$6.4 billion. That’s how much candidates, political parties, and interest groups spent on federal elections in 2016, according to the Open Secrets project at the Center for Responsive Politics. Especially in competitive races, huge amounts of money are invested in reaching voters through ads, phone banks, direct mail, and canvassing. Ostensibly, the goal is to persuade people to vote for a particular candidate.
A new paper by two California political scientists finds that the total effect of these efforts is zero, meaning that they have no impact on how voters vote. David Broockman, a Stanford University assistant professor, and Joshua Kalla, a doctoral student at the University of California, Berkeley, analyzed data from 49 field experiments—state, local, and federal campaigns that let political scientists access their data to evaluate their methods. For every flyer stuck in a mailbox, every door knocked by an earnest volunteer, and every candidate message left on an answering machine, there was no measurable change in voting outcomes. Even early outreach efforts, which are somewhat more successful at persuading voters, tend to fade from memory by Election Day. Broockman and Kalla also estimated that the effect of television and online ads is zero, although only a small portion of their data speaks directly to that point.
For obvious reasons, this kind of data is relatively rare, according to Broockman and Kalla: Few campaigns want to give outsiders carte blanche access to their operations, particularly when those outsiders might find their methods totally ineffective. But in many ways, it’s the most important kind of data available. It shows how actual voters and campaigns behave, rather than relying on theoretical surveys. The findings suggest that a lot of the time, energy, and money poured into traditional campaigning methods is wasted, and that the campaign operatives hawking tried-and-true tactics don’t have the evidence to back up their claims. It also casts doubt on the theory of the swing voter who can be persuaded with enough flyers, ad exposure, and conversations with earnest volunteers. In reality, Broockman and Kalla find, direct outreach is most effective at improving voter turnout, suggesting that campaigns should focus on getting their core supporters to the polls than reaching out to a mythical middle.
Before they published this paper, Broockman and Kalla ran into minor social-science fame when they debunked a major study showing that voters can be persuaded to change their views on same-sex marriage if they meet a gay or lesbian person. A short while later, the pair published their own paper on a campaign to change minds about transgender people, finding that persuasion is, in fact, possible. This new study suggests that intentionally curated, issue-specific persuasion campaigns may shift people’s views more easily than partisan political campaigns.
I spoke with Broockman and Kalla about their findings and the abject failures of most political campaigns. Our conversation has been edited for length and clarity. For more on this topic, read Molly Ball’s 2016 piece about the impotence of political consulting.
Emma Green: So, is political campaigning useless?
Joshua Kalla: The short answer is ‘no.’ There are lots of things that campaigning can accomplish. Two decades of research on voter registration and hundreds of field experiments show really cost-effective ways to increase turnout in the base.
But on persuasion, yes, we find that on average, there are very small effects.
David Broockman: Certainly, for those who hope to win elections or change each others’ minds, the conclusion should prompt a rethinking of the way to change minds.
Green: As you point out, a ton of money is spent every election cycle on television ads, directing-mail campaigns, canvassing operations, etc., which are all designed to persuade voters. How should we grapple with the fact that all this spending is largely ineffective?
Kalla: A lot of campaign operatives think there’s this big pool of moderate, undecided voters that we can spend money on to persuade them to our side. That strategy is probably not the right strategy. And we should be skeptical of big claims of persuasion.
“In God we trust, and everyone else bring data.”
Green: Is all that money that goes into campaigns wasted?
Kalla: All the money is being poured into the same time and the same place. It’s hard to imagine that the hundredth TV ad that a person views is really worth it from a monetary perspective, versus that same money spent in a different race or a lower race. There’s a case to be made that too much money is being spent in the same ways and on the same people.
But the takeaway from this paper should not be that campaigns should stop. Campaigns do a lot of work that is measurable in return on getting voters to vote, and persuading voters. It’s just a question of how the money is spent.
Green: I’m curious what your findings say about the “horserace” politics television and print media love to hype. If early persuasion techniques don’t work, and on-the-ground canvassing and campaigning doesn’t work, how meaningful are the “horserace” speculations?
Kalla: The first order of understanding an election and how people vote is partisan identity. Most people vote based on whether there’s a D or an R next to their name. Unpacking that should be more the focus than the horserace.
Green: Your last major study showed the potential for success in personal, transgender-related persuasion campaigns. How do your new findings fit with that study?
Kalla: Ballot-measure campaigns—which are similar to that issue-persuasion work—are very persuasive. One commonality in the ballot measures and the transgender work is the lack of partisanship. We don’t see persuasive effects in general elections where a Democrat is talking to a Republican. But in ballot-measure campaigns and primaries and the transgender work, it seems that persuasion is possible.
Green: Is it just a category difference—that people can connect on issues, but not about some guy who’s running for office?
Kalla: My best guess on the difference is partisanship, where there’s a clear party cue as to how you’re supposed to vote. Most Americans view themselves in a partisan lens. When it comes time to vote, it’s less a function of a person running for office than a person with a party label beside his or her name.
Green: But what about the roughly 39 percent of Americans who identify as independents?
Kalla: A lot of independents tend to be what political scientists term as “closeted partisans.” They might not explicitly identify with a party, but if you ask them which party they lean toward, they’ll often give you an answer. Their behavior tends to look a lot like the behavior of people who explicitly identify as partisans. And explicitly within our experiments, we look to see whether people who identify as independents end up being more persuasive than partisans. And we find there is no difference.
“Humans are very good at fooling themselves, especially when there are consulting fees involved.”
Green: Our democracy is based on this romantic idea that encounters in the public square—conversations, essays, speeches, etc.— have the power to change how people view the world. If you’re saying that’s basically not true, where does that leave us? Are we all just destined to remain isolated in the prisons of our own convictions?
Kalla: I want to draw a distinction with the transgender canvassing work. That was very much focusing on getting people to be introspective and think about times that they or their loved ones have been discriminated against, and how that made them feel, and how that real, lived experience informs their views on non-discrimination laws and views toward LGBT people. That’s close to an ideal of how we want democracy to function.
That’s not the type of discourse you see in campaigns. I don’t think TV ads or every glossy postcard is really going to lead to enlightened discourse among the American public.
Green: Is the other lesson here that all campaign strategists are hacks?
Kalla: There’s a great line: In God we trust, and everyone else bring data. One thing our paper says is, hey, these claims that campaign strategists and polling gurus make are testable. Unless someone is going to actually test their claim and do it in an unbiased way, and share the results whether or not it makes them look good or bad, I think our starting assumption should be that most of the claims that are out there should be treated skeptically.
Broockman: If you think about medicine in the 18th and 19th century, there were a lot of people who weren’t dumber than we are now. They were the same people; the just didn’t have the same data. A lot of honorable people believed certain things about leeches and blood-letting and whatever else. And now we know that’s all garbage.
That’s to say that just because there are certain beliefs that many people hold about what works in campaigns doesn’t mean those things are likely to be true. Humans are very good at fooling themselves, especially when there are consulting fees involved. We need to move to a higher standard of evidence. If I were a campaign candidate or donor—just like if I were a patron in the 19th century—that’s what I would demand.
That revolution that happened in medicine is what we’re starting to see in campaigns. And if medicine is any guide, there’s no reason to think that the anecdotes of highly paid, highly respected people are going to be accurate.