People carry around biases—subconscious or otherwise—about social groups and often treat members of different groups differently. But until now, it has been a challenge to capture links between such underlying stereotypes and their effects on how people act.
Across six studies, researchers from the University of Pennsylvania and the University of California, Berkeley, found that stereotypes about two well-studied traits, warmth and competence, play a large role in behavior by changing how much people care about equity. Capturing this in a computational model, the researchers were able to predict how people would behave in both laboratory and field studies.
“‘Warmth’ corresponds to how good or bad you think a person’s intentions are likely to be toward others. ‘Competence’ corresponds to how capable you think the person is of acting on those intentions,” says Adrianna (Anna) Jenkins, an assistant professor in Penn’s Department of Psychology. “We found that stereotypes about the warmth and competence of different groups had opposite effects on people’s willingness to tolerate unequal distributions of resources with members of those groups.”
Jenkins and colleagues, who published their work in the Proceedings of the National Academy of Sciences, started from the baseline that people in general dislike unequally distributed resources, what’s known in economics as “inequity aversion.” Someone who has more resources relative to another person is said to experience “advantageous inequity,” and someone in the lower-status position experiences “disadvantageous inequity.”
The researchers’ first goal was to determine, in a lab setting, how stereotypes about different social groups’ warmth or competence influence people’s willingness to impose such inequality. They recruited 725 study participants to play a version of the “Dictator Game,” in which each person received $10 and a single social characteristic about one of 20 possible partners—a nursing occupation, for example, or Japanese nationality—then decided how much money to keep and share. In one sub-group, participants received an actual payout for the amount they retained.
To measure stereotypes about the groups, the researchers also asked a separate 252 participants to rate the 20 partners on warmth and competence.
The results were telling. When people had more resources, they paid closest attention to their partner’s warmth, and, when they had fewer resources, they cared more about their partner’s competence.
“In other words, the warmer the partner group was perceived to be, the less willing people were to take more money than that other person,” Jenkins explains. “The more competent the partner group was perceived to be, the less willing people were to take less money than their partner.”
To test their model’s ability to predict how people might treat others, they then trained it on 19 of the 20 partners, and had the model guess the outcome for the 20th. It guessed very well—in fact, the model predicted just as well when trained on as few as half of the partners.
“This is basically asking, what’s the interplay between subjective stereotype information and objective monetary value in driving people’s decisions in this game?” Jenkins says. “By capturing that interplay, how much of people’s behavior can we actually explain? It turns out, quite a bit, almost twice as much as previous models.”
For Jenkins and her collaborators, which included Ming Hsu, senior author on the paper and an associate professor at Berkeley, the next question asked, “What relevance do these findings have in the real world?” To answer that, they turned to two past field studies that documented treatment disparities. One looked at callback rates for job applications, the other on response rates of professors asked to mentor students.
From 318 participants, the researchers collected ratings on the warmth and competence of about 60 names used in those studies, names stereotypically associated with different genders and ethnicities.
“We showed participants a name and asked, ‘How competent or how warm do you think a typical person with this name might be?’” Jenkins says. “Because we realized people might have hesitations about reporting their stereotypes, we asked them to tell us what their first impression would be, even if they thought it might be inaccurate.”
Much like with the lab-based studies, the model was able to predict the outcomes of the field studies with very high accuracy.
“Scientifically and practically, this is a step forward,” Jenkins says. “It means that these intangible things that are sometimes thought of as wishy washy—that is, stereotypes about different social groups—actually have meaningful structure that can be quantified and used to study and predict human behavior, not only in the lab but also in the wider world.”
It also goes beyond the binary fashion in which discrimination typically gets studied: male versus female, black versus white, with one group designated the pair’s winner. The research found, for instance, that in one version of the resource-allocation process of the Dictator Game, “lawyers” were penalized 67 cents for their lack of perceived warmth, and 29 cents for their perceived competence. “Nurses,” on the other hand, were rewarded 47 cents for their perceived warmth, but penalized 16 cents for their perceived competence.
Jenkins says the consistency across studies of participants’ views of different social groups surprised her. She also notes that, in real life, rarely do people make decisions about others based on a single piece of social-group information, as was the case in this work. For this reason, she and her colleagues are already working on additional studies looking at how the mind integrates social-group markers with each other and with other information about a person to drive valuation processes.
“People constantly need to make decisions about how to behave toward others. Often, those decisions rely on information we can’t access directly, so one way we fill in those gaps is to turn to stereotypes,” she says. “But when whole groups of people rely on stereotypes in a similar way, they can become systematic, contributing to some of the biggest disparities that we see in society.”
Funding for the research came from the National Institute of Mental Health (Grant MH098023) and the National Institute on Drug Abuse (Grant DA043196).
Collaborators on the work included Ming Hsu of the University of California, Berkeley; Pierre Karashchuk of the University of Washington; and Lusha Zhu of Peking University.