When Benjamin Franklin emerged from the Constitutional Convention in Philadelphia in the summer of 1787 he was asked what form of government had been agreed upon? Franklin replied, “A Republic, if you can keep it.” Our analysis of World Values Survey data from countries as diverse as the U.S. Japan, Chile and Indonesia shows why Benjamin Franklin was right to be moderately skeptical.

A lively debate is in progress to explain why “the end of history” has been partially upended. Francis Fukuyama finds the economic stagnation of working class incomes to be the root of the anti-establishment politics in 2016 in both the Republican and Democratic parties. Ronald Inglehart and Pippa Norris reject economic determinism and utilize a cultural backlash theory to explain rising support for populist political parties in Europe and the United States. Inglehart and Norris say older, white, less educated males have become alienated over the past half century, rejecting the post-materialist consensus of governing elites and turning instead to the xenophobic nationalism offered by “strong leaders.” Roberto Foa and Yascha Mounk identify youth and withdrawal from political participation as characteristics of new authoritarianism in Europe and the United States.

This article contributes to the on-going debate on anti-democratic attitudes in the United States, Japan, Chile, and Indonesia. It is part of a larger SAIS project on authoritarianism and democracy in twelve countries, using data from four different rounds of the World Values Survey (1995, 2000, 2005, and 2010). Our conclusion is that democracies remain fragile for political reasons, not because of social class or culture clash, but because democracies contain within themselves substantial population blocs which are either ambivalent about democracy or opposed to it, and these groups, under particular circumstances and with the right leadership can be mobilized to weaken or destroy democracy. Mass attitudes about the legitimacy of the democratic political structure are a reality that transcends cultural, economic and class distinctions, and these well-established political beliefs form a potentially explosive element that may be difficult for public policy to defuse.

In 2016 Donald Trump trumped the Republican Party, and the Democratic Party felt “The Bern” of socialist Sanders. Only one of the candidates was authoritarian by personality, but what is most striking is the intolerant certitude of both sets of followers. As divergent as these two movements were ideologically, they were united in their low trust in the system and intolerance of opposing viewpoints. What happened in the United States was a rebellion against party structures by groups who felt their interests ill-served by established party leaders and who became attracted to simplistic solutions such as “revolution,” “building a wall” and ending bad trade deals.

In 2011, before anyone ever thought about “Trump-for-President,” our data show 40 percent of the American population was expressing distinctly non-democratic views, providing the raw material for anti-establishment political insurgencies. In the United States non-democratic groups have been increasing in size over the last 20 years, and are driven primarily by social rather than economic factors. Data from Japan, Chile, and Indonesia reveal robust non-democratic groups within these functioning democracies. For example, in Indonesia in 2011, those who were ambivalent or alienated from democracy equaled 56 percent of the adult population (substantially higher than in the U.S.). This suggests how tentative support for democracy is in the 3^rd largest democracy in the world. Likewise in Chile, discontent with democracy remains ominously large. In Japan, the support for democracy attains a middle position between the United States, on the one hand, and Indonesia and Chile on the other. But in all of these examples the proportion of adults who either feel ambivalent about democracy or reject it outright is high enough to serve as the basis for destabilizing non-democratic movements.

Each wave of the World Values Survey includes four questions posing radically different alternative forms of government. The respondents are asked to evaluate each political system.

After more than two centuries of uninterrupted democratic rule, in 1995 nearly everyone interviewed in the United States (93 percent) says democracy is “very good” or “fairly good,” but unexpectedly they do not universally reject army rule, government by a strong leader, and a system run by “experts.” In 1995, 24 percent endorse “having a strong leader who does not have to bother with parliament and elections.” Also, 36 percent of Americans are attracted to a society run by technocrats rather than the government. Seven percent even endorse army rule.

In the United States between 1995 and 2011, support for the democratic political system declined while ambivalence and alienation from democratic government grew. Support for strongman rule grew to 31 percent, for technocratic rule to 49 percent, and even army rule increased to 16 percent while positive evaluations of the democratic political system decreased by 9 percent from 1995 to 2011.

Many of the 2,232 U.S. respondents in 2011 indicated their ambivalence about democracy by selecting opposites, for instance by endorsing both democracy and having “a strongman who does not have to bother with parliament or elections.” One-quarter of the Americans who say democracy is “very good” or “fairly good” also say that rule by a strongman is “very good” or “fairly good.” In other words, a quarter of all American adults are ambivalent about the difference between democracy and strongman rule. In addition, another 7 percent are pure authoritarians (saying rule by a strongman is good and democracy is bad), and 10 percent reject both democracy and authoritarian government. If we combine the 25 percent who are ambivalent, with the 7 percent who are pure authoritarians and the 10 percent who are alienated from both kinds of political systems, 40 percent of the adult population of the United States supported the democratic political order less than wholeheartedly in 2011. In fact, the proportion of pure democrats fell from 71 percent in 1995, to 64 percent in 1999, to 61 percent in 2006, to 59 percent in 2011. Before anyone had even heard of Donald Trump, there was a very considerable bloc of adults ready to be roused to challenge the current system.

Human beings are very complex, particularly with regard to their political attitudes and how these attitudes relate to actual behaviors. But which variables best explain the differences between the pro-democratic and more authoritarian respondents: age, gender, ethnicity, income, employment, social class, education, media exposure, involvement in non-electoral forms of participation, membership in civil society organizations, or something else?

In the United States most of the economic variables, such as income differences, did not make significant independent contributions to distinguishing between those who are inclined toward non-democratic as opposed to democratic political outcomes. This is true, even though non-democratic blocs in both political parties in the U.S. tout the rhetoric of economic discontent. Those who are more authoritarian do not report they are any more dissatisfied with their personal economic situation than those who are more democratically inclined. Similarly, those at the authoritarian end of the scale are not more likely to be unemployed. Variations in the ability to save do not predict authoritarian attitudes. Differences in reported economic well-being do not consistently differentiate between those who are pro-democratic and those who are less democratically inclined. Favoring a democratic type of government involves more than economic determinism; otherwise Singapore and China would be democratic and India and Indonesia would not. To steal a phrase from Mother Goose, more seems to be involved than just “which little piggies had roast beef and which little piggies had none.”

Second, there is no strong relationship between perceiving oneself as “middle class” and whether an individual favors an authoritarian system. Accumulating a brace of consumer goods and adopting a middle class identity do not cause individuals to favor democracy. From Weimar Germany in the 1930s, to Thailand and China today, perceiving oneself as “middle class” is not a particularly powerful or universal predictor of unqualified support for democracy. In contrast to other studies, we find no dramatic shift to authoritarian attitudes among “the rich” in the United States.

Third, social and political factors are driving attachment or detachment from the democratic political system in the United States. The single most important factor is exposure to education, and this is true for 1995, 1999, 2006, and 2011. Those who are less educated are more inclined to authoritarian methods of governing. The second most important factor is age. Apparently, Americans grow not only older and wiser but more democratic, with those 18-29 more likely to be impatient with democratic ways while individuals over fifty are more likely to reject different forms of authoritarian rule. Ethnicity plays a strong role. African Americans and Hispanic Americans are more inclined to authoritarian forms of government than Caucasians. Whether one is a Southerner, even a white Southerner, does not predict the inclination to either democracy or authoritarianism. This finding casts doubt upon the cultural backlash thesis in which old, white males are supposed to be turning to authoritarian populism. In three of the four waves of the WVS there were no differences between men and women in their support, or lack of support, for democracy. However, women in 2011 in the United States are more likely to be attracted to non-democratic systems than men. This gender difference was also present in the 1999 wave of the WVS but it did not reach statistical significance. The jury is still out but apparently gender does not consistently predict devotion to the democratic form of government.

Fourth, there is good news for the survival of democracy in the United States. Even though the proportion of the general population holding authoritarian views may be higher than anticipated, the good news is that non-democratic respondents are less interested in politics, less involved in day-to-day political activities between elections, and they are less likely to vote in national elections. Even though 40 percent of the adult population in 2011 may be ambivalent or hostile to democracy, these same individuals are systematically less likely to discuss or engage in politics. Although 40 percent of the population may have the potential to be mobilized in favor of non-democratic alternatives, the authoritarian bloc is systematically less willing to engage in politics. The question for 2016 is whether the personality of a Trump might be able to shake the discontented minority out of its normal political torpor. Anecdotal evidence from 2016 suggests that Trump accomplished this in the primaries, but there is probably a ceiling beyond which his appeal will not climb in the general election. Watch this space.

Fifth, the non-democratic forces in America are split among Democrats, Republicans, and Independents. In 2011 only 60 percent of respondents identified themselves as either Republicans or Democrats, but self-identified Democrats were slightly more likely to support non-democratic political systems. However, the important fact is that the non-democratic forces were spread across the parties and the independents. This makes it unlikely they will coalesce around an anti-democratic candidate from any one party.

How do the patterns just described for the United States differ or remain the same in places such as Indonesia, Japan, and Chile? All politics is local, and it should not be surprising that there is no universally applicable set of variables accounting for undemocratic attitudes.

The proportion of U.S. respondents identifying with the democratic political system has been declining over the past twenty years, but it is still 13 percent larger than in Japan and 26 percent larger than in Chile. Chilean respondents have become more positive about democracy across the four waves of the WVS, moving from just 25 percent in the upper half of the democracy scale in 1996 to 30 percent in 2000, 34 percent in 2006, and 36 percent in 2011, but the proportion of non-democratic to democratic citizens is still roughly 2:1. Women who are less educated, employed in jobs requiring limited skills and less interested in politics or in participating in politics between elections are more likely to favor more authoritarian forms of government.

Japan’s case is difficult to interpret. There is no clear trend, and roughly 30 percent of respondents opt out of answering the four-part question by selecting “don’t know.” Thirty-seven percent of Japanese respondents were pro-democratic in 1995, increasing to 40 percent and 42 percent in 2000 and 2005, only to decline to 32 percent in 2011. In Japan, being female, less educated, and less interested or participatory in politics remain powerful predictors of non-democratic attitudes. Economic variables such as income, employment, social class, and occupation do not distinguish between being pro-democratic from being non-democratic.

Our findings reveal no mono-linear, universal trend among the countries. Attitudes change with the time in response to circumstances. One finding that is characteristic of all four countries is that authoritarian views correlate highly with lack of interest and low levels of participation in politics during and between elections. Those who are ambivalent and the alienated from the political system are consistently less involved in making choices for the political system.

U.S. attitudes remain more pro-democratic than Japan, and much more democratically inclined than in Chile or Indonesia. History may play a role here. The United States has never tasted the bitter fruit of dictatorial rule. Japan’s constitutional framework has been remarkably stable and very difficult to change over the last 65 years. Chile is often touted as the Latin American beacon of democracy, but the attitudinal data reveal the persistence of a large non-democratic bloc, and the same is true in Indonesia. Perhaps this can be partly understood because Indonesia and Chile experienced rapid economic growth under dictatorial rulers less than twenty years ago. For some portion of the adult populations, Pinochet and Suharto did not leave wholly negative legacies. In the alchemy of public opinion, dead dictators somehow gain partial absolution.

Thankfully, more than mass attitudes are involved in democratic stability. Whether a non-democratic movement can threaten long-term political stability depends on the quality of the constitutional framework, the size of the non-democratic bloc, and whether it can be united behind a single candidate. If an anti-democratic movement can attain approximately 35-40 percent of the popular vote in a parliamentary system, it may be able to lead a coalition government without the 50 percent that might be required in a typical presidential system.

Around the world constitution writers and jurists remain concerned with ensuring the rights of individuals and groups to withstand periodic onslaughts from non-democratic forces that may attempt to ride a wave of populism into the seats of power. The American Founding Fathers designed a limited government, based on the rights of the individual, one that divided power among the executive, a bi-cameral legislature, the judiciary and the states of the Union. Today in countries where the non-democratic base is large (for instance, in Chile and Indonesia), democracy will remain fragile until the legislative and judicial branches have strengthened. Adequate performance and the passage of time should make it possible for these institutions to preserve the rule of democracy, even in situations where non-democratic forces may attain near majority proportions.

The cast of leaders generated by a political system at any given point in time may be the most important variable of all. The most effective democratic leaders stay close to popular opinion while appealing simultaneously to what Lincoln called the better angels of our nature. The worst democratic leaders are those marshaling a plurality (or even a majority) to take power by appealing to fear and identity politics, subsequently utilizing these forces to abridge the rights and privileges of groups they stigmatize.

A democratic political order can endure if the current generation of leaders remains politically attuned to the public and trustful enough of one another to alternate in power and to realize that the heart and soul of democracy is the diffusion of authority among different branches of government and throughout civil society—“A Republic, if you can keep it.”