Incentives are a potent tool for shaping human behaviour, but they’re famously tricky to get right. New research,. published in Basic and Applied Social Psychology, has uncovered an alarming wrinkle that complicates incentives even more: they may make powerful people less ethical. 

The study, led by Jessica Swanner at Iowa State University and Denise Beike at the University of Arkansas, involved 250 undergraduates at the University of Texas at El Paso. The researchers set up a scenario in which students were asked to inform on a fellow student who had confided in them. 

Students were told they would be given course credit to participate in computer tests on unconscious memory. When they arrived for the study, however, a researcher told them that a previous student had crashed the computer, so instead they would be participating in a study on interpersonal interactions and decision making. Each student was paired with a participant who was secretly a confederate of the researchers. The pairs discussed a series of personal questions, culminating with the confederate either admitting to or strongly denying crashing the computer. 

Next, the researchers randomly assigned each student to a boss (high-power) or employee (low-power) role for the next task. But before the task started, a researcher individually asked students for their help in figuring out who crashed the computer. The researcher offered additional course credit to half of the students for their assistance. (The other half, constituting the control group, wasn’t offered anything.) The researcher then asked the student whether their partner mentioned crashing the computer, prompting the student up to eight times to turn in their partner. 

If the student said yes — regardless of what the confederate actually said — they were asked to sign a statement testifying that their partner crashed the computer. The researcher noted how many prompts were needed before the student signed the statement, as well as whether the student tried to change the terms of the agreement or the wording of the statement. Later, the students were quizzed on what exactly the confederate said, the incentive the researcher offered, and how confident they felt about remembering the details of each. 

For most combinations of factors (high-power or low-power role, partner confession or not, incentive or not), the researchers found that offering an incentive had almost zero effect on whether the student turned in their partner. All students, both high-power and low-power, were much more likely to sign a true incriminating statement than a false one. In other words, everyone was more likely to say the confederate had admitted to crashing the computer when the confederate had actually admitted to it. But high-power students were three times more likely to sign a false incriminating statement when they were offered an incentive.

High-power students who were offered an incentive also stood out when it came to the number of prompts. All students were prompted up to eight times to turn in their partner (for example, “Are you sure they didn’t mention anything about crashing the computer?”). When the confederate had admitted to crashing the computer, all students became increasingly likely to turn in their partner for the first five of the prompts, after which additional prompts had no effect. But when the confederate had denied crashing the computer, high-power incentivized students became more likely to turn in their partner the more they were prompted to do it; everyone else levelled off after four prompts.

This study isn’t the first one to show that power causes people to focus more on their own rewards than on others’ well-being. Although an incentive may be intended to encourage proper behaviour, it may instead make a powerful person do whatever is necessary to secure the reward. 

The study’s findings are significant for the business world, where powerful people can have a huge effect on the careers and day-to-day work of those beneath them, but the implications go much further. The paper mentions two examples of incentives gone very awry: First, in 1994, Kenneth Wyniemko was wrongfully convicted of criminal sexual misconduct, breaking and entering, and armed robbery. He was sentenced to 40–60 years in prison despite the victim saying she had little opportunity to view her assailant. Among the evidence against him was a fellow inmate’s incriminating testimony, provided in order to avoid a life sentence. Second, in 2014, the mayor of Edmundson, Missouri, told the local police that their salaries would be increased according to the number of traffic tickets they wrote. The police department was later accused of making false citations in record numbers, especially against African-American citizens. 

The researchers point out that incentives are not inherently harmful; problems tend to arise when the behaviours that are incentivized pit powerful people against weaker ones. One way to counteract how incentives can corrupt may be to incentivize cooperative behaviours, such as employee satisfaction or retention. Keeping in mind how incentives can affect people’s behaviour, in good ways and bad ways, can help us use incentives more effectively — and fairly.

This article was originally published in HBR.