In late September 2015, then-candidate Donald Trump sat for an interview with Scott Pelley of CBS’s “60 Minutes” in which he railed against trade agreements. “We’re being defrauded by all these countries,” Trump barked. Pelley responded in dismay, “It’s called free trade and it’s a plank of the Republican platform.”

Two and a half years later, now-President Trump has taken an ax to that platform, imposing controversial new tariffs on imports from China, the European Union, Canada and Mexico, moving the world closer to a full-fledged trade war.

As Republican orthodoxy on the issue has shifted, so too has public opinion. When asked whether trade agreements are good or bad for the country, voters are increasingly divided by party. During the 2016 campaign, Republican opinion shifted somewhat dramatically from supporting to opposing free trade — although that opinion has shifted back somewhat. Meanwhile, Democrats’ support went up during the campaign — and has remained markedly higher than in the past.

What’s going on here?

Trump’s opposition to free trade is changing the way the public thinks about the issue. Political scientists have long argued that many Americans decide where they stand on issues by taking cues from the things said by elites in government or voices in the media. Party leaders set the agenda, and the party faithful adjust their own views accordingly.

However, in a just-published article, we argue there is more to that story. Partisans gravitate toward their own party’s norms, but those norms are communicated by both elected leaders as well as “ordinary” partisan peers. What you think about tariffs may depend not only on what people like Trump say but also on your beliefs about what other people like you think.

Here’s how we did our research

We conducted two online survey experiments on a representative sample of 1,023 Americans recruited by Survey Sampling International between Aug. 16 and Aug. 23, 2016.

Respondents were asked to read excerpts from (fictional) news stories about two contentious policy debates: the Trans-Pacific Partnership (TPP) trade agreement in one experiment and the educational standards initiative called Common Core in the other.

In each experiment, respondents were randomly selected to receive different versions of the stories. Some received only details about the policy debate itself while others received additional survey results about where the parties stood on the issue. Those who received the survey information were presented either with poll results about the policy leanings of ordinary citizens who belong to each party or poll results revealing where Democratic and Republican members of Congress stood on the matter.

Throughout the experiments, the poll numbers were fabricated for purposes of experimental control. We debriefed our respondents, explaining this deception at the end of the study.

Here’s what we found 

In both experiments, people were just as willing to “follow” partisan peers as they were elites in Congress. For instance, when Democrats read that fellow Democrats opposed the TPP while Republicans supported it, Democratic participants’ support for the agreement was close to half a point lower on a five-point scale than their fellow Democrats who hadn’t been told about anyone’s opinion.

Similarly, when told the same information about where rank-and-file Democrats and Republicans stood on the TPP, Republican respondents’ support for the agreement was 0.4 points higher than those who weren’t given any opinion data.

Those who read about where the parties in Congress stood on the TPP reacted similarly, although the differences were more modest. In another era, it is possible that congressional elites might have more influence. But as Julia Azari explained, we’re in a time of “weak parties and strong partisanship.” People are at least as likely to lean on what “people like them” think as they are to follow the views of their elected leaders.

The desire to belong shapes American opinions

Our findings are in keeping with a growing political science literature on how the desire to conform to group norms shapes ideology. Our social identities drive our politics to a significant degree, as political scientist Lilliana Mason argues in her recent book, “Uncivil Agreement: How Politics Became Our Identity.”

In keeping with these findings, we found the people most susceptible to partisan “peer pressure” were those who identified most strongly with their party, Democrats or Republicans, as a social identity. Partisan social identity didn’t seem to predict respondents’ willingness to conform to the views of their party’s members of Congress, however.

Following one’s partisan tribe may have its downsides. When motivated by belonging or by wanting the other side to “lose,” individuals may not consider what careful policy would best serve their own or the country’s interests.

As political scientists Alan I. Abramowitz and Steven W. Webster have noted, partisans are increasingly tied together by their strong dislike of the opposing party. This form of identity-driven partisanship may push people to put party above nation, to justify ends over means and contort their views to fit their loyalties.

But one upside is this: Partisans do seem to listen not just to elites but also to fellow partisan peers. That means different factions within parties can actually sway others’ policy stances, and presumably the party’s direction. Such coalition building has long been seen as a cornerstone of democracy. But it may require forms of listening that our contemporary media and democratic system may not be particularly well suited to foster.

Benjamin Toff is assistant professor of journalism and mass communication at the University of Minnesota.

Elizabeth Suhay is assistant professor of government at American University.