When It Comes to Impeachment, False Negatives are More Dangerous for Democracy than False Positives
So, here we are. Though observers have long expected something like this result, it looks to be a matter of hours or at most a handful of days before the US Senate votes both to deny a call for further witnesses in President Trump's impeachment case and to acquit him.
This outcome is what scholars of Latin American politics would expect, too. Latin America has seen a lot of presidential impeachments, enough to begin to draw some general conclusions. Among them: all impeachments are political; the outcome depends more on the strength of the president's legislative coalition than the evidence; and a bad economy and massive protests can tip the scales against the president. Given those findings, all lights look green for Trump. Research on Latin American politics indicates the results in the US are pretty much foreordained.
But does this mean that Trump's acquittal will just be politics as usual? Messy, unsavory, and factional, but ultimately democratic? To put it another way, can the outcome of an impeachment trial actually threaten democracy? If so, when?
A view from statistics and democratic theory provides reason for anxiety about the US right now.
Let's start by discussing why presidential democracies need a mechanism for impeachment. The notion of “rule by the people” is a fiction in the post-1776 world -- a beloved and important one, a founding myth. Absent a system where “the people” can actually themselves somehow direct national policy, thoughtful people like the US Founding Fathers have invented institutions that attempt a one-off approximation. So, what we call “democracy” today entails free and fair elections with universal suffrage, but it also requires accountability between elections to check abuses and to prevent consolidation of power that would subvert future free and fair elections. In presidential systems, we achieve that accountability between elections via a complex system of inter-branch checks and balances -- legislators have the legal right to investigate members of the executive, and the ultimate penalty is impeachment. Parliamentary systems provide that accountability in other ways: the executive (i.e., Prime Minister and Cabinet) is constantly answerable to Parliament, and the executive can go at any time. So, impeachment is a solution to the problem of inter-election accountability that is distinctive to presidentialism.
But, in the real world, we know impeachment is political. Rules never get implemented perfectly. Party politics always plays a role. Sometimes presidents who are unpopular but haven’t really violated serious and important rules get impeached on technicalities -- arguably the case for Brazil’s Dilma Rousseff. And other times, presidents who are engaged in objectively undemocratic behavior get off the hook -- arguably the case with the Venezuelan legislature’s 2016 attempt to impeach Nicolas Maduro.
Are some forms of politicized impeachment worse than others? Or can democracy survive all sorts of unsavory impeachment processes?
Hold tight -- we’re going to get nerdy for a minute. If you’ve taken a class in introductory statistics, you might recall an arcane discussion of Type I and Type II errors. These concepts can help us understand the risks of different kinds of politicized impeachment processes. To take an example that popped into my head out of thin air, let’s say our stats prof writes out on the blackboard a “null hypothesis” and an “alternative hypothesis”:
The null hypothesis is: “Donald Trump did nothing wrong.”
The alternative hypothesis, then, must be that “Donald Trump did something wrong.”
The goal of hypothesis testing is to determine whether to accept or dismiss the null hypothesis, based on evidence. So, our stats prof explains, we are at risk of committing two different kinds of errors:
A “Type I” error involves incorrectly dismissing the null hypothesis -- concluding that DT did something wrong, even though he actually didn’t -- that is, a false positive.
A “Type II” error involves incorrectly accepting the null hypothesis -- concluding that DT did nothing wrong, even though he actually did -- that is, a false negative.
Here’s the thing: both Type I and Type II impeachment errors are problems -- but Type II impeachment errors pose an immediate threat to democracy.
The 2016 impeachment of Dilma Rousseff arguably constitutes a Type I error. The root cause of her impeachment was probably the recession and her inability hold her legislative coalition together. However, she was actually impeached on relatively minor charges of fiscal juggling to maintain a budget that looked balanced on paper. Assuming such budgetary improprieties wouldn’t normally lead to impeachment, we could say that hers was a “false positive” impeachment. But, despite her party’s indignant claims they were victims of a “coup,” her impeachment did not really threaten to derail future elections. Brazilian democracy has problems -- I’ll come back around to that in a minute -- but its future is not immediately, existentially threatened.
By contrast, the Venezuelan legislature’s inability to remove Nicolas Maduro in 2016 via a recall referendum arguably constitutes a Type II error. Maduro’s behavior likely merited impeachment, but the constitution and executive machinations stymied the effort.
Removing a president whose behavior probably didn’t merit impeachment has democratic risks. The Academy Award-nominated documentary The Edge of Democracy eloquently expresses the disillusionment of Dilma Rousseff’s supporters. People who think their vote doesn’t matter might support coups or violence, or they might just withdraw from political participation. Type I error impeachments could also help anti-democratic elites consolidate power -- a risk that has not yet materialized in Brazil (that’s a long discussion for another day).
But a Type II error, a failed impeachment process, poses bigger risks. Legislative failure to hold an executive to account might signal that the executive has already consolidated power beyond normal democratic bounds. More importantly, a Type II error signals what the executive can expect going forwards. Absent normal inter-election checks and balances, we can only trust the executive’s own self-restraint to avoid inter-election rules violations and consolidation of power that inhibit future free and fair elections. And, um,“self-restraint” and “Trump”...I won’t finish that sentence.
In this regard, the US Senate’s likely refusal to call witnesses stacks the deck in favor of a Type II error. Failure to remove a president is nothing new in the context of Latin American politics. But what’s worrisome for democracy is that today’s vote may signal to the executive that, going forward, he can avoid not only removal, but even the sunlight of real legislative scrutiny.