In 2017, then-President Donald Trump said he believed that the U.S. must “fight fire with fire” in response to terrorism and that he considered torture an effective means of interrogation.
Some Republicans may have shared his view; others may have been swayed by his words. But to what degree?
When MIT researchers Ben Tappin, Adam Berinsky, and started to investigate how Americans incorporated cues from party leaders into their views on policy issues, they expected respondents to fall in line behind their party leader of choice: in this case, Trump or Joe Biden.
The research team expected that party loyalists would agree with Trump’s stance on torture or Biden’s view on affirmative action, as well as with party-specific opinions on any one of dozens of hot-button issues.
In new research published in the journal Nature Human Behaviour, the researchers found that Democrats and Republicans reacted to persuasive messaging in ways that didn’t always align with their party leader’s stated positions.
This work challenges the view that party loyalty distorts how Americans process evidence and arguments. “Our results are clear and unequivocal: Learning the in-party leader’s position on an issue certainly did influence partisans’ attitudes — but it did not cause the partisans to ignore or discount arguments and evidence that ran counter to the leader’s position,” said Rand, professor of management science and brain and cognitive sciences at MIT Sloan.
Rather, respondents seemed to integrate leader cues and countervailing persuasive messages as independent pieces of information in determining whether they agreed or disagreed with various policy issues, he said.
Following party cues
The study, conducted in September 2021, involved more than 5,000 American partisans — Republicans who voted for Donald Trump in 2020 and Democrats who voted for Joe Biden in 2020.
Respondents completed a survey spanning 24 policy issues, including affirmative action, the death penalty, the estate tax, foreign aid, tariffs on Chinese imports, and restrictions at the U.S. border.
In the study, respondents gave their opinions about several of the policy issues on a seven-point scale that ranged from “strongly disagree” to “strongly agree.” Before giving their opinion, they were shown a cue about the position of their party leader on the issue, a persuasive countervailing message and evidence that went against their leader’s position, both the cue and the message, or neither the cue nor the message.
For example, some respondents were asked whether the military should be allowed to use interrogation techniques such as waterboarding to gain information from a suspected terrorist. After answering that question, Trump-voting Republicans who were assigned to see a persuasive countervailing message read a paragraph that cited a 2014 Senate report in which the CIA concluded that such techniques rarely provided reliable intelligence or got suspects to cooperate.
The message concluded, “The question then becomes: is it worth violating international law by torturing people — who are effectively innocent until proven guilty by a jury — for mainly useless information? The answer is No. America is better than that.”
Critically, informing respondents that Trump supported waterboarding did not cause them to ignore or discount the persuasive message.
In fact, exposure to the message caused respondents to shift their attitudes toward the message, and this attitude change was a similar size even when they knew that their favored party leader took the opposite position. For example, some respondents who voted for Biden might nevertheless feel that illegal immigrants should not have access to government-subsidized health care or in-state tuition rates at publicly funded colleges.
This pattern of results was consistent across a wide range of policy issues, as well as across demographic groups and different types of partisan cues.
“We found no evidence that countervailing cues from favored party leaders meaningfully diminished partisans’ receptivity to persuasive arguments and evidence — in contrast to what one would expect if party loyalty distorted partisans’ information processing,” the researchers write, adding that the findings contrast with the idea that party loyalty overrides people's values and interferes with the way they process counter-partisan messages. "If such interference and distortion does occur, our findings suggest that it is relatively minor, uncommon, or may be avoided with ease.”
Trying to bridge political chasm
Pew Research Center surveys indicate that Republicans and Democrats are more divided along ideological lines — and that partisan antipathy is deeper and more extensive — than at any point in the past two decades. And a 2012 Northwestern University study concluded that “polarization intensifies the impact of party endorsements over substantive information and, perhaps ironically, stimulates greater confidence in those — less substantively grounded — opinions.”
Rand said that’s a commonly accepted viewpoint — in fact, before this latest experiment, it was his own. “We tend to look at people on the other side of the divide as blind sheep, doing whatever their leader tells them to do, thinking whatever their leader tells them to think,” he said. “Then they look at us and say the same thing. But we suggest that people on both sides are not driven by partisan motivations but instead are more rational and reasonable.”
He said he hopes that the Nature Human Behaviour results may impel people to be less polarized and temper their dislike of people with different political beliefs.
The study has implications beyond politics. The fact that sound evidence and rational arguments can change minds is also applicable to marketing, sales, public service announcements, and other types of persuasive messages.
Appealing to people's values and identities
The authors noted that their results highlight a distinction between two key questions in political psychology: To what extent do party cues influence people’s attitudes, and why do they exert their influence? While there is relative consensus on the first question — party cues reliably influence people’s attitudes, sometimes a great deal — the second question remains unsettled.
The study makes the case that party cues are influential not because of blind loyalty to the party leader but because they contain useful information for individuals that they then weigh against countervailing information.
The range of issues and arguments in the study was larger than those of previous research but, by necessity, was still limited; the authors acknowledge that different persuasive messages containing alternate arguments and evidence might have produced different results.
The researchers believe that the messages might have worked by appealing to people’s values and identities rather than those of their party — and that party loyalty could not override the relevance of these other factors. “People have various identities and motivations that are not reducible to their party,” the researchers write. Those could include religious affiliation, gender, geographic region, economic circumstances, and ethnic background, among others.
Rand said the results have caused him to recalibrate his own beliefs. “I feel, in some sense, it’s a much more optimistic take on human reasoning than the general zeitgeist of, ‘We’re post-truth and everyone is blinded by politics,’” he said.