Photo by Joe Raedle/Reuters/Pool
Americans Are United … in Dissatisfaction with Their Choices
Analysis of the American Electorate at the Start of the 2016 General Election
In our continuing series on the RAND Presidential Election Panel Survey (PEPS), we're examining what Americans think about the upcoming presidential election. In addition to asking people about their voting intentions, we're asking about their beliefs and opinions on some of the nation's most-pressing issues and examining how these opinions change over time. In our analysis of the latest round of the RAND PEPS, conducted during and after the party conventions (2,601 total respondents between July 25 and August 25, 2016), we look at, among other issues, how respondents have changed (or not changed) their support for particular candidates between December 2015 and July/August 2016, as well as which combinations of beliefs and attitudes were associated with candidate support in this latest wave.
PEPS Wave 4 will follow up with the same group of respondents during the two weeks prior to the first debate between Democratic nominee Hillary Clinton and Republican nominee Donald Trump (September 26). A fifth wave will collect information following the final debate (October 19), and a sixth wave will collect information immediately following Election Day (November 8).
Clinton Versus Trump? The Fastest-Growing Segment Is "None of the Above"
Because our survey follows respondents over time, we can assess how candidate support evolves, as shown in Figure 1. Based on self-reported voting in 2012, we see that as of December 2015, Clinton supporters included not only a sizable majority of 2012 Barack Obama voters but also a substantial number of people who voted for neither Obama nor Mitt Romney in November 2012. Between December 2015 and March 2016, Clinton did suffer losses, as erstwhile supporters became uncertain, decided to support Bernie Sanders, and even went over to Republican candidates. But after Sanders’s concession, the Clinton campaign gained more supporters from his campaign than it had lost to it in the previous period. However, the Clinton campaign did not gain back as many supporters from the "unsure" category as it had lost to that category previously. Despite this, as of August 2016, Clinton’s total support base stood at a formidable 38.4 percent of the population.
While Clinton inherited the majority of Obama voters, Trump did not inherit the majority of 2012 Romney voters. Instead, his following came almost equally from those who voted for Romney and those who voted for neither Romney nor Obama in 2012. While Trump drew from this outside pool, his rivals split Romney voters into many small slices, amplifying Trump’s advantage. By March 2016, the Trump campaign had gained very little ground, losing roughly an equal number of supporters to others as it gained from them. However, none of his remaining rivals managed to unite the non-Trump Republican vote, and the number of non-Trump Republicans shrank as many became uncertain of whom to support. Following the respective concessions of his rivals (and the Republican convention), Trump gained about three quarters of Ted Cruz supporters, as well as about half of John Kasich and Marco Rubio supporters. As of August 2016, Trump’s total support base stood at 28.6 percent of the American population.
Both candidates may have cause to be concerned at losing a full third of their party’s supporters to third parties and voter uncertainty.
While Clinton and Trump fought their respective primary fights, Americans became increasingly uncertain and disaffected. In December 2015, the vast majority supported either a Democrat or Republican candidate; less than 15 percent were uncertain (listed as "unsure" in the PEPS) or supported a third party ("other"). That number grew to 22.8 percent by March 2016 and 33 percent by August. In other words, the number of people supporting a third-party run ("other") is now larger than the Sanders campaign at its height, and the number of Americans supporting neither major party ("other" + "unsure") is now greater than Trump’s total support base. Only 67 percent of people who favored a Democrat in December favored Clinton in August, and only 66 percent of people who favored a Republican in December favored Trump in August. Both candidates may have cause to be concerned at losing a full third of their party’s supporters to third parties ("other") and voter uncertainty ("unsure").
Global Integrationist Liberals Want Clinton, Isolationist Conservatives Want Trump, and Everyone Else Wants Better Options
In a previous report in this series, we examined how respondents felt about various political issues and then used mathematical techniques to identify two spectrums that together predict respondents’ answers to most ideological questions. One spectrum ended up looking remarkably like the traditional liberal versus conservative spectrum. The second was more complex, reflecting socioeconomic status, global isolationist ideology, and attitudes toward social groups outside one’s own. (To better understand the analysis presented here, we recommend reading that report and its follow-up.) We found that those with higher socioeconomic status also tended to be less likely to hold isolationist opinions and more likely to have positive attitudes toward social groups outside their own. In Figure 2, we group Americans according to where they fell on these two spectrums. While the groups are located all over the ideological map, the largest concentrations tend to be mildly liberal, mildly isolationist, and of lower-middle socioeconomic class.
The largest, most cohesive group of conservatives is mostly unsure, reflecting the unorthodox progression of the 2016 Republican primary.
Clinton supporters are mostly liberal, less isolationist, and more likely to have higher socioeconomic status. Trump supporters are mostly conservative, more isolationist, and more likely to have lower socioeconomic status. Proportionally speaking, supporters of third parties ("other") are most concentrated among wealthier, less-isolationist conservatives. Americans with centrist leanings comprise the largest segment of those who remain unsure of which candidate to support. Interestingly, the largest, most cohesive group of conservatives is mostly unsure, reflecting the unorthodox progression of the 2016 Republican primary. Trump has provoked an unusual amount of opposition among stalwart conservative leaders for his controversial statements and positions, creating uncertainty for some within the Republican base.
Figure 2 shows two elements that may interfere with Trump's chances of success. First, the groups with a majority favoring a third-party run (in yellow) are decisively conservative. This could lead to a third party splitting off a segment of the conservative vote. Moreover, while these voters are a relatively small slice of the conservative block as a whole, they are a large portion of wealthy conservatives. This could cut into Trump's fundraising pool, as illustrated by Republican fundraiser Meg Whitman's high-profile decision to back Clinton. Second, there is a region in the heart of Trump's ideological territory where the majority of voters are unsure (see the purple circle near Trump's label). This suggests that Trump may have difficulty maintaining unity among his core supporter groups. This friction among conservatives is embodied in the growing discord between members of the "alt-right" and more-traditional Republicans. However, Figure 2 is problematic for Clinton's chances too. Trump supporters are in the majority among at least three groups that fall on the liberal end of the spectrum. These liberal voters of lower socioeconomic status are more sympathetic to Trump than Clinton — a significant incursion into traditional Democratic party territory. Clinton's diminishing support for free-trade agreements may be a response to the strong feelings of this constituency.
Clinton Supporters Really Like Equal Opportunity, Trump Supporters Really Dislike Gun Control, and Both Have Intense Feelings About Immigration
Clinton and Trump supporters may not align exactly with the traditional party coalitions, but they do have strong differences on certain issues. Using a statistical method called logistic regression, we identified political belief items for which Clinton and Trump supporters were most likely to have differing opinions and then calculated the gap in policy support. For example, if 90 percent of Trump supporters favored issue X and only 60 percent of Clinton supporters favored that issue, then we would find a policy support gap of 30 percentage points.
Based on our analysis, Clinton supporters tend to be most ideologically united on efforts to provide currently disadvantaged groups with more equal opportunities in the United States. Compared with Trump supporters, Clinton supporters are more likely to agree that women and African Americans face unfair disadvantages and that the government should take action to level the playing field. This includes more-positive feelings about feminists and traditionally feminist issues, such as opposing bans on abortion.
Trump supporters tend to feel strongly about a more diverse group of issues, including antipathy for gun control and support for sending ground troops to fight the Islamic State (IS) terrorist group, and they have a general inclination to consider themselves more patriotic.
The two groups of supporters tend to disagree most on the issue of immigration. Clinton supporters are more likely to favor amnesty for unauthorized immigrants, while Trump supporters prefer building a wall along the Mexican border.
Figure 3 summarizes these significant policy support gaps, reporting differences on a given issue (e.g., wiretapping by the National Security Agency [NSA]), group (e.g., labor unions), or belief (e.g., "I'm very patriotic").
If Americans Could Vote Against the Candidates, Many Would
While Clinton and Trump supporters disagree on many topics, they are unified in their dissatisfaction with the candidates. For example, the vast majority of Americans think that Clinton and Trump are immoral and untrustworthy. In fact, less than 20 percent of Americans think the word "moral" describes either candidate "extremely well." Slightly more Americans believe this about Clinton (15 percent) compared with Trump (8 percent), but neither candidate is held in high esteem. In fact, when asked how well the word "moral" described the candidates, about half marked "not well at all" (see Figure 4).
In fact, dislike for the candidates is a major driving force in the 2016 presidential election. We asked our respondents for whom they would vote if given the chance to cast a positive vote for their candidate or a negative vote against their candidate's opponent. Based on their responses, we estimate that the number of Americans casting a negative vote would be about as large as the number casting a positive vote. In other words, among supporters of the two major parties, half would rather vote against a candidate than vote for one.
Among supporters of the two major parties, half would rather vote against a candidate than vote for one.
Broken down by pro and anti support, the percentage of pro-Clinton support (18.9 percent) is only slightly higher than pro-Trump support (16.7 percent); see Figure 5. However, negative votes shift the balance decisively in Clinton's favor, suggesting that much of Clinton's lead in supporters actually comes from Americans seeking a way to oppose Trump rather than from those who are enthusiastic about a Clinton presidency. If the election had been held in August, a negative vote option had been offered, and every American voted, Clinton would net 6.2 percentage points of the vote, and Trump would have a final tally of −1.9 percentage points of the vote.
These results differ according to the liberal versus conservative and global integrationist/higher socioeconomic status versus isolationist/lower socioeconomic status divides discussed earlier (see Figure 2).
Liberal isolationists are currently both the largest segment of the American population and the most undecided.
A strong majority of global integrationist/higher socioeconomic status liberals support Clinton. However, roughly two-thirds of that support comes from Americans seeking a way to vote against Trump; see Figure 6. The majority of conservative integrationists support Trump, with nearly equal numbers voting for Trump and against Clinton. The majority of conservative isolationists also support Trump, but respondents in this camp cast more votes for Trump than against Clinton. Liberal isolationists are currently both the largest segment of the American population and the most undecided. Some of them were Sanders supporters who have not decided to support Clinton or Trump, and many are torn between the appeal of Clinton's progressive platform and Trump's isolationist leanings. In this sense, the Democratic party's move to integrate much of Sanders's platform — for example, skepticism on trade agreements, enthusiasm for free college tuition, and a renewed focus on combating inequality — is likely a reaction to the mood among this large segment of the American public. In general, isolationist/lower socioeconomic status Americans (the bottom quadrants) are the most likely to support neither Clinton nor Trump at the moment. They likely represent the largest remaining opportunity for a candidate to gain support.
Concluding Thoughts: Clinton Leads, but the Most Significant Trend Is Overall Dissatisfaction
As of the end of August, Clinton retained a strong and consistent lead in support among the American public, which is not too surprising given the composition of the electorate. The United States in 2016 has a slight tilt toward the left (Figure 1), which is mostly counter-balanced by higher rates of turnout among conservative voters. Perhaps the most significant trend, however, is voter dissatisfaction. In terms of alignment, the number of people supporting neither a Republican nor a Democrat has gone from less than 30 million to more than 80 million in seven months (Figure 1), and only 28 percent of the country disagrees with the statement, "People like me don’t have any say about what the government does" (Figure 3). In terms of the candidates, 80 percent of Americans are not strongly convinced that either major-party candidate is moral (Figure 4), and 31 percent would rather vote against a candidate than for one (Figure 5).
In terms of the campaign, the ideological battle lines of this election are increasingly being fought at an angle to the traditional red versus blue lines, with both candidates winning support among groups that traditionally would have been part of the opposing coalition (Figure 2). Many Americans are restless with the traditional alliances underlying both parties (Figure 2) and are dissatisfied with their choices (Figures 1, 5, and 6). Unless the candidates can shore up support in their respective coalitions, the unresolved friction of this election season might carry over when the next president tries to implement the policies proposed on the campaign trail.
-  For complete details on the methodology behind the RAND PEPS, see Michael Pollard and Joshua Mendelsohn, Methodology of the 2016 RAND Presidential Election Panel Survey (PEPS), Santa Monica, Calif.: RAND Corporation, RR-1460-RC/UCLA, 2016a. As of June 10, 2016:
-  Some of these respondents would have been too young to vote, but most were indeed old enough to cast a ballot.
-  Joshua Mendelsohn and Michael Pollard, Politically Polarized, Ideologically ... Complicated: Mapping America’s Ideological Terrain, Part 1, Santa Monica, Calif.: RAND Corporation, RR-1625-RC, 2016a. As of June 30, 2016:
-  Socioeconomic status encompasses factors, such as income and education, that contribute to a person's position in society. It is similar to the idea of economic class.
-  Mendelsohn and Pollard, 2016a; and Joshua Mendelsohn and Michael Pollard, What the Ghosts of Candidates Past Can Tell Us About the General Election: Mapping America's Ideological Terrain, Part 2, Santa Monica, Calif.: RAND Corporation, RR-1650-RC, 2016b. As of September 9, 2016:
-  This figure builds on a previous analysis (Mendelsohn and Pollard, 2016a) that mapped the beliefs of the American public into a two-dimensional space based on respondents' answers to more than 30 survey questions, such that people who gave similar answers were placed at the same point on the map. The x-axis and y-axis were mathematically determined based on patterns in the answers. The x-axis roughly corresponds to liberal (left) versus conservative (right) differences, while the y-axis corresponds both to socioeconomic status differences (top = higher income and education) and to preferences for global integrationist or isolationist beliefs (bottom = more isolationist).
-  David A. Graham, "Which Republicans Oppose Donald Trump? A Cheat Sheet," The Atlantic, August 18, 2016. As of August 25, 2016:
- David E. Sanger and Maggie Haberman, "50 G.O.P. Officials Warn Donald Trump Would Put Nation's Security 'at Risk,'" New York Times, August 8, 2016. As of August 25, 2016:
- Robert Schlesinger, "Drip, Drip, Drip," U.S. News and World Report, August 5, 2016. As of August 25, 2016:
-  David A. Graham, "Which Republicans Oppose Donald Trump? A Cheat Sheet," The Atlantic, August 18, 2016. As of August 25, 2016:
-  Jonathan Martin, "Meg Whitman, Calling Donald Trump a 'Demagogue,' Will Support Hillary Clinton for President," New York Times, August 2, 2016. As of August 17, 2016:
-  Doug McKelway, "'Alt Right'" Conservative Movement Embraces the Trump Campaign," Fox News, August 25, 2016. As of August 25, 2016:
-  Domenico Montanaro, "A Timeline of Hillary Clinton’s Evolution on Trade," NPR, April 21, 2015. As of August 17, 2016:
-  The organization's name transliterates from Arabic as al-Dawlah al-Islamiyah fi al-'Iraq wa al-Sham (abbreviated as Da'ish or DAESH). In the West, it is commonly referred to as the Islamic State of Iraq and the Levant (ISIL), the Islamic State of Iraq and Syria, the Islamic State of Iraq and the Sham (both abbreviated as ISIS), or simply as the Islamic State (IS). Arguments abound as to which is the most accurate translation, but in the PEPS, we refer to the group as IS.
-  Each wave of the survey has asked respondents to opine on several policy proposals, social groups, and beliefs about society. For policy proposals (listed as "issues" in the figure), respondents had the option of marking that they "strongly favor," "somewhat favor," "somewhat oppose," "strongly oppose," or are "not sure" about the policy. For social groups, respondents had the option of marking that they have a "very favorable," "somewhat favorable," "somewhat unfavorable," or "very unfavorable" opinion or are "not sure" about a particular social group (such as feminists). For beliefs about society, they had the option of marking that they "strongly agree," "somewhat agree," "neither agree nor disagree," "somewhat disagree," or "strongly disagree" with the belief statement. Figure 4 uses data from these responses.
-  Figure 4 reports on respondents' answers to a single survey item. It follows the format established in Pollard and Mendelsohn, 2016b.
 In Wave 3 of the PEPS, we asked respondents the following:
Imagine that voters could choose to vote "against" a candidate instead of "for" a candidate. Each voter still has only one vote. Each "against" vote is called a "negative vote" and is counted as −1. The candidate who receives more "for" votes after the negative votes are subtracted wins the election. If the option to vote either "for" or "against" were available, how would you vote? (You can only select one option).
We gave respondents the option to vote for or against Trump, Clinton, or someone else, as well as the option to mark "will not vote" or "not sure," thus allowing voters to cast a vote either for a candidate (traditional vote) or against a candidate (negative vote).
The inclusion of this question in this round of the PEPS was funded by the Negative Vote Association, a nonprofit organization in Taiwan exploring the implications of adding a "negative vote" option to Taiwanese elections. In a U.S. election year in which voter dissatisfaction is at a record high, we thought that the concept held merit as a policy research question, so we allowed the Negative Vote Association to fund the fielding of this question; however, our analysis of the resulting data was not funded by the Negative Vote Association.
-  The July 25–27 and August 12–25 polls for Clinton and Trump are outside the margin of error (±2.8 and ±4.9, respectively), meaning that it is statistically unlikely that the difference between Clinton and Trump is due to sampling imprecision alone. The July 28–August 3 and August 4–11 polls are within the margin of error (±3.3 and ±4.7, respectively), meaning that the two candidates could have been tied, with sampling imprecision making it look like Clinton was ahead. The overall poll is outside the margin of error (±1.8), meaning that, during PEPS Wave 3 as a whole, Clinton’s lead over Trump was statistically unlikely to be the result of chance alone.
-  The projected election results are calculated by multiplying the likelihood of voting by the likelihood of voting for the Democratic party candidate, the Republican party candidate, or someone else.
-  Michael Pollard and Joshua Mendelsohn, "RAND Kicks Off 2016 Presidential Election Panel Survey," blog, Santa Monica, Calif.: RAND Corporation, 2016b. As of August 25, 2016:
-  Michael Pollard and Joshua Mendelsohn, "The Latest from the RAND Presidential Election Panel Survey," blog, Santa Monica, Calif.: RAND Corporation, 2016c. As of August 25, 2016: