district that shifted to Clinton. Republican legislators who hold either of those types of districts — as well as a much broader swath of GOP districts — should be very worried by what has occurred at the legislative level over the past several months. Likewise, Democrats do not necessarily need to choose between targeting state houses in places like Iowa where Trump did well in 2016 or states like Arizona or Virginia, where Trump is generally weaker than other recent Republicans.

For example, in Iowa Democrats have already done well in special elections with a 45-point rout in HD89 (Clinton +11, Obama +28) and a hold by 9 points in HD82 (Clinton -21, Obama +2). If Iowa Democrats continued to perform at the national 2017 average compared to Hillary Clinton, they would gain 5 seats in the State Senate and 12 in the State House by the time every seat is up. Under this scenario, the State Senate would move from a 10-seat GOP majority (20 D / 30 R) to a two-seat advantage (24 D / 26 R) and an 18-seat Republican advantage in the House (41 D / 59 R) to a 6-seat Democratic lead (53 D / 47 R).

More alarming for Iowa Republicans would be if legislative Democrats continue their ability to outpace not just Hillary Clinton’s 2016 share but the average of that and Barack Obama’s 2012 performance at the rate that’s been happening nationally. In Iowa, in the next three years this would leave Iowa Democrats with a 20-seat lead in the Senate and a 22-seat lead in the House. That’s not meant to predict Democratic dominance of the Iowa legislature, but it does show the expansive playing field available to legislative Democrats playing offense as well as how well Democrats have done in special elections to this point.

Another state that drives home Democratic opportunities is Virginia, which will be an early test with State House of Delegate elections in November 2017. Virginia Republicans currently hold a roughly 2:1 House advantage (34 D / 66 R). Using the more conservative baseline of outperforming Barack Obama’s 2012 vote share (whose margin was slightly more narrow than Clinton’s) would result in a 10-seat Democratic majority (55 D / 45 R). Using an average of the outperforming 2012 and 2016 would provide Democrats with more than 60 State House seats (61 D / 39 R).

There are numerous reasons why special elections may not translate to on-year elections, including the distinction between open-seat versus incumbent/challenger races and midterm turnout dynamics versus largely one-off special elections. We’re also not making race-by-race or even chamber-by-chamber predictions. State dynamics will be meaningful: so far in 2017 Democrats have flipped three deep-red seats in Oklahoma where GOP Governor Mary Fallin is unpopular, but in Connecticut Democrats have performed below Clinton’s and Obama’s performance on average. Other factors including the number of open seats, top-of-the-ticket performance and campaign spending, make specific projections a fools’ errand. A look at two of those specific factors:

1. Incumbency. Unlike special elections which are open seats by definition, incumbency will matter in 2018 and beyond. Republicans hold about 1100 more legislative seats than Democrats nationwide, so they benefit more fro incumbency. Many Republicans have already survived elections in blue/purple seats, and they’ll be tougher to knock off than winning an open-seat race. When we add a-5 point “incumbency advantage” to each party’s vote if they currently hold a seat (even those who are retiring and term-limited), Democrats make fewer gains and flip 14 chambers instead of 20 because they do not win either chamber in WI, PA, or MI. When we add a 10-point advantage for a sitting legislator, Democrats only flip 7 chambers because they also don’t flip IA, FL, VA House, and NH + MN Senate). That said, many Democratic incumbents were washed away in 2010 and 2014 who had been assumed to be invulnerable to a wave.

2. Turnout. There’s no question that Democrats are turning out at higher rates than Republicans right now, and a theory goes that this gets negated or attenuated in higher-turnout midterm elections compared to very-low-turnout special elections. So far, the average special election has had 44% of the total votes that were cast in the same district in 2014 (calculated for districts had contested 2014 legislative races). Democrats have overperformed less as special-election turnout has climbed (2017 turnout as % of 2014 turnout is on the x axis, Democratic over-performance is on the y axis):

With that said, the one outlier on the far right of the graph is causing all of the downward trend, which is the “poor” result in Obama/Clinton Delaware Senate District 10 where Democrats posted a 17-point win in a seat they only carried by 2 percent in 2014. Remove that district and turnout has had zero real effect on Democratic over-performance so far:

A wave election is already building, judging by the dozens of races that have already taken place. These elections also indicate sizable Democratic gains are on the table at the legislative level that could dramatically reshuffle the map of national partisan control ahead of redistricting. Our recommendation for Democratic legislative caucuses is that at minimum, anywhere either Clinton or Obama ran within single digits or better should be on their 2018 target lists at this early stage. And while these state legislative elections are crucial on their own — they play a big part in redistricting in many states — these results also should spook Republican candidates and operatives at the statewide and Congressional level.

Brian Stryker (@brianstryker) and Zac McCrary (@ZacMcCrary) are partners at the Democratic polling firm ALG Research

A note of thanks: this analysis owes very much to Daily Kos, which has an invaluable resource we use all the time for tracking special election results and Presidential performance by legislative district. All president-by-district and most special election results were taken from their herculean data-collection effort, with the balance from local election websites. The full data this article is based on is saved on this Google spreadsheet, which contains a sheet for 2017 results, two sheets for combined 49-state looks, and calculations from each individual state.