All else equal, having a white guy carry out an experiment can change the results in big ways.
To understand how humans make decisions, economists and psychologists often study how people play “games” that mirror real-life choices. The results of these games underpin many theories about why people act they way they do. But a new paper—titled “The white-man effect”—shows that having a white or black researcher influences what people do in these games, calling into question just how much we really know about how people make decisions.
The authors went to 60 villages across Sierra Leone to play ”the dictator,” a decision game with simple rules. There are two players, a ”dictator” and a “subject.” Researchers give cash to the dictator and tell him or her to share any amount with the subject. You might think that dictators keep all the money for themselves, but they usually give some away, often up to 50% of the cash. This shows that people don’t always act only in their own self-interest.
But the authors, led by Jacobus Cilliers of Georgetown University, found that generosity varies greatly based not on the characteristics of the people playing, but of those running the game.
In their experiments, four of the people administering the game were black Sierra Leoneans. There was also a fifth “supervisor,” who was either a white American or another black man from Sierra Leone. When the supervisor was white, dictators gave 19% more money to their subjects than if the supervisor was a fellow black black Sierra Leonean.
The authors were careful to control for other factors, by choosing similarly friendly, male, and well-educated supervisors, and by not telling these supervisors that they were in fact part of the experiment.
These results come as more research relies on such games to explain the behavior of people in Africa and other developing regions. The authors collected data on the number of mentions of dictator games on Google Scholar in papers on these regions, and the trend is clear:
So what might make Sierra Leoneans share more just because there is a white guy around? The authors have a theory.
People who are being studied, regardless of the circumstances, tend to act base on what they think researchers want from them, even if they’re explicitly told to act normally. In Sierra Leone, it looks like the respondents thought the white supervisor expected more generosity than his black counterpart. This, they explain, reflects a perceived power relationship between researcher and respondent.
Respondents will try to please any supervisor, black or white. But if they see him as being particularly rich or powerful, as they did with the white American, they will try even harder. That explanation is backed up by the fact that relatively powerful respondents in the study—like household chiefs or religious leaders—tended to give less than others when the white supervisor was present.
The implication is that every aspect of a study matters. Decision research has been criticized for attempting to explain all of human behavior based mainly on studies of undergraduates in rich democracies. That has led to repeating such research in other parts of the world, as the chart above shows. But that might not be enough.
“Behavioral studies that offer ‘cultural’ or other contextual explanations for variation in generosity should be taken with a grain of salt, unless we are confident that such differences aren’t driven by simpler explanations such as who was in the room at the time,” said Bilal Murtaza Siddiqi, an economist at the World Bank and one of the paper’s co-authors.
Studying a handful of villages in a single country does not prove that the race of the researcher will always influence outcomes. The point is that it might.
The authors don’t speculate as to what might happen if “rich” respondents played games administered by “poor” supervisors. Would the dynamic flip, with rich dictators not feeling obligated to please their supervisors and keeping close to 100% of the money?
In short: Next time you read about a study that shows how we decide about this or that, consider the white-man effect.