Ostentatiousness may not just be obnoxious among colleagues, but might actually harm long-term performance by fracturing trust and social bonds, according to a study out of Yale’s Human Nature Lab.

The study compared collaboration in a series of simple games among players given different numbers of “points,” which were later exchanged for money. Through repeated games involving almost 1,500 participants from around the world, the study found that in situations of perceived high inequality — when one player had significantly more points than the other ­— the overall outcome, both in terms of aggregate wealth and inequality, was worse.

“Wealth inequality today has exceeded the levels of the Gilded Age of robber barons,” said Nicholas Christakis ’84, head of the Human Nature Lab and senior author of the study. “We have seen that more unequal places tend to be poorer, and we wondered whether one caused the other.”

The main finding of the study was that, over the course of the game, the end state of inequality in the system was relatively low in all but one of the test groups. Only when there was high inequality to begin with and players could see the wealth of other players was inequality drastically higher than the other end states.

The results of the study have vast social applications, said Christakis. One such application might be an argument for non-disclosure of salaries in offices with a wide pay range. If two coworkers needed to work closely together, he said, they might each perform better if there was no knowledge of the other’s salary.

Each round of the game was assigned a certain Gini coefficient, a metric for measuring inequality where zero is perfect wealth equality and one is perfect wealth inequality. Throughout the game, players could choose to end some of the relationships they had been assigned if, for instance, the other player was not cooperating. The Gini coefficients for games without initial inequality or without visible wealth levels all converged at around 0.15, while the high-inequality, visible wealth games converged at around 0.4.

Players were incentivized to win by cash bonuses paid to those with the most cash at the end of the 20-minute game. They received $3 just for playing, but could earn as much as $9 or $10 if they played well. In the game, social networks were more frequently broken and defections more frequent when wealth disparities were visible.

“We don’t know why people defect more in the visibility situations, but it seems like people think that there is something unfair in the inequality situation,” said Hirokazu Shirado GRD ’20, am author of the study and a researcher in Christakis’s lab.

What is clear is that defection is related to knowledge of inequality, not inequality itself, noted Andrew Bean ’16, a founding member of the Yale Undergraduate Network Science Society, an organization that researches social networks. In the real world, of course, it is hard to find situations where those two things are separate, Bean said, noting that it can be extremely difficult in lab settings to recreate all the variables at play in the real

“In order to develop a well-controlled experiment, we have to play these abstract games, and attempt to build them so that they correspond to real-life situations,” he said. “Oftentimes, I think this can be done with a great degree of success, but there is always the risk that they do not.”

For that reason, many network labs, including Christakis’, conduct research in real-world situations, where many factors are outside of the researcher’s control, he said.

Going forward, Shirado said, the researchers hope to address the study’s limitations by expanding the demographics it reaches. Having participants from less wealthy areas with little to no access to stable Internet connections would add a new layer of data, said Shirado.

“Some poor people cannot attend our games because their Internet connection is poor,” said Shirado. “The demographics may be a little bit skewed because of this experimental setting.”

According to the Central Intelligence Agency, Lesotho has the highest Gini coefficient at 0.63 and Sweden has the lowest at 0.23.