Case Studies

Cheating

Question: We know that dishonesty is seen in societies worldwide, but to what extent is there variation between countries in the amount of dishonesty? This research tested whether what country someone is from affects levels of dishonesty through the use of an abstract, novel task.

Hypothesis: We predicted that there would be differences in levels of dishonesty between countries. Specifically, we believed that in countries where corruption is high there would be a greater norm of dishonesty being acceptable.

Experiment: Participants in several different countries (with varying levels of corruption) were given a digital dice rolling task on an iPad. In this task, participants are provided with an opportunity and a financial incentive to cheat. Participants rolled a die and were paid based on the number of dots they rolled. At the beginning, participants are instructed to choose either the top or bottom of the die before rolling without revealing their choice to anyone else. Participants “roll” the die, see an image which displays both the top/bottom, and indicate which side they have chosen. Thus, on any roll where the participant originally decides to select the unfavorable side, they have the opportunity to cheat and claim to have originally chosen the higher value side. While it was not possible to tell if any individual participant was lying, it was possible to determine the statistical likelihood of dishonesty across the whole sample. With 20 trials per subject and 200 subjects per country, the sample size of rolls was sufficiently high that there was a low probability of an honest sample deviating significantly from choosing the favorable side 50% of the time. Thus, by comparing the observed portion of favorable rolls with the expected 50% we obtained an estimate of how honest subjects were in each country.

Results: For all 10 participant samples, the mean proportion of favorable reports on the die task was significantly above chance. However, our results suggest that country-level cultural variables have limited influence on generalized dishonesty.

The mean proportion of favorable reports on the dice task was significantly above chance in all of the samples. For example, the distribution of outcomes in American students is shown below.

cloth

However, our results suggest that country-level cultural variables have limited influence on generalized dishonesty, as shown in the graph below by the similar results across countries.  

cloth

Application: At a policy level, these findings suggest that programs aimed at promoting general morality are unlikely to have lasting effectiveness. Instead, programs aimed at establishing honesty norms for specific behaviors may be more effective. While culture may influence norms for dishonesty in specific situations, it seems it has limited impact on generalized dishonesty.