Were Political Scientists Too Pessimistic About American Democracy?

Joe Biden soundly defeated President Trump in the 2020 presidential election. Although this outcome was predicted by most respected polls, as well as by poll aggregators like 538 and The Economist, the margins were closer than anticipated. And the outcome nevertheless came as something of a relief for the tens of millions of Americans who held their breath as the votes were counted last week.

Political scientists ought to be the most relieved. The past four years have seen a marked surge in popular writing about the state of American politics, often by scholars who do not work on American politics. Groups such as Bright Line Watch and the Authoritarian Warning Survey have conducted surveys and advocated for greater public awareness about the threats to American democracy posed by the Trump administration. The Democratic Erosion project has produced course materials for higher education on the problem of democratic backsliding around the world and at home. The new media have been saturated with commentary (mine included), by political scientists as well as by smart observers like Sarah Kendzior and Masha Gessen, about what is normal and what is not. The general tenor of this commentary was watch out: this is not normal, and this administration is a threat to democracy in America as we know it.

The pollsters and forecasters are devoting some effort right now to thinking about why their polls and forecasts were off. It stands to reason that political scientists who raised serious questions about the health and durability of American democracy might want to do the same. Do political scientists need to reckon with their understanding of American democracy?

tl;dr The answer is yes, we political scientists need to reckon with our understanding of American democracy. But we still do not have a good way to know how strong America’s political institutions are. This means that we cannot know if we were too pessimistic or not, but reasoning about why that is tells us something about how we ought to think about the question.

A Personal Perspective

My own writing over the past four years has been shaped by Trump’s election and by his presidency. I’ve written a number of posts on this blog about Trump and American democracy. Some of those posts ended up with bigger audiences than just this blog’s readers: this piece on how normal life under authoritarianism is showed up in Vox. Here’s a piece on Trump’s personal inclinations that was published in The Washington Post‘s Monkey Cage blog. Here’s one on regime cleavages in Politico. And here’s a piece in Perspectives on Politics–the American Political Science Association’s most public-facing academic journal–that I wrote with colleagues and friends from the American Democracy Collaborative.

There is no secret that I think that President Trump is an abject disaster as a president: needlessly divisive, absolutely corrupt, shockingly immoral, and plainly incompetent. But these conclusions are neither novel nor perceptive. They may in fact be views shared by many of his copartisans, and even by his supporters. These supporters may simply also like the fact that his administration is good for business, that he is appointed pro-life Supreme Court justices, and/or that he allows white identitarians to acknowledge their whiteness. America has had corrupt, immoral, racist, and incompetent presidents before.

The theme that underlines my writings on the Trump presidency is that is it not simply that Trump is a bad president. Rather–inspired by my knowledge of democratic backsliding and erosion around the world–it is that his actions as a candidate and as a president undermine popular beliefs that elections are the established practice for allocating political authority, and the partisan apparatus that supports him is making it worse.

My view of democracy is minimalist: it is nothing else than a nonviolent procedure for determining who gets to form a government that may legitimately pass laws. It has value even though that is all that it is (see Przeworski [PDF] for the best articulation of this point). Laws may be bad, elected politicians may be venal, voters may be stupid, but so long as voters agree that elections are how we resolve intractable political disputes, then winners can win and losers can lose.

An example may serve to illustrate my point. Although I opposed the Patriot Act in the early 2000s, it did not strike me as an essential threat to electoral democracy. It was a threat to other things: liberty, due process, and so forth. In writing about Trump, my overwhelming concern has not been on his racism, misogyny, corruption, or incompetence. Problems as they are, my concern has been with how Trump’s rhetoric, his actions, and the behavior of those who support him undermine electoral democracy in the United States.

I am only one voice among many political scientists who sounded alarms about Trump’s presidency. I don’t represent them, nor do I speak for them. But I suspect that the reasons why so many political scientists have been so vocally and visibly disturbed by the Trump administration is because they view his presidency as qualitatively different than other administrations that they opposed (especially Reagan and Bush II, neither of whom were popular among political scientists).* Trump’s behavior calls into question the belief that American political institutions are strong enough to withstand a challenge to electoral democracy.

Reckoning with Institutions

So that is where I want to focus: on our understanding of American political institutions. Among political scientists, we can compare those who study American politics (“Americanists”) and those who study primarily the political systems of other countries (“comparativists”). I think it is safe to say that by and large, Americanists and comparativists had different baseline orientations to the Trump presidency:

  1. Americanists have a deeper understanding of not just America’s constitution and system of laws, but also how the bureaucracy functions, how informal norms structure Congressional proceedings, what executive actions are customary, and how things get done. Buried within all of these–but built into their very structure–is a profound status quo bias. It is hard to steer the ship of state. It is also hard for any group, even a well-resourced and dedicated one with four years to do its work, to penetrate all of the layers of America’s complex political system in order to make truly fundamental changes. Because federalism and the administrative state.
  2. Comparativists have a deeper understanding of what is possible from the experiences of other countries. Actions, decisions, and events that are totally abnormal in the context of American politics itself often prove to be quite recognizable from a comparative perspective. The field site where one explores the implications of a president consistently alleging that elections are being stolen is not the United States, it is Hungary, Bolivia, or Thailand. If one wanted to look for the antecedents of democratic breakdown, one must necessarily look elsewhere: to Weimar Germany, the collapse of the French Fourth Republic, the erosion of Indonesian democracy in the 1950s, or the failure of open politics in Russia in the 1990s.

The challenge is how to apply the comparative experiences to the American case. We have good comparative experiences from which to learn, but America is unique.** My view has always been that we have a good mass of evidence from the comparative experience about how political institutions such as America’s work (for example, I wrote about presidentialism just a couple months back). I am not satisfied with the proposition that America’s uniqueness contains within it some kind of anti-authoritarian quintessence.

In broad strokes, one main lesson from the comparative experience of democratic decline is that it is often subtle, gradual, and hard to notice. People want to keep watch for the next Reichstag fire, but as I wrote in Vox in 2017:

Most Americans conceptualize a hypothetical end of American democracy in apocalyptic terms. But actually, you usually learn that you are no longer living in a democracy not because The Government Is Taking Away Your Rights, or passing laws that you oppose, or because there is a coup or a quisling.*** You know that you are no longer living in a democracy because the elections in which you are participating no longer can yield political change.

The gradualness of democratic decay is also a theme in Levitsky and Ziblatt’s How Democracies Die, which is perhaps the most well-known statement by academic political scientists of the predicaments of our time. Many Americanists came to agree with the comparativists’ perspective over the course of the Trump administration. Those who considered Trump to be “just a bad president” or “an odious politician who’s dividing America” or even “responsible for terrible laws that will set America back decades” but not “making decisions that threaten the very foundations of American democracy” tended to be pretty quiet (there is not a market, after all, for “this is bad, but normal”).

One possibility is that given the challenges of knowing how durable American democracy is, but confronted with what seems to be a novel and highly unpleasant form of politics that has obvious parallels with the authoritarian styles of Putin and Erdogan, political scientists worked themselves into a tizzy. Blog posts like mine, tweets about what is abnormal, and other such commentary may have functioned like ghost stories. Or even like a kind of “fash-porn.”****

And note how insidious this process might have been. If the doom-sayers are able to draw on comparative experiences because there are no exact American parallels upon which to draw, it becomes impossible to rebut the comparativists’ views by referring to the particularities of the American experience.

America’s Democracy Test

In the event, however, the Democratic Party nominated an old white lifelong politician and a multiracial coastal elite who also happened to be woman, and proceeded to defeat Trump (a populist plutocratic demagogue) and Pence (a stuffed-shirt Christian conservative straight out of central casting) handily at the ballot box. My worst fears–that there would be outbreaks of violence on Election Day, or a concerted Russian or Iranian attempt to hobble electronic voting systems on Election Day, both of which I predicted in semi-public settings–did not come to pass. Instead, America’s elections went off just fine, controlling for mass pandemic.

Everyone should pause just a moment to reflect on that fact. Carrying off an election in America’s decentralized federal system is an amazing achievement. Most people devote no thought at all to how it works, and that itself is amazing. In the most polarized election of all time, in a pandemic, with a plainly authoritarian president supported by a party willing to break all manner of governing norms in order to get its way, something like 150 million Americans voted, their votes were counted, and the president was defeated.

Let me put on my comparativist hat and observe the following: if you think that this is not a big deal, you’re wrong. I welcome the fight for equal access, against the rampant inequalities that Americans face when they wait in line for hours to vote or are unjustly forced to certify their identity or are portrayed as not counting because they make the mistake of living in a city rather than out in the country. But solving these problems is nothing compared to convincing an entire country to hold elections, putting the infrastructure together to do so, and then doing it.

So then. Knowing the outcome, how do we reason about the Trumpist threat to American democracy, and the strength of American institutions?

It seems plain to me that political scientists need to update their prior beliefs either about the strength of American institutions or about the magnitude of the Trump administration’s threat to American democracy, or both. I have no answer as to how to update these two beliefs. But we need to think seriously about how the structure of American politics makes America’s version of presidentialism special, how federalism makes a top-down takeover of institutions difficult, and other factors that might mean that comparative lessons do not travel so easily to the American context.

There is another option, of course. It could be that institutions were exactly as vulnerable and Trump’s presidency was exactly as threatening as feared, but that this provoked exactly the response needed to defend institutions and defeat Trump. Perhaps this is true. But it is also the most self-serving possible interpretation of the past four years that I can imagine. I’m embarrassed even to write it.

Where We Are After the 2020 Presidential Election

An entirely orderly election under very difficult circumstances notwithstanding, the following things are all true right now:

  1. The President has not conceded the election to President-Elect Biden.
  2. GSA administrator Emily Murphy is refusing to allow the Biden transition team access to resources and information needed to begin the transition process.
  3. The president fired SecDef Mark Esper, which is abnormal in the lame-duck period when it comes to inner cabinet secretaries. Since I wrote that sentence, in fact, all top civilian defense officials have been fired or resigned.
  4. Lindsey Graham wants a joint committee in the Senate to analyze mail-in ballots.
  5. Georgia’s two Senators want their state’s Republican Secretary of State to resign because they oppose the electoral outcome that Biden is slightly ahead of Trump (they view this as being about “honest and transparent elections”).
  6. Pompeo, Pence, and McCarthy are in race to see who can debase himself most thoroughly in insisting that Trump will remain the president.
  7. The Trump Team plans to conduct mass rallies against alleged voter fraud, keeping the issue alive indefinitely.
  8. We can see the emergence of a GOP mythology–what some might term a mass psychosis–that Trump didn’t really lose. This will force the GOP to rehearse Trump’s “fake loss” for years.

I could go on, but there’s too much to keep track of. I will say, however, that I was not prepared for how hilarious the Trump Team’s efforts to avoid defeat have been.

How should we think about this? On one hand, the Trumpists’ pleading and denialism is pathetic. On the other hand, and there are millions of people who now believe that the election has has somehow been stolen. They were primed to believe it by months of preemptive Trump whining and the far-right disinformation regime, so this was utterly predictable. But I don’t think that this is all performative self-abnegation.

But so what? I think it is possible to heed the cautionary message about telling ghost stories that I shared above while still identifying real concerns. Having seen the conduct of the election and the strength of the opposition to the president, I suspect that an orderly resolution to the Trump-Biden transition is the most likely outcome. I emphasize that institutions are not self-executing: individuals must act as if the law still binds. But I am more optimistic that they will than I was in late October. The role of expectations here is key. The more that ordinary Americans hear that Biden won the election, and the more that the regular arms of government outside of Trump’s direct authority simply perform their roles, the more difficult it will be to prevent an orderly transition.

To gain real insight, we need more analysis from experts in the mechanics of American politics. I can create analogies all day to the final days of liberal democracy in Sukarno’s Indonesia, but at this particular moment they probably create more heat than light. One lesson that does travel, though, is about the essential role of uncertainty in this process: the fundamental uncertainty in political transitions applies just as much to the American case as to any other.

But the nonsense coming out of the Trump Bunker is still dangerous to the long-term health of American democracy. There is a dangerous dualism emerging. Most Americans of both parties think that Biden won the election. But most Republicans believe the election was unfair. We have already crossed the Rubicon, in the very specific sense that Trump will not go willingly, he must be removed. But of course, we’ve known that all along.

My long-term concern is what would you do if you thought that elections were being stolen, or that you had no chance of winning them? The Trump administration, enabled by the GOP as an institution, is systematically cultivating the belief that the conduct of American elections are not legitimate. And, moreover, that the electoral bias is an essentially partisan one that divides real Americans represented by the GOP from elites, non-Americans, and bad Americans. What would you do if you thought that?

Take a beat before answering. On January 21, 2021, this will either be you, or one of your political rivals.


* Take note here: I am not implying that Trump is sui generis. In fact, I think that the politics that Trump exploits is just the fulfillment of long-term developments in American politics. Some of these are structural: the constitution, the legacy of slavery. Some of these are choices made by GOP elites, like Newt Gingrich’s style of politics. Trump is not an accident or an aberration, but he is not the same as what has come before.

** This is a trivially true statement. It is not profound. The same challenge would exist if we tried to, say, apply the lessons of the Jones Act to Uruguay.

*** The fine Norwegian drama Occupied is actually fantastic on this point. It tells the story of a hypothetical Russian occupation of Norway. All of the politicians are complicated. Even in the country that innovated the Quisling, it’s hard to tell who is the quisling.

**** I did not invent this term, but I don’t think it’s been used quite this way before (*ahem*).

Americans Are Already Set to Conclude that the 2020 Presidential Elections Were Rigged

With less than a week to go until the 2020 presidential election, Americans are ready for a seemingly interminable election season to come to an end. But one thing seems increasingly certain: Election Night won’t be the end of it. From the long wait times anticipated as swing states like Pennsylvania count absentee ballots, to the president’s continued refusal to acknowledge that he will abide by the outcome, you can bet that when you wake up on November 4 the election won’t yet be over.

This is worrying. Americans are used to believing that election night produces decisive results, because for all presidential elections except for 2000 it has. But a delayed announcement of the winner will be more than an annoyance, it will undermine faith in the results of the elections. In a recent Supreme Court ruling, Justice Kavanaugh wrote that it might be legitimate to not count votes “to avoid the chaos and suspicions of impropriety that can ensue if thousands of absentee ballots flow in after Election Day and potentially flip the results of an election.” Of course, Justice Kagan’s dissent (PDF) that “there are no results to ‘flip’ until all valid votes are counted” reveals exactly the issue: stopping the vote count to avoid suspicions of impropriety will generate the very suspicions of impropriety that Kavanaugh seems to want to avoid.* But no matter. The president himself wants a final result on November 3, because he believes that the more votes are counted, the worse off he will be.

Whatever the particular of how this election is resolved, though, will American have faith in the result, even if their preferred candidate has lost? Or will they conclude from the election that the election was fraudulent, unfair, illegitimate, or somehow rigged?

Some new survey data can shed light on these questions. Shana Gadarian, Sara Goodman, and I have been working for the past six months on a NSF- and Cornell Center for the Social Sciences-supported project on the politics of the COVID-19 pandemic. And the other day we received our most recent round of survey data, fielded between October 15 and October 21, which contains an important piece of information that we can use to anticipate how Americans will interpret the election results.

As part of the survey,** we asked respondents a simple question.

To what extent do you agree or disagree with the following statement: If Trump/Biden wins the election, we will know American elections are rigged.

The words “Trump” and “Biden” were randomized in the survey, so half of our 1871 respondents were asked to evaluate a Trump victory, and other half to evaluate Biden victory. Randomization combined with a large sample sizes means that we can compare average responses to the two versions of the question to get a good sense of what Americans will believe in these two hypothetical states of the world—one of which will become actual in the coming weeks.

Now, this is a stark question. And it is easy to give an expressive answer: yeah sure, if my guy loses of course this whole thing is rigged! It is pretty costless to give extreme answers like these, and this means that perhaps people are just sharing their emotional reactions rather than their actual beliefs or values. Moreover, rigged is ambiguous. Does this mean illegally manipulated? Or just preordained? Unfair in some vague sense? Something else entirely?

But respondents had five possible answers that they could give, and even if they felt some misgivings they did not have to respond “Strongly Agree” if their candidate were not declared the winner. I, for example, would answer “Neither Agree nor Disagree” to this question if my non-preferred candidate were to win. And even if respondents’ answers are entirely expressive and emotionally driven, we know for certain that the airwaves and social media will be plastered with exactly those sorts of expressive claims from elites associated with the losing candidate. So these results might give us a hint as to just how much latent or mobilizable sentiment might be out there in the American mass public.***

Here is a graphic that summarizes our main results by comparing Trump wins versus Biden wins versions of the question by the respondent’s partisan affiliation (Democrat, Republican, or something else).

These results are, in a word, extreme. To wit, the modal Republican respondent strongly agrees with the view that if Biden wins, we will know that the elections were rigged. And the modal Democrat respondent strongly agrees with the view that if Trump wins, we will know that the elections were rigged. Those who aren’t affiliated with either party (either because they are nonpartisan, or because they identify with a third party) are most likely to strongly disagree that elections were rigged. But even among those respondents, a surprisingly large number would agree or strongly agree that elections were rigged no matter who wins.

How much do these differences reflect to partisanship itself, versus other things like demography or context? To see this, we estimated some OLS regressions where we interacted partisanship with our randomized Biden/Trump variable, and included a large set of covariates, flexibly adjusting for nonlinear associations by urban/rural, education, income, gender, race, and state fixed effects. This is a pretty conservative approach. Here is what we find when we use this model to predict what respondents’ levels of agreement are.

So… these are massive differences in responses among the partisans, and they cannot be explained by the demographic and geographic covariates that we have available to us.****

But it is also interesting to ask about those other demographic differences as well. To do this, we estimate four more OLS regression models. In each, we add to the two-way interaction of party ID and Biden/Trump a third interaction: by age categories, by race, by gender, and by education level. Below, we show differences across these demographic categories (averaging across and hence controlling for the differences in partisan composition across these groups).

We can see here that—after accounting for partisan differences—there are few differences in the perceived legitimacy of a Biden versus Trump election across these demographic categories. Two exceptions stand out: women would be more likely to find a Trump election illegitimate than a Biden election, and Black Americans are far more likely to find a Trump election illegitimate than a Biden election. That last result is particularly striking, highlighting the essentially racialized nature of this presidential election and its implications for electoral legitimacy in the United States.

There is more to dig into in these data. Got another idea of a covariate to explore? Put it on the list! But these data suggest that we can expect some pretty polarized reactions to the election next week. Partisan competition may be producing a fundamental political divide over the institutions and legitimacy of democracy itself.

I hasten to emphasize once again that these could just be expressive views, emotional reactions that don’t reflect behavior or broader political orientations. But if savvy politicians know how to exploit such emotional reactions, we could be in for a dangerous ride. Do we live in a world in which politicians will exploit such emotional reactions to news and events that their supports don’t like? It’s a waiting game right now.


* Look at the big brain on Brett! Also, laughably bad with his sources. Whelp. It is what it is.

** The survey is nationally representative. We can demonstrate this statistically.

*** If the president is on Twitter objecting to the results of the election, you can bet that he will use the word rigged.

**** Even with all of the state fixed effects and other covariates, the t-statistic on the Biden × Democrat coefficient is 20. Twenty.