Every presidential election is different. But there is different and then there is 2016.
As the nomination contests and their surviving candidates have made clear, this year’s election looks to be very different. Most notably, neither of the two highly controversial major party candidates are well thought of by an unusually large part of the electorate. Substantial majorities of Americans hold generally unfavorable rather than favorable views of both Democratic Party candidate Hillary Clinton and Republican Party candidate Donald Trump.
On the Democratic side, despite being her party’s ultimate establishment candidate, Hillary Clinton enters the campaign with a warehouse of controversies (e.g., huge Wall Street speaking fees, the handling of the Benghazi fiasco, and the Clinton Foundation’s fund-raising). This includes a year long FBI investigation into her mishandling of confidential and top secret emails on private unsecured computer servers while serving as Secretary of State. Though unindicted for handling classified information “in a grossly negligent way,” the FBI characterized her actions as “extremely careless.”
On the Republican side, while lacking the experience necessary for a long history of controversies in public office, Donald Trump has his own trail of business scandals (e.g., Trump University, multiple bankruptcies) and continues to demonstrate an uncanny flair for gratuitous insults and outlandish offenses (e.g., mocking a reporter with a disability, implying Ted Cruz’s father was implicated in JFK’s assassination, and other offensive comments directed at ethnic groups, women, and public figures of both parties). Regarding Trump as shallow, unprincipled, vulgar, boorish, and belligerent, many pragmatic conservative (aka, establishment) Republicans at least tentatively have decided to sit this election out, leaving the party unusually fractured even after its convention.
With all of this, the specter of a Clinton versus Trump race suggests that we should all expect the unexpected, but what should be expected? That is where election forecasting models come in.
In this symposium, we have again gathered eight election forecasters or forecasting teams to present their models’ predictions of the 2016 national two-party presidential vote. Four also venture forecasts of the congressional elections. A ninth forecasting team offers a presidential prediction based on combining a number of forecasts based on very different approaches.
Each of the election forecasting models is based on a statistical analysis of historical data capturing how different pre-campaign contexts (aka, the fundamentals) have predicted vote divisions in the past. These predictors tap into the electorate’s early vote inclinations, its predispositions that affect reactions to developments during the campaign, and performance and issue information likely to be considered in the course of the campaign. An electorate favorably disposed initially to the in-party and its candidate, in generally a good mood, and having positive information to be raised during the campaign is likely to elect the in-party candidate. Reverse the initial disposition, the mood, and the valence of information to be digested in the campaign and the outcome is also likely reversed.
So what is the context leading into the 2016 national elections? There are a number of ways to take a reading of the context, but before getting to the individual forecasts, it might be useful to review six indicators of the prevailing winds in this year’s general election campaign.
Open seat elections or, more accurately, elections in which the in-party is seeking more than a second term in the White House are historically much more competitive than elections in which a party has held the presidency for only a single term (Abramowitz 1988; Campbell 2000a; 2000b; and 2010; Mayhew 2008; Norpoth 2000; Weisberg 2002) A comparison of the pair of bars on the left side of figure 1 illustrates the difference. A party having held the presidency for a single term and seeking a second has won seven out of eight times since 1948. Carter’s defeat in 1980 was the only loss. Going back further in electoral history leaves the conclusion intact: second terms are nearly sure things. In contrast, parties seeking to extend their presidential tenure beyond a second term face competitive contests. There have been nine such elections since 1948, six of them open seat elections, and all have results within the range of 45% to 54% for the in-party. Democrats in this election are seeking a third consecutive term. Outlook for 2016: Close Race.
Going back further in electoral history leaves the conclusion intact: second terms are nearly sure things. In contrast, parties seeking to extend their presidential tenure beyond a second term face competitive contests.
Since the realignment of the parties and their fortification by a polarized public (Campbell 2006 and 2016; Abramowitz and Saunders 1998; Abramowitz 2010), presidential elections have become hyper-competitive. The right-side pair of bars in figure 1 compares the spread of the in-party presidential vote in 10 elections from 1948 to 1984 to the seven elections since then. The standard deviation of the vote in the more recent series is only about half of what it was in the earlier period (3.3 compared to 6.5). Figure 2 may explain why. The parties are now much more dependent on their ideological bases than they had been. Democrats draw a bigger percentage of votes from liberals than they had and Republicans draw more from conservatives than had before. A Pew poll in April found that 41% of Democrats and 45% of Republicans think their opposing party is a “threat to the nation’s well-being” (Pew Research Center 2016). Polarization has hardened party lines, reduced the volatility of the electorate and compressed the vote division. Outlook for 2016: Close Race.
Although its electoral influence is routinely exaggerated, the growth in the economy is important both directly as an issue and as a backdrop for the election, a factor affecting the public’s general perception of the in-party’s performance and policies. The economy through a presidential term matters, but the most crucial period of economic performance is in the second quarter of the election year. This is recent enough to affect how many voters see the trajectory of the economy as their candidates preferences are taking form.
So how has the economy been doing? Figure 3 displays the quarterly growth rate of the real GDP following the end of the Great Recession in June 2009, according to the National Bureau of Economic Research (2016). As one might gather from the Federal Reserve’s stimulus monetary policy, the recovery from the Great Recession has not been robust. The public appears to expect growth of about 3 percentage points and that has been about the post WWII average. Even in recent decades (1990–2006), real GDP in the median quarter has been 3.1 percentage points. Over the last six years, however, growth in the median quarter has been a full percentage point lower. The August growth rate report for the second quarter this year (1.1%) is even weaker, a good deal below average and the politically neutral.
Although anemic growth works against the Democrats, the impact should be somewhat muted by the fact that the incumbent is not in this year’s race (Nadeau and Lewis-Beck 2001; Campbell, Dettrey and Yin 2010; Holbrook 2008). Outlook for 2016: Tilt to Republicans.
Public opinion prior to the start of the general election campaign can be read at different levels of specificity. At its most general level, Americans have been asked how things are going in the nation. A very large majority of Americans are dissatisfied with the direction of the nation. When you combine the sluggish economy noted above with national and international terrorism and racial unrest, it is not surprising that fewer than a quarter of Americans polled in the first three weeks of July think America is moving in the right direction, while nearly 70% think the nation is on the wrong track (Real Clear Politics 2016). A similar satisfaction question asked by Gallup in June arrives at a slightly higher satisfaction number, but still under 30% (Gallup 2016). This sour attitude is not good news for the party that has held the White House the last seven plus years. Unhappy voters are inclined to make a change. Outlook for 2016: Favors Republicans.
A second public opinion barometer often used in forecasting models is the presidential job approval rating. When the public generally approves of the president’s performance and record, voters are inclined to vote for continuity. A generally dissatisfied public is inclined to vote for change. These ratings are also colored by partisanship, polarization, and incumbency. In open seat elections, for instance, presidential approval ratings may not be quite as indicative of support for the in-party as they are when an incumbent president is running. Presidential approval ratings for the departing presidents in the seven open-seat elections since 1948 are ranked along with the election’s outcome in table 1. There were two decisive in-party losses with presidents rated in the low 30s or lower (1952 and 2008) and one decisive win with a president rated in the mid-50s (1988). President Obama’s rating of around 50% suggests the election could go either way. Outlook for 2016: Close Race.
While the public’s views of national conditions and the sitting president’s job performance offer important clues about what to expect from voters in November, their early assessments of the candidates also are suggestive precursors. Elections are normally a good deal more competitive than preconvention polls would indicate, but they still provide a clue about the vote division (r with vote of .74). The four major (non-internet) preference polls in the week prior to the first (Republican) convention had Democrat Hillary Clinton with a narrow lead (52% to 48%) over Donald Trump, but the four polls had undecided voters ranging between 9% and 20%. Outlook for 2016: Tilt to Democrats.
These early signs and readings of the context in which the campaigns will be run would suggest that 2016 was shaping up with a tilt to the Republicans until they nominated the bombastic Donald Trump. It was not a “sure thing” for the Republicans before their nomination and it (amazingly, to briefly editorialize) does not appear to be a “sure thing” for the Democrats after the Republicans’ astonishing nomination. But though informed, this is loose conjecture. A more systematic appraisal combining many of these fundamentals awaits the models explained in the following articles.
Although each model has its own track record and the accuracy of each forecast should stand judgment on its own, as background, it is worth bearing in mind that many of these forecasts were quite accurate in the 2012 election.
Each of the forecast models in the symposium (with the exception of the Pollyvote combining forecast) is based on a statistical analysis of historical data capturing how different pre-campaign contexts have predicted the vote divisions in past elections. Each has different sets of predictors, different data sources, different specifications and measurements, and different timings. They have different histories, as well. Some forecast models have long records with minor tweaking over the years and others have undergone major renovations. Over time these revisions have produced some convergence in the models’ predictors and specification, including timing (when a model’s forecasts are most accurate). Although each model has its own track record and the accuracy of each forecast should stand judgment on its own, as background, it is worth bearing in mind that many of these forecasts were quite accurate in the 2012 election.
Seven of the forecasts made more than eight weeks before the 2012 election and, in several cases more than three months before the election, were within one and a half percentage points of the actual two-party popular vote (Campbell 2013).
To provide an overview and facilitate your comparison of the forecasts, I have extracted and organized the predictions of the models in table 2. Six of the 11 forecasts predict that Hillary Clinton will defeat Donald Trump with 51% or more of the two-party presidential vote. Two predict a Trump victory over Clinton by the same margin. The remaining three forecasts predict a very close election. The median forecast predicts that Clinton will win 51.1% of the two-party national popular vote.
Finally, while all eyes are understandably turned on the presidential election spectacular (like passing a car accident on the highway), there are important congressional races as well. For these races, the context more clearly favors one party: the Democrats. After their gains in the 2010 and 2014 midterm elections, Republicans hold more seats in the House of Representatives than at any time since 1931. They are at an 85-year high water mark. They are unlikely to gain any more seats and unlikely to hold steady at their current numbers. In the Senate, after their large gains in 2010, Republicans are defending more than twice the number of seats as Democrats. With Republicans defending 24 seats to only 10 seats defended by Democrats, Republicans are quite likely to sustain significant losses. But again, these are general expectations, the analyses of the forecasting models in four of the following articles provide a more systematic and specific expectations. Table 3 collects the basics of these four forecasts.