Democracy is expensive. During the 2016 general election, candidates spent nearly $7 billion dollars on their campaigns. (More than $2.5 billion was spent just on the presidential contest.) This money paid for attack ads on television and direct marketing in the mail; it went to voter outreach in swing districts, fancy consultants and targeted ads on Facebook. The goal of all this spending was simple: to persuade more Americans to vote for them.

Did it work? Were those billions well spent? According to a new paper by Joshua Kalla of UC-Berkeley and David Broockman at Stanford, the overwhelming majority of campaign activity failed to persuade voters. As they bluntly state, “We argue that the best estimate of the effects of campaign contact and advertising on Americans’ candidates choices in general elections is zero.” Not close to zero. Not even one or two percent. Zero.

Kalla and Broockman come to this shocking conclusion by conducting the first meta-analysis of campaign outreach and advertising. Based on a review of forty field experiments, they found that the average effect of all these professional interventions was negligible. (Or, to be exact, - 0.02 percentage points.) While two of the forty studies did find a significant shift in voter behavior, Kalla and Broockman rightly note that these studies looked at interventions with limited applicability. (In one case, the candidate himself knocked on doors, while the other intervention relied on an onerous survey that most voters would never answer.)

However, Kalla and Broockman weren’t content to re-analyze the null results of the past. Given the “dearth of statistically precise studies” the political scientists decided to conduct nine of their own field experiments. They teamed up with Working America, the community organizing affiliate of the AFL-CIO, to study the impact of canvassing in a variety of different campaigns.

The good news, at least for the political industrial complex, is that Working America had an impact during primaries and special elections. Take the Democratic primary for the mayor of Philadelphia. Kalla and Broockman estimate that a Working America canvass conducted six weeks before election day boosted support for their endorsed candidate by approximately 11 percentage points. A similar effect was observed during a special election for a seat in the Washington State Legislature.

However, the effect size shrank to zero when Kalla and Broockman looked at attempts to influence voters during the general election. When Working America tried to persuade people in Ohio, North Carolina, Florida and Missouri to vote for their candidates for the U.S. Senate, Governor and President, the scientists consistently found no impact from the interventions. As they write, “we conclude that, on average, personal contact—such as door-to-door canvassing or phone calls—conducted within two months of a general election has no substantive effect on vote choice.”

This doesn’t mean campaigns are irrelevant. Candidates can still shape voters’ preferences by changing their policy positions and influencing the media narrative. However, Kalla and Broockman do present solid evidence that most of the stuff campaigns spend their billions on is essentially worthless, at least in the general.

Why are voters so hard to persuade? One likely cause is our hyper-partisan age, which has been exacerbated by online filter bubbles. (Republican Facebook is very different from Democratic Facebook.) As Kalla and Broockman write, “When it comes to providing voters with new arguments, frames, and information, by the time election day arrives, voters are likely to have already absorbed all the arguments and information they care to retain from the media and other sources.”

The key caveat in that sentence is “care to retain.” While voters are inundated with information about the election, they are depressingly good at ignoring dissonant facts, or those arguments that might rattle their partisan opinions. (Roughly half of Trump voters, for instance, think that he won the popular vote and that President Obama was born in Kenya.) The end result is that partisanship dominates persuasion; the vast majority of voters vote for their side, with little consideration of candidate or policy details. In a primary election, those partisan cues are less obvious, which means voters are more open-minded about the actual candidates. Persuasion stands a chance.

President Trump's success depends on these trends. His rhetoric and norm violations are consistently directed at a highly specific (and very conservative) slice of the electorate. This approach might be toxic for the body politic, but it does reflect a certain realism about the limits of persuasion. After all, if the other side can’t be reached, then moderation is for chumps. Modern politics isn’t the art of compromise – it’s the act of targeted arousal. (And Facebook makes such targeting extremely easy.)

President Trump's key insight was that all those norm violations would exact a minimal price at the ballot box. When it was time to vote in the general, he knew that partisanship would dominate, and that even those offended Republicans would hold their noses and vote for their guy.

Is there a solution? Not really. I am, however, slightly encouraged by recent research on human ignorance. In a classic study conducted on Yale undergraduates, the psychologists Leonid Rozenblit and Frank Keil asked people to rate how well they understood the objects they used every day, such as toilets, car speedometers and zippers. Then, the students were asked to write detailed descriptions of how these objects worked, before reassessing their understanding.

The quick exercise revealed that most people dramatically overestimate their understanding. We think we know how toilets work because we flush them several times a day, but almost nobody could explain the ingenious siphoning action used to purge the bowl. As Rozenblit and Keil write, “Most people feel they understand the world with far greater detail, coherence and depth than they really do.” They called this mistake the illusion of explanatory depth.

This same illusion is ruining our politics. In a 2013 study, a team of psychologists led by Philip Fernbach found that the illusion of explanatory depth led people to overestimate their understanding of political issues such as the flat tax, single-payer health care system and Iran sanctions. As with the toilet, it wasn’t until people tried to explain their knowledge, along with the impact of their chosen policies, that they realized how little they actually knew. Interestingly, acknowledging the unknown also led them to moderate their political opinions. As Steven Sloman and Philip Fernbach write in their book The Knowledge Illusion, “The attempt to explain caused their positions to converge.”

The practical lesson is that political persuasion isn’t just about slick videos and clever framing. In fact, most of that stuff doesn’t seem to work at all. (Charisma is no match for cognitive dissonance.) Rather, to the extent persuasion seems possible, it seems to be conditional on voters recognizing their own lack of knowledge, or at least grappling with the complexity of the issues. Sloman and Fernbach put it well: “A good leader must be able to help people realize their ignorance without making them feel stupid.”

There is a smidgen of hope in this research. If Trump represents the triumph of hyper-partisanship—he’s most interested in reaffirming the beliefs of his base—these findings suggest that candidates might also persuade voters by emphasizing the hard questions, and not just their partisan answers. At the very least, such rhetoric makes moderation more appealing.

This was an underappreciated part of the Obama playbook. While the former professor was often criticized for his long-winded and nuanced responses, that nuance might have been more persuasive to voters than another set of rehearsed talking points. As President Obama once observed, when asked about the challenges of the Presidency: “These are big, tough, complicated problems. Somebody noted to me that by the time something reaches my desk, that means it’s really hard. Because if it were easy, somebody else would have made the decision and somebody else would have solved it.”

Obama understood what the science reveals: If you want to change someone else’s mind, yelling out your answers won't work; facts are not convincing. Instead, try beginning with an admission of doubt. (Recent research by David Hagmann and George Loewenstein shows that "expressions of doubt and acknowledgment of opposing views increases persuasiveness," especially in the context of motivated reasoning.) We are most persuasive when we first admit we don't know everything.

Kalla, Joshua L., and David E. Broockman. "The Minimal Persuasive Effects of Campaign Contact in General Elections: Evidence from 49 Field Experiments." American Political Science Review (2017): 1-19.