Quick quiz: what percentage of Americans do you think were worried about climate change in the early 1990s, and how did that figure shift over the next few decades?
One might reasonably assume that many more people are concerned about the issue now than in the past. After all, the world has been besieged by wildfires, droughts, and more powerful storms, with constant coverage of these disasters in the media. But the real answer is that 61% of people surveyed in 1990 expressed concern about climate change—only slightly less than the 65% who did so in 2019.
In a recent study, Jason Dana, an associate professor of management and marketing at Yale SOM, and Adam Mastroianni of Columbia Business School found that these mistaken assumptions are common. When they asked people how much attitudes had shifted on a variety of social issues, participants’ average guesses were often wildly off.
“They have systematic misperceptions of how attitudes are changing,” Dana says.
And these mistakes weren’t random. They tended to be biased in one direction: people often assume that past opinions were much more conservative than they really were, and that attitudes had therefore changed much more than they had in reality. In some cases, participants were even wrong about the direction of change. They thought that a position had become more popular when the reverse was true, or vice versa.
Such misperceptions could affect the prospects for genuine change. The study found that when people believe public support for a position is rising, they think that lawmakers are more justified in pushing for that position—even when they disagree with it themselves.
“Attitude change legitimizes policy,” Dana says.
The study grew out of Dana and Mastroianni’s previous work on misperceptions of other issues. For instance, Dana had found that people often think that many items have become more expensive over time (in terms of how long one needs to work to be able to afford to buy them) when the opposite is true. Similarly, Mastroianni had found a widespread belief that people are becoming less moral, but survey data looking at markers of actual behavior suggests that this impression is mistaken.
The researchers investigated people’s perceptions of public opinion on 51 social issues, covering topics such as gender, race, religion, immigration, abortion, and gun control. For each issue, the team gathered data on how attitudes had changed from Gallup, the Pew Research Center, the General Social Survey, and other sources.
Then they asked 943 study participants to guess how public opinion on those issues had shifted from the earliest to the most recent years in the poll data. For instance, people estimated what percentage of people said they would vote for a qualified female presidential candidate in 1972 and in 2010.
In general, the study participants’ guesses were badly wrong. They overestimated how much social change had occurred for 57% of issues, underestimated it for 20%, and guessed the wrong direction of change for another 20%.
People believed that willingness to vote for a female president had leapt from 32 percent to 70 percent, when in fact it had gone from 74 percent to 96 percent.
For instance, people believed that willingness to vote for a female president had leapt from 32% to 70%, when in fact it had gone from 74% to 96%. They thought that feelings toward Black Americans (measured on a warm/cold “feelings thermometer”) were much more unfavorable a half-century ago; in reality, they were similar to people’s feelings today. And they guessed that support for an assault weapons ban had increased from 41% to 56% over the last two decades, when it had actually declined from 58% to 48%.
Why was this happening? The bias appeared to be partly driven by an assumption that people were much more conservative in the past, and therefore that much more progress in the liberal direction had occurred.
To confirm this hypothesis, Dana and Mastroianni performed another study in which they asked participants to rank how liberal or conservative the two positions on each issue were. The team then calculated a score capturing how partisan each person considered each issue.
People systematically overestimated shifts in the direction they considered liberal, and this bias was even more prominent on issues that they thought were highly partisan. Participants seemed to assume that if a position was very liberal, “people must have really been against it in the past,” Dana says.
The team doesn’t know exactly how this misperception formed. One possibility is that people’s impressions are informed by TV shows and movies about past eras, which might not accurately reflect true attitudes. Or participants might have mistakenly assumed that major events, such as the 2008 election of Barack Obama, heralded a tectonic shift in attitudes. (People had estimated that stated willingness to vote for a Black presidential candidate skyrocketed from 32% to 74% from 1978 to 2010, when the change was actually much smaller: 85% to 97%.)
To explore how these misperceptions might affect policy, the team ran experiments about three issues: immigration, assault weapons, and the death penalty.
In the first two experiments, people were were asked about their support for a piece of legislation after being told about the real numbers on attitude shifts from polls, and after hearing the numbers estimated by earlier study participants. For instance, they were asked how justified Congress would be in limiting immigration if support for this position had decreased from 42% to 35% (the actual shift in public sentiment), or if support had increased from 42% to 56% (the incorrect estimate from the study).
When people thought public support for a position was increasing, they said that lawmakers would be more justified in pushing through that policy. And this pattern held true even when there was no difference in the final percentage of supporters. In a third experiment, participants were told to imagine that support for banning the death penalty had either risen from 33% to 50% or dropped from 67% to 50%. They felt that lawmakers were more justified in enacting the ban in the first scenario.
“They think policy change is more appropriate if support has been growing than if support has been shrinking, even if the level of support right now would be the same,” Dana says.
The results show that mistaken assumptions about the public’s attitude on an issue can have a real-world impact. If people believe that most people don’t support a particular policy, they might give up on fighting for it. Or in the opposite scenario, they may become complacent. If they assume social change is already moving in the direction they want, the authors write, “they might not do what is necessary to secure victory in the first place.”