As children, we’re taught the difference between right and wrong, good and bad. Growing up, we internalize these values and hopefully learn how we should behave in situations that challenge our moral values. At first, these lessons seem black and white, but over time we also learn that the world is full of grays: good people sometimes break the rules. Behaving dishonestly has its rewards.
For example, people lie to justify why they were late, peek at a classmate’s test during an exam to obtain a better grade, and fudge their tax filings to earn some extra cash. Researchers have shown that people will cheat, steal, lie or mislead to gain external rewards, like money, but only to the extent that they can preserve their own moral self-image–a balancing act, which researchers Nina Mazar, On Amir, and Dan Ariely coined the theory of self-concept maintenance.
Based on this idea, we started wondering: To what extent is our underlying tendency to act dishonestly shaped by the cultural context in which we operate? Existing research indicates that countries vary in corruption, cultural values, and norms for dishonesty in particular situations, such as plagiarism on college essays. We became interested in the degree to which such cultural differences might be related to ordinary citizens’ underlying tendencies to act dishonestly. In other words, across different countries, how consistent is the tendency to behave slightly dishonestly?
Across different countries, how consistent is the tendency to behave dishonestly?
To find out, we and our co-authors Lars Hornuf, Juan Tafurt, and Dan Ariely conducted a large cross-cultural project that involved collecting data from more than 2,000 individuals in five countries: China, Colombia, Germany, Portugal, and the United States. Although all of these countries are modern, large-scale societies, they differ in geographical region, corruption, and cultural values. In each country, we sampled from two different populations: students in public universities and adults from the general population in coffee shops. This allowed us to compare dishonesty across participant samples within countries, and across participant samples between countries.
All participants completed a die-rolling task, which created a dilemma between following the rules and earning more money by cheating on the task. In addition, we conducted a prediction study with separate samples from those same countries and asked them to estimate the levels of dishonesty in their own country and the other four countries. Therefore, our research allowed us to assess two things: 1) people’s predictions about dishonesty on our experimental task, and whether dishonesty would vary across participants from different countries, and 2) how actual dishonest behavior on the task varied across our participant samples
How the die task works
To examine dishonesty, we used a die task administered on iPads. The task involved rolling a virtual die over 20 repeated trials. On each trial, participants were instructed to mentally select a side of the die (top or bottom) before rolling the die and to remember their choice. After rolling the die, participants viewed the outcome of the roll (the number of dots on the top and the bottom) and were then asked to indicate which side they had chosen prior to rolling.
Participants earned the equivalent of 10 cents for each dot on the reported side, incentivizing them to report the side with more favorable earnings. For example, if a participant mentally selected “top” and then rolled a “two”, they would see two dots depicted on top and five dots depicted on the bottom. That participant would then be faced with the decision of whether to honestly report selecting “top,” or dishonestly report selecting “bottom” in order to earn more money. In this task, it is impossible to distinguish dishonesty from good luck on any individual trial, so a participant’s decision about whether to cheat is never revealed. However, when results are aggregated across the 20 trials for each participant and across all participants in each sample, the average proportion of favorable rolls chosen in a sample can be compared against chance (50 percent) to assess the magnitude of dishonesty.
Before we reveal our results, take a guess at what we found. On what proportion of rolls do you think people reported selecting the more favorable side of the die? To what extent do you think dishonesty varied across our participant samples from different countries? And which country–China, Colombia, Germany, Portugal, or the United States–do you think houses the most dishonest participants?
Predicted Dishonesty
We asked those same questions to a separate sample of participants from our five countries of interest to predict the results of our study. Here is what they predicted we would find:
Prediction #1: When asked to guess the die task results, people predicted moderate levels of dishonesty, estimating that participants would select the more favorable outcomes on 70% of trials.
Prediction #2: People predicted that dishonesty would vary across countries, and their predictions were related to perceptions of corruption and cultural values.
Prediction #3: Predictors showed a home country bias–against their fellow citizens. Predictors thought citizens from their own country would be the most dishonest of the bunch!
Observed Dishonesty
Here is what we actually observed when we administered the die task:
Observation #1: Dishonesty was prevalent, but limited in scope. The average proportion of high earning sides reported in our total sample was 58 percent, and averages for all of our 10 samples were significantly above 50 percent–the expected proportion if participants were completely honest. Only a few participants reported the high earning side on most or all trials, indicating that cheating was driven primarily by many participants inflating their reports by a small amount.
Observation #2: Individuals across the five countries studied were remarkably similar in their tendency to behave dishonestly. When looking at our 10 participant samples, the average proportion of high earning sides reported ranged from 53 percent to 63 percent.
Observation #3: We found that differences within countries–comparing student and public samples–were larger than differences between countries. In every country except one (Portugal), students tested at public universities behaved significantly more dishonest than the general public testing in coffee shops.
In sum, we observed that the tendency to cheat–but only to a limited extent–was robust across five countries. Moreover, people expect much more variation in dishonesty across countries than we actually observed.
Why did students cheat more than the general public?
Contrary to people’s predictions, the most significant differences we observed were within countries, between university students and the general public. Why did students tend to cheat more? There are a few possible explanations: not only were students younger, poorer, and more educated, students and the general public were tested in different settings (university lab settings vs. coffee shop). While we did everything we could to keep the procedure as consistent as possible, it is difficult to know which of these factors explains students cheating more than the public.
What does this study say about our morality?
The news and popular media tend to propagate “bad apple” view of morality–that there are good people and bad people, and that the bad apples are the ones who do bad things. Thus, when we read the news stories of embezzlement, fraud or adultery, we can easily shake our heads, decry the current state of affairs, and relish our own “good apple” status.
Another view of morality, embedded in classical economics, is the “rational actor” model. According to this framework, we are only honest in order to avoid potential punishment. The “rational actor” view is often traced back to Adam Smith, the forebear of traditional economics, based on his classic volume Wealth of Nations (Ironically, in an earlier volume, The Theory of Moral Sentiments, Adam Smith put forward the notion that we are not just selfish and calculating.)
Our research suggests an alternative to each of these notions. Building on the theory of self-concept maintenance, our findings suggest that the tendency cheat–but only to a limited degree–is robust across countries (at least, across the five countries studied). Around the world, it appears that people are mostly rule abiding–but will push the envelope to the extent that they can still view themselves as honest, upright people.
