James Farley/Booz, Allen & Hamilton Research Professor; Professor of Managerial Economics & Decision Sciences
Georgy Egorov’s researcher ears perked up during the 2016 presidential campaign when people started talking about how then-candidate Donald Trump’s rhetoric was changing what was considered “normal.”
Egorov, a professor of managerial economics and decision sciences at Kellogg, had conducted previous research that touched on the issue of social norms—those unwritten rules that govern most people’s behavior. Few of us like deviating too far from what’s deemed normal, lest we end up stigmatized for our actions or views.
But social norms do, of course, change over time. Yet little is known about what, exactly, causes social norms to change, or how quickly those changes can occur.
So Egorov and colleagues decided to study this question in the context of Trump’s anti-immigrant rhetoric. Social norms had previously deterred people, in general, from voicing xenophobic opinions. But there was candidate Trump, calling to ban all Muslims from entering the U.S. and describing Mexican immigrants as “bringing drugs, they’re bringing crime, and they’re rapists.”
While politicians from both parties denounced such sentiments, the researchers wanted to know if evidence of Trump’s popularity among voters would make people more likely to express xenophobic views.
They found that, indeed, that happened.
In a series of experiments, participants were more likely to publicly express support for an anti-immigrant organization when prompted to think about how popular Trump was in their community. Additionally, people indicated that they were willing to be forgiving of those who might be feeling pressured by new social norms to support anti-immigrant causes.
“What we see here is that new information about the distribution of opinions in society can actually change social norms, and that happened quite quickly,” Egorov says. “The way Trump campaigned, and especially the fact that it increased his popularity, revealed some new information that increased the expression of xenophobia.”
The researchers, who included Leonardo Bursztyn of the University of Chicago and Stefano Fiorin of the University of California, San Diego, conducted their first experiment two weeks before the 2016 election.
They recruited roughly 450 participants online in eight solidly Trump-leaning states to answer a short research survey.
“The way Trump campaigned, and especially the fact that it increased his popularity, revealed some new information that increased the expression of xenophobia.”
Participants were then asked if they would authorize the researchers to make a small donation to an anti-immigrant organization on the participants behalf as a thank you for their participation. To bolster the organization’s xenophobic credentials, its founder was quoted as saying that the persistence of U.S. culture “requires a European-American majority, and a clear one at that.” Participants who approved the donation were given a small additional payment.
Importantly, just before being asked about the donation, half the participants were told that Trump’s odds of winning their home state were 100 percent, according to the website PredictWise. Additionally, half the participants in each group were told that their decision on the donation would be anonymous, and thus made in private, while the other half believed that the researchers would be able to see who had donated.
Among participants who had not been told about Trump’s popularity, there was a big difference between those who would donate publicly versus privately: 34 percent compared to 54 percent. But, when told ahead of time that Trump was supported by their state’s electorate, that difference disappeared. In that group, 46 percent donated publicly and 47 percent did so privately.
The information about Trump’s popularity, the researchers write, “causally increased the social acceptability of the [donation] to the point of eliminating the original social stigma attached to it.”
The researchers repeated the study a week after the election’s surprising outcome. This time, however, they did not give participants any information about Trump’s popularity; they let the election results speak for themselves.
Postelection, the researchers found virtually no gap between those who were willing to donate publically, 48 percent, versus those who donated privately, 49 percent.
The researchers admittedly hurried to conduct this experiment, assuming, as did much of the country, that Trump would lose and thus their window to study his rhetoric would close. But, after Trump’s unexpected victory, they were able to conduct a similar, more comprehensive experiment.
This time they focused on 1,600 participants recruited by a survey firm in and around Pittsburgh. That geographic area offered the advantage of having a particular voting pattern: Pittsburgh’s metropolitan area voted in favor of Trump, while Pittsburgh’s Allegheny County favored Hillary Clinton.
This allowed the researchers to truthfully tell half the participants about Trump’s local victory and half about Clinton’s.
Again, social norms appeared to play a role in whether people authorized a donation to the anti-immigrant group. Among those who were told of Clinton’s win, only 20 percent were willing to donate in public compared to 30 percent who thought the donation would be private. As in the previous study, this difference essentially disappeared among the group of participants who thought that Trump won locally, indicating that there was no perceived social stigma to supporting the anti-immigrant group if their fellow Pittsburghians also supported Trump.
The researchers also added a new question to this study, asking participants to predict the share of Pittsburgh voters who would agree with anti-immigrant statements such as, “Both legal and illegal immigration should be drastically reduced because immigrants undermine American culture and do not respect American values.”
The participants’ responses suggest that people were more willing to express xenophobic views because they thought such ideas were not unpopular. Indeed, those told about Clinton’s victory predicted that 42.6 percent of local voters agreed with the statements compared to an estimate of 50.1 percent among those who were told about Trump’s win.
Trump’s victory also gave the researchers time to create a mathematical model to help explain their initial results. The model predicted that people would judge others less harshly for the same behavior if they thought those people were making decisions because they felt pressured by social norms.
“He or she will give you the benefit of the doubt,” Egorov explains.
To test this, the researchers conducted a final experiment in 2018 that built off their one in Pittsburgh.
They used a survey company to recruit 1,830 people from across the U.S. who self-identified as Democrats and were thus less likely to share Trump’s anti-immigrant views.
These participants were “matched” with participants from one of the four experimental groups in the previous study: those who thought Clinton won and could donate in public, those who thought Clinton won and could donate in private, those who thought Trump won and could donate in public, and those who thought Trump won and could donate in private. The new participants were shown the same text as the Pittsburgh participants and were then told that the person they were matched with had authorized the donation to the anti-immigrant group.
Next, the Democratic participants got to choose how much of $2 they would share with their fellow player. The researchers wanted to see how differently these Democrats would treat their teammate for expressing xenophobic views in the four different conditions.
In three of the conditions, the participants gave nearly the same amount on average—about 80 cents. But, in the one condition where public pressure could be at play—when people had been told of Trump’s local victory and their donation would be public—the Democratic participants were more forgiving, giving their teammates an average of 88 cents.
One explanation for this result, Egorov says, is that while these Democratic participants may have had a hard time understanding another person’s anti-immigrant views, they likely could empathize with feeling pressured by social norms to act a certain way.
“Maybe that’s something where they can put themselves in the other person’s shoes,” he says. “They can empathize.”
Egorov stresses that this research did not look for or find evidence that Trump’s rhetoric actually caused people to become more xenophobic. It focused only on a willingness to publicly endorse the cause.
It’s unknown, however, if changes in social norms around speaking out on a topic could eventually change people’s beliefs on that issue. Future research could delve into this, Egorov says, shedding light on debates over the limit of free speech. For example, is it better to ban speech by Holocaust deniers, as Germany does, or allow it as a First Amendment protection, as Americans do?
“Most of us likely agree that it’s bad for people to have these views,” Egorov says. “But what is most effective in preventing the proliferation of these ideas? That’s a question for future research.”