This is a guest post by Craig Allen Smith, Professor Emeritus, North Carolina State University
Donald Trump stunned America and the world by defeating Hillary Clinton in the 2016 American presidential election. Twelve days ago I wrote of the candidates’ rhetorical puzzle and suggested three strategic challenges facing them. Clearly, Trump addressed those challenges with greater effectiveness. Today’s post is an attempt to explain how he did so.
Closing the Gap
The U. S. Constitution made American presidential elections a tournament of state elections, not a national popular vote. Candidates therefore need a plurality of votes in states that control at least 270 electoral votes. State polls of “likely voters” aggregated daily by elecoral-qote.com, 270towin.com, and 538.com projected the states that seemed “safe” or “likely” for Clinton and Trump. My post of 25 October used the Electoral-Vote.com tipping point to report that Clinton needed to win those states where she then led by 6% (or more) whereas Trump needed to win every state where he trailed by 6% or less, and I inferred that Trump needed to dramatically improve his standing in Pennsylvania (-6%) and Nevada (-5%) to have a chance.
By Election Eve of 7 November, thirteen days of heavy campaigning and reporting had changed that scenario. The electoral-vote.com tipping point showed Clinton needing to win every state she led by a statistically insignificant 3% (not 6%). Trump had solidified his leads in Texas and Mississippi such that his +5% states offered 161 electors (up from from 100), flipped both Arizona (from -4% to +2%) and Nevada (from -6% to +1%), and maintained his slight leads in Ohio, South Carolina and Georgia. The states where Trump led by even 1% thus offered 221 electoral votes, but winning all of those states would still leave him 49 electors short of victory.
Where might Trump find those 49 electors? State polls had shifted toward Trump to create a tie in Florida (29) and a statistically insignificant 2% Clinton lead in North Carolina (15). By beating those polls’ margin of error Trump could reach 265 electors. Moreover, Clinton’s leads in Michigan (16 electors), Colorado (9), and New Hampshire (4) had dropped to 3%. In short, by election eve Trump had moved into position to win 294 electors by winning the states where polls found him within a statistically insignificant margin of 3%.
Converting Polls to Votes
The campaign to “get out the vote” (GOTV) is crucial. Barack Obama’s GOTV efforts twice resulted in historic vote totals, and the Clinton campaign followed his model. In October electoral-vote.com reported that “The Democrats have 5,100 paid staffers in the battleground states. The Republicans have 1,400.” By their convention in July Trump’s advisors reasoned that winning the Republican nomination would commit the Republican Party to do his GOTV work for them. That fit the needs of Republican leaders wary of Trump who were investing their resources in “down ballot” races for the Senate, congressional seats, and governorships. Although many of Trump’s supporters disliked establishment Republicans nearly as much as they disliked Democrats he could be confident of their votes. What Trump needed was a concerted Republican effort to turn out traditional Republicans who disliked him.
As of this writing the incomplete vote totals show that Trump and the Republicans outperformed the polls in the states he needed to win. Specifically, his votes exceeded his polls in South Carolina (by 13.6%), Ohio (6.6%), North Carolina (5.8%), Michigan (3.3%), New Hampshire (2.8%), Georgia (1.7%), Florida (1.4%), Colorado (0.8%), and Arizona (0.2%); Trump’s vote fell below his polls only in Nevada (-2.4%). More impressively, he took 30 electoral votes from Clinton’s column by outperforming his polls in Wisconsin by 8% and Pennsylvania by 5.1%. By outperforming the polls Trump won those states’ electoral votes, 128-19.
Bad Polling or Impressive Turnout?
Should we infer that the polls were wrong or that Republican turnout was impressive? The available evidence suggests both. The Republican turnout hypothesis finds support in their down ballot senate races. Republicans handily won the “close” senate races in Indiana and Florida by 8%, won three of the five “toss-up” races, and won the Wisconsin seat in an upset. The Republican turnout effort trumped that of the Democrats.
But polling was problematic as well. The “surprise” states were Wisconsin, Pennsylvania, Michigan, North Carolina, and Florida; but because none put Trump behind on election eve by more than 5%, they were not technically wrong. More problematic were the interpretations of the polling data by reporters, who paid too little heed to the statistical margin for error. But dare any pollster claim success when the outcome is so far off the mark?
One poll — the LATimes/USC poll is being highlighted as “the one that got it right”. It daily tracked a panel of about 3,000 voters and asked panelists for their candidate preference and their likelihood of voting. For several months it was the outlying poll that showed a Trump lead. But notice that the LATimes/USC poll projected a 3.2% Trump win of the popular vote that he currently trails by 0.2% and it paid no attention to their respondents’ states or electoral potential. It seems apparent that their method should guide future polling and extend it to the state level, but it seems equally apparent that it, too, was less than prescient.
The American people have been polarizing for many years. We began migrating into likeminded communities in the 1970s, then found ways to watch likeminded cable television channels and niche news, then accelerated the process through social media. Today all Americans and many people around the world are asking who we are as a people. Consider this election day paradox: 69.2% of respondents said the the country was on the wrong track and 56% approve of President Obama’s job performance (only Bill Clinton had greater approval eight years after his election).
How, then, did American voters sort themselves into the Trump and Clinton coalitions? According to data from CNN’s exit polls, Trump found votes among married white males over 45 who did not graduate from college and but earn more than USD $50,000 and attend religious services at least monthly. Clinton found votes among unmarried non-white females under 45 who graduated from college earn less than USD $50,000, attend religious services infrequently.
Perhaps more informative than the demographics are the issue positions associated with the candidates. These positions permeated the candidates’ rhetoric in speeches, debates, and advertising. Trump found votes among those who disapprove of President Obama and are angry about the federal government for doing too much and want a president who can bring change especially with regard to immigration and terrorism by deporting illegal aliens, building a border wall, escalating the fight against ISIS, and changing trade deals to increase jobs. Clinton found votes among those who approve of President Obama, the federal government, international trade, and progress in the fight with ISIS and who want the federal government to do still more by offering legal status to illegal aliens without building a border wall.
The CNN data help to explain the apparent paradox between approval of Obama and disapproval of the direction of the country — the numbers reflect different populations. And yet they suggest new paradoxes within the Trump coalition. First, Trump voters want the federal government to do less but want it to build a border wall, deport illegal immigrants, and change our trade agreements. Second, television commentators attributed the anger and call for change to economic dislocation, but Trump did better among those earning more than USD $50,000 annually and lost among those who regard the economy as the most important issue. Those paradoxes deserve our attention.
Resolving the Paradoxes: 7 Hypotheses
There will be no shortage of commentaries seeking to explain this election. Let me simply propose some observations and hypotheses for further study.
- This was not a referendum on two personally unpopular candidates. The exit polls undermine this canard. Only 18% of voters regarded both candidates unfavorably and just 2% regarded both favorably; 41% regarded only Clinton favorably and 36% regarded only Trump favorably. There is a disconnect between the media narrative and the exit poll data because fewer than one in five voters looked unfavorably upon both candidates.
- Democrats were unlikely to win with a different nominee. We are now hearing that Bernie Sanders, Elizabeth Warren, or Joe Biden could have won, which presumes that a candidate to Clinton’s left could have won more votes. Perhaps. But Clinton won 84% of liberal votes and 52% of moderate votes —specifically, 90% of Democratic women and 47% of independent women. Among voters 35% were conservative, 39% independent, and only 26% liberal. The additional votes needed to prevail were unlikely to be found among liberals or Democratic women, but from the larger pools of independents and Republican women. Indeed, Sanders worked with Clinton to craft “the most progressive platform ever” and it ran headlong into a right of center electorate.
- Trump’s vote emerged late. CNN’s exit polls show that Clinton led 52-45% among the 60% of voters who decided by the end of August; Trump won among the 13% who decided in September, the 12% who decided in October, and 14% who decided in November. Surely, the September shift is partly attributable to Clinton’s pneumonia that seemed to validate warnings about her general health and may have contributed to her unfortunate reference to Trump’s supporters as a “basket of deplorables”. Just as surely, the November shift must be at least partly attributable to the FBI’s reopening of their email investigation eleven days before the election (by the time they closed the inquiry ten days later millions of votes had been cast). But the October shift toward Trump is more curious.
- Trump won October despite the Debates and the Video. October was dominated by two presidential debates and the vice-presidential debates. They drew large audiences and there was a general consensus that Clinton won them going away, but 57% of Trump’s supporters said that the debates were not important to their decision. Between the first two of those debates a video emerged in which Trump was heard saying lewd and unseemly things about his approach to women. Yet Trump won October deciders 51-37% apparently because his converts were undeterred by those episodes; indeed, the October deciders may have been alienated by the Clinton campaign’s attention to the video.
- Republican GOTV efforts used the Supreme Court to broaden Trump’s vote. The success of the Republican effort to turn out the vote can be seen in their success in other races. One question suggests how they motivated Republicans wary of Trump to vote for him: 21% of all voters said that the next president’s Supreme Court appointments were the “most important factor” in their decision, and 56% of them voted for Trump. This seems significant because it suggests that 11.7% of all voters chose Trump primarily to gain control of the Supreme Court — a number of voters sufficient to swing several states all by itself..
- The election does not show that Americans are racist sexist, and nativist. Many Americans are struggling with this hypothesis today. We should note, first, that Trump was elected with less than a majority of the popular vote. Second, we calculated that 11.7% of those Trump voters considered the Supreme Court nominees to be the most important factor in their decision. Third, we saw that Trump won the late deciders 47-42% to overcome a 47-49% deficit among the 85% who decided prior to the last week of campaigning. It seems unlikely that so many voters would become suddenly racist, sexist, and nativist just a week before voting. Trump did build a constituency by fanning the embers of prejudice and intolerance, but many who ultimately voted for him did so with difficulty and for other reasons.
- This election was was a conflicted call for change. When CNN asked voters what candidate quality mattered most to them they heard four responses: Can bring change (39%), right experience (21%), good judgment (20%), and cares about me (15%). Trump won 83% of the 39% who wanted change (or 32% of all voters). But those voters re-elected 26 of 28 incumbent senators and 432 of 435 members of congress. It appears that incumbents represented their constituents who want their adversaries to compromise or change.
The 2016 American presidential election will long be studied as a shocking upset. One of the world’s wealthiest men created a populist movement to become the only American president with no military or political experience. He ran a unique campaign: he used his own money rather than raising funds, he had a small staff and ran through three campaign mangers, he relied on Twitter far more than television ads, he repeatedly charged that the polls and election were “rigged” against him, and he insulted key Republicans, Mexicans, Muslims, refugees, women, and the persons with disabilities. He proceeded to defeat 16 Republican rivals and the party establishment, then defeated Clinton — a former First Lady, Senator, and Secretary of State supported by two popular presidents. Pollsters and journalists, as well as citizens, were blindsided by the outcome.
This post has shown that Trump narrowed Clinton’s lead in the final two weeks to statistically insignificant leads in the crucial states. The Republican party’s turnout efforts and concern for the Supreme Court apparently induced many Republicans and independents to grit their teeth and join the enthusiastic Trump base. Only 18% regarded both candidates unfavorably and Clinton held the advantage on favorability. Moreover, potential Trump voters cared little that Clinton won the debates, and they may have recoiled at the Clinton campaign’s attention to the lewd video.
During August many journalists, pundits, and politicians foretold the demise of the Republican Party, but that party now holds the White House, the Senate, the House of Representatives, most state governorships and stands ready to approve a conservative nominee to the U. S. Supreme Court. Fissures within the Republican community will go a long way to deciding the nature of the Trump presidency.
Now it is the Democrats who are in shocked disarray. More Democratic than Republican congressional seats are at risk in 2018. If the new administration shrinks the federal government by delegating responsibilities to the states, it is Republicans who control most states. And by 2020 everyone who has been considered a potential Democratic president — Hillary Clinton, Bernie Sanders, Joe Biden, and Elizabeth Warren — will be well over 70.
But the pressing question today is how a polity so polarized will find the common ground necessary for effective self-governance. The argument over the White House has ended, and a new conversation requires mutual understanding and trust that are difficult to find.