An old but still-common method of encouraging honesty is the venerable “oath”: an explicit promise to be honest. People have used oaths since ancient times, says Janis Zickfeld, a social psychologist at Aarhus University in Denmark. “If you’re a witness in court, you have to swear an oath to tell the truth, for example,” he says, “or in the Netherlands they have this Dutch banker’s oath for people working in the financial industry concerning moral work practices.”
Researchers believe oaths have the power to increase honest behavior, but they often come with penalties for transgressions, so the power of the promise itself is unclear. Can merely making a promise to be honest without any punishment involved lead people to behave more honestly, even when there’s reason to lie?
A study led by Zickfeld that was published in last month’s issue of Nature Human Behaviour suggests that it can, but the wording of the oath matters. The way in which the oath is made, and its timing, also make a difference. The findings suggest that oaths could be low-cost tools to curb dishonesty, but to confirm this assumption, these results need to be replicated in real-world settings.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
In their study, Zickfeld and his colleagues recruited 21,506 people from the U.S. and the U.K. on an online platform to play a game that simulated paying taxes. Participants performed a sorting task in which they earned money based on how quickly they finished. They then reported how much they had made, which was taxed at 35 percent. Participants could lie without consequence but were also told the collected tax would go to the British or American Red Cross.
The researchers generated 21 “honesty oaths,” including a “baseline” oath: “I hereby declare that I will provide honest information in this study.” Two oaths had phrasing that was either more severe—“I hereby swear upon my honor …”—or more specific: “… that I will provide honest information when reporting my final income from the sorting task.” The other 18 employed strategies such as emphasizing social bonds of empathy or trust (“To earn the trust of my fellow citizens …”), highlighting aspects of self-image, such as responsibility (“I understand that it is my responsibility to report honestly”) or character (“I am an honest person …”), or appealing to social norms (“I understand that most people agree that reporting honestly is the right thing to do”). A control group of 953 participants performed the task without taking any oath.
A quarter of participants were dishonest to some degree; 7 percent falsely reported zero income from the task. “We have 75 percent of people being fully honest, which is nice,” Zickfeld says. “Even if there’s no consequence, people still prefer to be honest.” This meant that 14 percent of the potential tax revenue, totaling about $930, was lost. The researchers also calculated a “tax compliance” score, the percentage of income honestly declared. A completely honest participant scored 100 percent, those who reported no income scored 0 percent, and everyone else fell somewhere in between. This tax compliance figure averaged 82.3 percent for participants in the control group, compared with 86.2 percent averaged across all the groups that made oaths. Overall, including an oath increased compliance by 3.9 percent. This might not seem like much, but note that most people in the study were completely honest to begin with, so this was driven by changing the behavior of a minority of participants.
The oaths were far from equal in preserving honest behavior. Only 10 were effective, and some were much more effective than others. “Ten were statistically significant and increased tax compliance from between 4.5 to 8.5 percentage points,” Zickfeld says.
In the control group, 31.3 percent of people were dishonest, and 22 percent of tax revenue was lost as a result. In the group that took the most effective oath, 18.5 percent of people were dishonest, which reduced tax losses to 11.6 percent. “The most effective pledge nearly cut cheating in half,” Zickfeld says.
This approach involved making the baseline oath more specific. The next most effective spelled out the meaning of dishonesty or the rules to be followed (“I understand that misreporting is forbidden in this study”). Casting honesty as all-or-nothing (“either the reporting is honest or it is not”) and appealing to social norms were also both modestly effective. Oaths highlighting social bonds or self-image did not have a significant effect.
Almost everyone thinks of themselves as honest, and we behave in ways that allow us to maintain this positive self-image, so it is surprising that appeals to self-image were not effective. But oaths that explicitly reference the targeted behavior and spell out the rules of how to behave may make it harder to “wriggle out” of being honest without damaging one’s self-image. “One takeaway is that subtle hints are not sufficient,” says Shaul Shalvi, a behavioral economist at the University of Amsterdam, who was not involved with the research but wrote an accompanying commentary on the work. “It seems that when you want people to follow their promises, it’s good to be clear.”
The team also found that the timing of when the oaths were made mattered. Placing it right at the start of the experiment, before participants performed the sorting task, was less effective than placing it after the task, immediately before they reported income.
The researchers also tested whether retyping the oaths produced a better result than ticking a checkbox. “Typing was more effective for some [formulations], especially the more effective ones, but not on average,” Zickfeld says. Retyping an oath forces people to think about it more, so this might amplify effective oaths. “I think that’s exactly what you’re seeing there,” he says.
In the study, men were more dishonest than women on average, particularly if they were younger, and scored low on a personality dimension called honesty-humility. “On average, men are more likely to cheat [in these types of studies],” Zickfeld says. “Some say it’s due to being higher in risk-taking.” Participants from the U.S. were more dishonest than those from the U.K., which may indicate cultural differences, but the study used the British pound, so it could be that foreign currencies seemed more abstract, or less “real,” to the American participants.
Previous research on the effectiveness of honesty oaths returned mixed findings, so the new study provides much-needed data on what works and what doesn’t. Researchers now need to replicate these results for the effective oaths “and assess how robust each is across different cultures, settings and in the field,” Shalvi says.
Zickfeld and his colleagues are talking to their local hospital about assessing whether oaths can improve rates at which loaned medical equipment is returned. “That’s the current plan,” Zickfeld says.
Honesty oaths could ultimately be a useful tool. “The beauty of these interventions is: they’re very cheap,” Shalvi says. But this study is only a first step. “Once we know what effects are robust, we can go to policymakers and suggest randomized controlled trials in target populations,” Shalvi says. “Then, if that works, we can look at changing tax forms or whatever.”
