Wave Goodbye

Courtesy Crowd Sourcing Week

Courtesy Crowd Sourcing Week

Hillary Clinton’s resurgence in both national and state polling over the last two weeks has rekindled the notion among some that 2016 can be a wave election for Democrats. This line of thinking has been put into hyper drive after the release of the now-infamous tape of Donald Trump sharing his views on women. The general assumption is that if Clinton bests Trump by 5 or 6 points in the popular vote, her coattails will assuredly give Democrats control of the Senate and make eking out a majority in the House a real possibility.

Republicans have always had a tough path ahead of them this year: The Senate map is stacked against them, their majority in the House is bloated and they are due to lose seats, and with a historically disliked presidential nominee in Donald Trump, conventional wisdom held that the rout was on. With this in mind, should organizations plan on unified Democratic control in Washington DC?

This year has clearly been “interesting” to say the least, but the odds are still against a full wave developing.  To see why, first we need to figure out why and when waves happen, and what that means for the current election cycle.

When do waves happen?

Wave elections used to be evenly distributed between presidential and mid-term elections. If we define a wave election as a party gaining at least 20 House seats and at least 1 Senate seat, there were 21 wave elections from 1900 to 1980. This was an average of 2.6 wave elections every decade and about half of these were during presidential election years (10 in total) and 11 happened during the midterms. However, there has only been one wave election in a presidential election year since 1980, while midterms have continued to be more fertile ground for wave elections with three in the last 10 midterms. Overall this works out to 1.1 wave elections per decade. Going back to 1984, the party that didn’t control the Presidency has gained an average of 20.7 House seats and 4 Senate seats in midterm elections, while in Presidential election years, the President’s party gained just 2 House seats and lost less than half a Senate seat, on average.

As has been well documented, we are now in the era of hardening partisan preferences, where there are very few persuadable voters left to move. So without the possibility of mass defections from either party, what would it take to cause a wave election in the current political environment? Only a dramatic turnout imbalance between the two parties would reshape the electorate enough to cause such a wave. Unfortunately for Democrats hopeful that 2016 is their year, these massive turnout imbalances are incredibly tough to develop outside of midterms, where turnout patterns tend to fluctuate much more.

Why midterms are different

The incredible difference between midterm results and presidential elections has long intrigued political scientists. Since 1842, the President’s party has lost seats in 41 of the 44 midterm elections, and often they have been wave elections where the President’s party is sharply rebuked. A popular theory for this phenomenon was the “surge and decline” theory by Angus Campbell and James Campbell. It is based on the premise that presidential elections stir up a massive surge of voter enthusiasm and the President’s coattails allows the party to over-perform in the presidential year. Then, without the President’s coattails in the midterms, the electorate returns to its normal level which results in lower performance for the President’s party.

However, the “surge and decline” theory has waned in popularity over the years as a more nuanced understanding has emerged. The difference in the midterms and why the President’s party is usually the one to suffer is more likely due to a confluence of several factors: voters who are motivated to voice anger against the President, surge and decline, and reversion to the ideological mean.

A new normal?

All of the political science models we are looking at utilize a large range of elections to build their datasets. This has the benefit of allowing those aggregating the data to get a broad view of the midterm picture, but it may have caused them to miss more recent trends. We have repeatedly shown that voter patterns are hardening and becoming more predictable with each passing election. With this occurring, it makes the reversion to the mean ideology very unlikely. At the end of the day, if there are fewer and fewer persuadable voters, there simply won’t be much movement in the overall ideology.

The real differences then come from who turns out, as well as the anti-President voting that plays an important role. The three most recent wave elections (1994, 2006, and 2010) had very strong anti-President elements. In this case, the anti-President element means that the side which enjoyed the wave had fewer voters stay home in the midterm elections. Turnout will always be higher in Presidential years, so therefore we are dealing with a battle of attrition and the key to surviving is to make sure that less of your voters stay home the election two years after the presidential election.

If historical turnout patterns are analyzed, there is actually greater variability among the midterms than among presidential elections in recent years. Where the “surge and decline” model views the midterms as the return to the baseline and the presidential election as the unpredictable surge, we feel it is more appropriate to look at the presidential election as the baseline. If both sides are turning out at relatively even levels and there isn’t much party crossover, large gains in Congress simply aren’t possible. It is only when one side is more discouraged than the other that the potential for a wave exists.

Can 2016 be the next 2008?

It was fashionable in the months after Donald Trump sealed the Republican nomination to hypothesize that Democrats would be able to have a wave election this year. Taking back the Senate seemed easy and the House was within their reach. However, up until last week, the race had descended back into the trench warfare typical of presidential elections of the past few decades. Recent national polling has shown a clear bump for Clinton over the past five days, however there has not been much data showing big increases for Democrats down ballot. Currently, Democrats have an average lead in the generic head-to-head in national polls of 4.4 points[1] while the most recent NBC/Wall Street Journal poll has the Democrats with a 6-point average. To put this in context, the NBC/Wall Street Journal poll had the Democrats with a 13-point lead in early October in 2008.

Another motivating factor in 2008 for Democratic voters, or those leaning in that direction, was going from a Republican-held White House to a Democratic one, a situation which does not exist this year for Democratic voters.

If we look at the negative motivation in 2008, about two-thirds of voters disapproved of the job George W. Bush was doing as President in early October. Currently 42% feel the same way about President Obama, and a majority approve in most polls. Again, the motivation for new or less likely voters to turn out and vote down ballot is lower now than it was eight years ago.

Finally, in early October, then-candidate Obama was viewed favorably by 52% of likely voters. In current polling, Clinton is at 41%. Looking at this from the positive angle, there is also less reason to believe that voters will be as strongly motivated as they were in 2008 to come out and vote for the Democratic candidate and help Democrats down ballot at the same time.

In 2008, there were also a lot of motivating factors to get more Democratic voters out to the polls and we experienced an abnormal increase in voter participation among key demographics for Democrats, especially younger and minority voters. With less than four weeks to go, there is little data to point to anything but a normal turnout pattern for a presidential election year.

As the odds are still heavily against a wave developing, the Democrats are instead looking at small gains in both the House and Senate; they may not even win the Senate (although are favored to do so at the writing of this memo). This is exactly the scenario we projected in our PAAR models for the House and Senate. It will be interesting to see how far Democrats over perform the expected norm (if at all), but we will need to wait for November 9th to begin that process.

Nearly every regular voter on both sides of the political spectrum is motivated to turn out in Presidential elections and most people’s partisan preferences are set in stone. Without a huge imbalance in which side’s voters are turning out this year, it is virtually impossible for this cycle’s electorate to produce a wave election. Until there is a complete shakeup in voter-party alignment or the composition of either party’s coalition, wave elections will be regulated to midterms and presidential election years will be the standard base against base. 

If you like to play the odds, bet on a Republican held House and a closely divided Senate. On November 8th, we will know if this is another typical election year (outside of the Presidential race, clearly) or if 2016 becomes an outlier. But for those who need to start the planning process for 2017, the better odds are on a non-wave.


[1] Real Clear Politics on October 12, 2016. http://www.realclearpolitics.com/epolls/other/2016_generic_congressional_vote-5279.html