Study explores the effectiveness of honesty oath for reducing dishonest behaviors

20 Nov 2024, 12:30 by Ingrid Fadelli

This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:

fact-checked

peer-reviewed publication

trusted source

proofread

Credit: Zickfeld et al., Nature Human Behaviour, 2024.

Over the past decades, psychologists and policymakers have been trying to devise interventions that could dissuade individuals from engaging in dishonest social behaviors, such as tax evasion or fraud. One promising strategy they identified entails asking people to commit to being honest before they are given the opportunity to engage in dishonest behavior.

While many studies found evidence suggesting that honesty oaths can reduce dishonest behavior, very few explored how different interventions utilizing these oaths compare. Researchers at Aarhus University and other institutes worldwide recently carried out a study investigating the effectiveness of various ex-ante honesty oath interventions.

Their findings, published in Nature Human Behavior, suggest that while different honesty oath-based interventions can successfully curb dishonesty, their effectiveness varies greatly depending on their content. The researchers also identified an intervention that appears to be particularly promising for reducing dishonest behavior.

"We have been studying the fascinating reasons why people engage in unethical behavior, such as dishonesty or tax fraud, for some years," Janis H. Zickfeld, first author of the paper, told Phys.org.

"There is a whole literature on so-called nudges, which refers to small or subtle changes in the environment that can influence behavior and decision making. This ranges from reminding people of being honest to having them sign an oath or pledge of honesty."

Some scholars have questioned the effectiveness of the various honesty oath-based interventions tested in the past. As part of their previous research, Zickfeld and his colleagues conducted a meta-analysis of past studies in the hope of gaining a clearer idea of the extent to which these interventions are effective.

"A meta-analysis focuses on combining all available studies and effects on a relationship," said Zickfeld. "As part of our meta-analysis, we analyzed 124 effects and 39 articles that tested the effect of honesty oaths on honesty. We found a positive effect (g = 0.27 [0.19, 0.36]), suggesting that honesty oaths can increase honest behavior. However, effects differed quite a lot based on the study type with some studies showing no effect and a minority even negative effects."

When they conducted their meta-analysis, Zickfeld and his colleagues observed that the studies they reviewed used different formulations of honesty oaths, and these stark differences could have influenced the extent to which they were effective. In addition, they found that past studies had asked participants to express their commitment to these oaths in varying ways.

"Some studies just used check boxes, others signatures, and others re-typing the formulation, yet there was limited information about whether the exact way of committing had an effect," said Zickfeld.

"For this reason, we set up a megastudy aimed at testing the effectiveness of honesty oaths. Megastudies have become a popular concept in the social sciences recently. They typically focus on testing many interventions at the same time (20+) in a field-like setting (so something that resembles real-world behavior more closely)."

Generally, research studies testing psychological interventions only test 1 or 2 interventions at once. Megastudies like the one recently carried out by Zickfeld and his colleagues reach beyond this number, simultaneously testing a wider range of interventions.

Discover the latest in science, tech, and space with over 100,000 subscribers who rely on Phys.org for daily insights. Sign up for our free newsletter and get updates on breakthroughs, innovations, and research that matter—daily or weekly.

The team's mega-study had several objectives. Firstly, it was aimed at evaluating the overall effectiveness of ex-ante honesty oaths (i.e., oaths/pledges taken before a participant can engage in the behavior of interest) in reducing dishonesty.

The researchers then wished to investigate whether different oath formulations and ways of committing to an oath (i.e., re-typing or checking a box) influenced the effectiveness of interventions. They also looked at the impact of timing (i.e., whether an oath was placed directly before a participant could engage in dishonest behavior or a bit earlier), as well as personality traits and socioeconomic variables (e.g., gender, nationality, age, etc.).

"For the study, we invited several experts in the field of dishonesty from different disciplines such as psychology, economics, management or criminology," said Zickfeld. "A total of 44 experts contributed to the study by suggesting potential interventions that were based on previous studies or theoretical reasoning. We received 98 suggestions of interventions and after implementing a vote on the most suitable and perceived as most effective interventions, 20 were retained."

The 20 interventions identified with the guidance of experts, along with a baseline oath intervention (i.e., employing a generic oath) and a control intervention (i.e., employing no honesty oath whatsoever) were tested on a total sample of 21,506 participants living in the UK and US. Each participant was randomly allocated to 1 of the 22 possible interventions.

"The participants all took part in a so-called tax evasion game, a task typically used in economics to measure tax reporting behavior," said Zickfeld. "The game is rather simple. In the first step, people can gain money by completing a task. In our case, this task consisted of ordering 8 numbers into odd or even numbers."

At every round of this tax evasion game, the study participants had 30 seconds to sort 8 numbers, and whatever time they had left would be awarded to them as income. If they only used 10 seconds to sort the numbers, for instance, they would receive 20s as their income, which was transformed into 20 pence.

After they completed four rounds of this game, the participants' income was summed together. The researchers found that they earned £0.49 on average throughout the whole number sorting task.

"In the second step, participants then were asked to report their final income like they would do in a tax report," said Zickfeld. "Similarly to a tax report, we applied a tax rate of 35% to the reported income. The taxes were paid to the British or American Red Cross to simulate that taxes are also used for a greater good."

Notably, during this second stage of the experiment, participants could also underreport the money they had earned to pay fewer taxes. For instance, if they earned £1, they could report their win as £0 and pay no taxes.

"To test whether committing to an honesty oath could reduce possible underreporting in taxes, we asked participants to commit to an honesty oath before reporting their income," said Zickfeld. "The exact wording and procedure differed per intervention, but this read something like 'I hereby declare that I will provide honest information in this study' (baseline oath)."

Zickfeld and his colleagues found that approximately 1 out of 4 participants (25.1%) had underreported their income to some degree. Moreover, 7.1% of participants had reported no income when they had, in fact, won some money.

"£737.96 in taxes were lost due to this dishonest behavior (which was 14.1% of all possible taxes)," said Zickfeld. "We observed that, compared to the control intervention with no honesty oath, all 21 honesty oath interventions (including the base oath) increased tax compliance and decreased tax underreporting, so participants were more honest when asked to commit to an honesty oath. However, the effectiveness of these 21 interventions differed significantly from each other."

The researchers thus observed that even if all the honesty oath-based interventions had reduced dishonest behavior to some degree, some were only slightly more effective than the control intervention, while others achieved statistically significant results.

The intervention that appeared to be more effective in reducing dishonest behavior utilized an honesty oath that mentioned the specific behavior that was expected from participants. The researchers found that using this oath reduced the under-reporting of income by half.

"This intervention resulted in 18.1% underreporting their income to some degree, while the control (so no oath) resulted in 31.3%," said Zickfeld. "In addition, the frequency of fully dishonest participants was nearly cut in half from 9.4% in the control to 4.8% in this most effective treatment. This also reduced tax losses from 21.9% in the control to 11.6% in the most effective intervention."

Overall, the results of this study suggest that oaths that minimize ambiguity about the honest behaviors that people are expected to engage in are most effective in reducing dishonesty. While the content of oaths appeared to influence the outcome of interventions, how people committed to an oath (i.e., by re-writing it or checking a box) did not seem to make a difference.

The researchers also found that committing to an oath just before reporting income appeared to dissuade more participants from being dishonest. On the other hand, they did not identify any personality-related and socio-economic factors that moderated the effects of interventions.

"We did not observe many moderators that changed the findings," said Zickfeld. "Nonetheless, it seems that honesty oaths were most effective for combinations of variables that are typically known to be the most dishonest, namely young males tend to rank lower in the personality trait of honesty. This could be because for other combinations honest responding was rather high, so the oath didn't have a strong effect anymore."

The valuable insight gathered by Zickfeld and his colleagues could inform the development of future interventions aimed at minimizing different dishonest behaviors, including tax evasion and insurance fraud. In the meantime, the researchers have been conducting further studies exploring the impact of honesty oaths over time.

"One interesting question is how effective such oaths are over a longer time frame," added Zickfeld. "For instance, if I take a pledge of honesty when starting a job (as it is the case in the Netherlands with the Dutch bankers' oath in the finance sector) does this also affect my behavior months or even years later?

"Also, if we suggest to implement oaths in many procedures, would people get so used to them that they lose their effectiveness? Or in other words, are they mostly effective in our studies because they are rather uncommon in everyday life?"

More information: Janis H. Zickfeld et al, Effectiveness of ex ante honesty oaths in reducing dishonesty depends on content, Nature Human Behaviour (2024). DOI: 10.1038/s41562-024-02009-0.

Journal information: Nature Human Behaviour