My pride (and my editor) forbid me from starting this story with a knock-knock joke,1 but it does begin with knocking. Door-to-door political canvassing made headlines in April — including here at FiveThirtyEight — when scientists at Stanford and the University of California, Berkeley, published a paper claiming they could use the technique to increase support for transgender rights. This research, conducted by David Broockman and Joshua Kalla, studied canvassers from the Los Angeles LGBT Center, and their finding wasn’t just buzzy, it was scientifically significant. Scientists who study political persuasion told me that they’re going to be using the paper’s research techniques and testing its ideas for years to come. It represents the new and exciting frontier in a fast-changing field.
It’s also representative of how politically lopsided that field can be. Whether or not you think increasing the public’s acceptance of transgender people is a laudable goal, it’s definitely a liberal one. In that way, Broockman and Kalla’s finding is part of a bigger trend in the science of political persuasion — when the study of persuasion is colored by ideology, that color is almost always blue.
The scientists I spoke with said that most of what they know comes from one of two places:
- Politically generic studies that are aimed at increasing voter participation, regardless of party. This constitutes the vast majority of research.
- Studies where the organizations and messages being tested are decidedly left-leaning. “I think it’s right to say that the sheer number of randomized trials on the right is far smaller than the comparable number on the left,” said Don Green, a Columbia University political scientist who specializes in the study of persuasion. When scientists come knocking, the answer to the question “Who’s there?” is very likely to be “A representative of a liberal cause.”
Conservative campaigns and campaign advisers also do a lot of their own research, as well, but this work isn’t usually peer-reviewed and published, nor is it typically done in collaboration with academics. Partly, this is the result of how the scientists’ funding is structured. But it’s also because of a frequently overlooked fact: Scientists are people, too.
“We aren’t allowed to do partisan work with our research money from federal sources,” said Diana Mutz, director of the Institute for the Study of Citizens and Politics at the University of Pennsylvania. Even then, there are some kinds of activities that researchers can’t engage in, because of IRS rules that govern the behavior of non-profits, including charitable foundations and universities.”
Scientists say that this is where their own biases come into play. The people who study the psychology of political persuasion tend to be liberal, said Green, Mutz and Daron Shaw, a professor of government at the University of Texas at Austin. The people who earn doctorates in persuasion science and then leave academia to work for campaigns follow the same pattern. The researchers said they don’t think anyone is intentionally excluding conservative campaigns — something Dave Carney, an independent conservative campaign consultant, agreed with. But their own interests and personal connections more easily lead them to questions and collaborators on the liberal side.
Think of it as a problem of supply and demand, said Alexander Coppock, assistant professor of political science at Yale. There just aren’t that many people in academia actively seeking out collaborations with the right, and there aren’t many people on the right actively seeking out collaborations with academia.
Sure, conservative campaigns are doing tons of their own research. But that kind of data analytics is really aimed at understanding what will help an individual candidate get elected right now, Shaw said. It’s not necessarily applicable to other candidates or campaigns. Academic research is focused on bigger questions about the human mind and how it can be changed, he told me. There are some big, elephant-shaped holes in this public body of knowledge.
Collaborations between conservative campaigns and academia do happen, however, and they’re valuable to both the campaigns and the scientists.
In 2014, Shaw and Carney collaborated on a study of the effects of different kinds of advertisements in the Texas gubernatorial Republican primary. They found that ads — particularly ones on the radio and broadcast TV — were able to significantly raise candidate Greg Abbott’s favorability among likely voters but didn’t do much to turn “likely” voters into actual ones. The campaign found out which of its buys were most effective. The scientists added another hash mark to a growing tally of research that suggests a lot of the things campaigns do to get a candidate elected aren’t doing anything to get the candidate elected.
Green pointed to another recent study, presented at the 2015 Annual Conference of the Midwest Political Science Association, that found increasing quantities of targeted, direct mail did almost nothing to swing votes in North Carolina legislative elections — or to change voters’ minds. That study was a collaboration with the generally conservative-leaning North Carolina Chamber of Commerce. “Finding no mobilizing effect of partisan mail isn’t news,” Green said. “But the relatively limited persuasion effect is a corrective to the notion that you can just throw money at it.”
Cold cash and colder calling don’t seem to have much impact on the outcomes of elections, for either the left or the right, Green said. In general, the most interesting thing scientists have learned from studying persuasion across the political spectrum is that there don’t seem to be any big differences in how persuasion works at either end.
That matters to those who want to use the Broockman-Kalla study as the basis of their own work. Dave Fleischer, project director of the L.A. LGBT Center’s Leadership LAB, said he thinks the technique he helped develop — an intensive type of door-to-door canvassing where volunteers tried to elicit empathy in subjects by helping them relate their personal experiences to those of transgender people — would be difficult for conservative campaigns to use. “We’ve noticed that we’re uncovering a disjunction between voter’s opinions, that are often more conservative, and their actual lived experience, that is often more liberal. They become less prejudiced because they revise their opinions in light of real, lived experience,” he said. “I won’t go so far as to say that there’s a liberal trying to claw its way out of every conservative voter. But I will say that it’s closer to truth than the idea that there’s a lot of potential to use this on the conservative side.”
But persuasion research suggests he might be wrong. Carney, the conservative campaign consultant, can already imagine how the technique could be used by the right for canvassing related to the abortion debate. Instead of building empathy for transgender people, the technique could help voters emotionally connect to women who regret their abortions, or people who survived abortion attempts.
Although scientists are finding that persuasion doesn’t differ much from left to right, it’s still important to field-test hypotheses across the political spectrum, Coppock said. In fact, broad-spectrum testing is important precisely because it has shown so little difference.
For his dissertation this year, Coppock published the results of 20 survey experiments that attempted to change people’s opinions on issues such as capital punishment, the minimum wage and climate change. The studies were structured so that subjects were split into groups, and each group received different perspectives on the issues — some got an argument in favor of capital punishment, for example, others got one against it. He found that, while the persuasive effects were small — on a scale of 1 to 10, he was shifting people’s opinions by maybe half a point — they persisted through his 10-day follow-up.
What’s more, everyone was affected, no matter their previous beliefs and political leanings. Republicans could be shifted to slightly increase their belief in anthropogenic climate change. Democrats could be shifted to slightly decrease their support for raising the minimum wage. Across the board, people exposed to logical arguments were persuaded by them.
“Many people believe effects will be conditional on the kind of person you are, i.e., Republicans will behave differently from Democrats, because they see Republicans and Democrats behaving so differently from one another,” Coppock said. “It turns out that they don’t behave so differently in response to information.”2
CORRECTION (July 1, 4:37 p.m.): An earlier version of this article misstated the limitations of federally-funded research into partisan issues. That research (unlike electoral or ballot issues) can be funded by federal sources; it is not barred from receiving federal funds. The Broockman-Kalla paper was funded by an independent charity because the authors chose to fund it that way, not because of federal limitations.