Ethics declaration
All experiments reported in this paper were conducted according to the rules and regulations for good academic practice of the University of Heidelberg (https://www.uni-heidelberg.de/en/university/about-the-university/good-academic-practice), defined by the university’s Commission for the Safeguarding of Good Academic Practice (Kommission zur Sicherung guter wissenschaftlicher Praxis und zum Umgang mit Fehlverhalten in der Wissenschaft, Universitätsverwaltung, Dezernat Recht und Gremien, Seminarstr. 2, 69117 Heidelberg, Germany). The Commission waives the need for ethics approval for non-medical studies like the surveys presented in the current paper (https://www.uni-heidelberg.de/en/documents/rules-for-safeguarding-good-academic-practice-and-handling-academic-misconduct/download). All participants were above 18 years old and provided explicit written consent to participating on the first screen of the survey.
Study 1
Sample and design
We conduct our first study with U.S. participants on the Amazon Mechanical Turk (MTurk) platform and Chinese participants of the Shandong General Social Survey (SGSS). MTurk has been increasingly used as a platform to conduct social science surveys, as it offers convenient access to a highly diverse population in the U.S.17,18,19,20. The SGSS aims to serve as a representative sample of the Shandong Province in China, covering 203 communities in the province (with a population of about 100 million).
For the U.S. sample, we posted our study on MTurk including a short description of the task, the task requirement, as well as the expected payment for completing the task. After accepting the task, participants were redirected to our survey that is programmed in oTree21. We set a restriction that allows only U.S. residents to enter the survey. We also added a CAPTCHA in the welcome page to prevent bots from entering our study. A total of 2,110 respondents finished our survey, of which 440 were randomly assigned to the Bank condition, 413 to the petty welfare/tax fraud condition, 420 to the severe welfare/tax fraud condition, 432 to the emergency lane condition, and 405 to the endangered animal condition. About half of the participants in each condition were confronted with unethical behavior of a high-SES and the other half with that of a low-SES individual. For the welfare/tax conditions, the high-SES behavior regarded tax fraud, the low-SES behavior regarded welfare fraud. For each scenario, we also separately present results for respondents whose household income is higher or lower than the average, according to self-report with the neutral category being average income. For the relatively high (low) household income respondents, there were 80 (127) in the bank condition; 76 (116) in the petty welfare/tax fraud condition; 82 (136) in the severe welfare/tax fraud condition, 77 (127) in the emergency lane condition, and 89 (129) in the endangered animal condition.
For the SGSS, our vignettes were included as one of the sections in the SGSS’s general surveys. The surveys were conducted in person at the respondents’ home by 182 trained investigators in 2019. In total 923 individuals participated in our survey: 185 of them were randomly assigned to the Bank condition, 184 to the petty welfare/tax fraud condition, 176 to the severe welfare/tax fraud condition, 193 to the emergency lane condition, and 185 to the endangered animal condition. Again, we also present results by household income of the respondents for each scenario. For the relatively high (low) household income respondents, there were 17 (68) in the bank condition, 22 (57) in the petty welfare/tax fraud condition; 14 (65) in the severe welfare/tax fraud condition, 18 (62) in the emergency lane condition, and 22 (54) in the endangered animal condition.
Vignettes
In what follows, we list the vignettes used in the MTurk study. The SGSS study uses exactly the same wording, translated to Chinese and using Chinese Yuan instead of dollar amounts (using the exchange rate of $1 = 7.2 at the time of survey, and rounded to round Yuan amounts). The translated scripts are available upon request.
Condition: bank
Consider the following setting: A low-skilled worker [successful businessman] who does not check his private bank account very regularly discovers that he incorrectly received a payment of $700 some weeks ago. Given that the transaction had not yet been discovered and reversed, he decides to not report the payment, hoping that it may never be discovered.
Conditions: petty [severe] welfare/tax fraud
For low-SES: Consider the following setting: A low-skilled worker lost his job some time ago. He, therefore, started an eBay store. He has been earning profits that are similar to his previous wage. Yet, he did not report his new income to the government and kept collecting subsistence allowance that accumulated to $1000 [$10,000]. Since the government has not discovered the fact the he is ineligible for the allowance, he decides not to voluntarily report his income earned from the online store.
For high-SES: Consider the following setting: A successful businessman has some overseas business that he has not reported to the tax authority. From the activities abroad, he has been earning profits that he should pay tax for, with the tax owed accumulated to $1000 [$700,000]. Given that the government has not discovered his overseas activities, he decides not to voluntarily report his income.
Condition: emergency lane
Consider the following setting: A low-skilled worker who has to show up at a construction site [A successful businessman who has a business meeting] outside of town is stuck in a traffic jam caused by an accident. Because he wants to show up on time, he decides to pass the accident site on the emergency lane, which is strictly reserved for ambulance and police. Because emergency services are busy at the accident site, he is not stopped and punished, and arrives at the construction site (the meeting) on time.
Condition: endangered animal
Consider the following setting: A low-skilled worker has recently been sent to an African country by his company. [A successful businessman has recently travelled for business to an African country.] He finds out that he can buy rhino horn powder on the black market there, which is considered an important cure-all by his family. Since he knows that it is forbidden to buy rhino horn in his home country, as rhinos are classified as an endangered animal, he uses this chance to buy this powder to take home.
All conditions share the same questionnaire:
Study 2
Sample and design
The study recruited 2243 U.S. participants through MTurk, following the same recruitment procedure as in Study 1. We described a dictator game to the respondents, in which an allocator has to distribute $100 between herself and another person, the recipient. We informed them about the respective incomes of the allocator and the recipient. Respondents were randomly assigned to one of 28 conditions that differ by allocator and recipient income, with about 80 respondents per condition. We have consulted labor market data from the U.S. Bureau of Labor Statistics and the United States Consensus Bureau for the income distribution in the United States. The allocator’s income in our experiment ranged from $10,000 per year to $1,000,000 per year. As the vast majority (roughly 87%) of the U.S. population earned less than $100,000 per year according to the 2019 U.S. Consensus Bureau22, we adopted fine steps of $10,000 for annual income below $100,000. We then included fewer steps towards the annual income of one million dollars, using $200,000, $500,000, $800,000, and $1,000,000. For the recipient’s annual income, we employ two levels: $40,000 and $80,000. According to the Bureau of Labor Statistics23, these correspond to the salaries for a typical individual with a high school diploma and a Master’s degree respectively.
For each condition, we asked our respondents to indicate what amount of the $100 endowment the allocator should transfer to the recipient, and to provide some motivations for their answer (it was possible to leave the entry for the motivation empty). The background information regarding the income of the allocator and the recipient is available to the respondents throughout the decision process. Comprehension questions were included to make sure respondents only proceed to the decision stage if they understood the key parameters of the game.
Instructions
The instructions of the survey experiment read as follows.
Please consider the following situation carefully and let us know what you think is the most appropriate behavior of the person in this context.
Suppose that a group of researchers studying individuals’ social behavior has recruited a person with an annual gross income of $Allocator’s Income [$10,000–100,000 (with a 10 K step), $200,000, $500,000, $800,000, $1,000,000] to participate in a study. Let us call this person the allocator.
The allocator is randomly matched to another person with an annual income of $40,000 [$80,000]. This person, whom we call the recipient, cannot make any decisions. The income of the recipient is clearly communicated to the allocator and vice versa. Both people are completely anonymous to each other and to the researchers.
The researchers give the allocator $100, and the recipient matched to him/her has not been given any money. The allocator has to decide now, how he would like to allocate the $100 between him/herself and the recipient. In particular, the allocator can send any amount between $0 and $100 to the recipient. The recipient is told that the decision-maker received $100, the choice the allocator had, and the decision made.
Comprehension questions
We would like to make sure that the key information has been clearly delivered to you. Please answer the following questions below before we proceed to the next stage.
-
1.
What is the allocator’s annual income in USD? $_____ (correct answer depends on treatment)
-
2.
What is the recipient’s annual income in USD? $_____ (correct answer depends on treatment)
-
3.
How much money the allocator can distribute in USD? $ ____ (100).
Main questions
We are interested in your opinion now.
What amount, between $0 and $100, should the allocator send to the recipient:
$_____ (enter a number between 0 and 100).
Motivate why you think this is an appropriate amount:
[Participants are provided with the description of the scenario throughout the decision process]
Scoring of respondent motivations
Respondents provided open-ended motivations for the allocations they indicated as appropriate. For each respondent, we construct a variable Motivation, coded as 1 if the motivation provided by the respondent only contains the keywords related to inequality (and thus nothing related to fairness concerns). We use the following keywords to identify concerns over inequality: inequality, unequal, rich, poor, recipient needs, help, more, less, poverty, generous. If a respondent expressed both fairness and inequality concerns, we code the variable as 0. If the motivation text contains only “fair” as the keyword, we code it as − 1. Keywords are considered relevant as long as the string of the motivation text contains the keywords. For instance, if a respondent wrote the word “richer” instead of “rich”, we still count her as expressing inequality concerns. Respondents who expressed neither fairness nor inequality concerns are excluded from the motivation analysis.
Study 3
Sample and design
The study was conducted with 837 U.S. participants through MTurk, following the same recruitment procedure as in Study 1. Respondents were randomly assigned to one of four conditions that elicit expectations regarding the previously collected behavior of a sample of millionaires’ in a simple allocation game conducted by Smeets and coauthors7: 197 participated in the empirical expectation with millionaire recipient condition; 205 participated in the empirical expectation with low-income recipient condition; 216 participated in the normative expectation with millionaire recipient condition; and 219 participated in the normative expectation with low-income recipient condition.
Instructions
The instructions of the survey experiment read as follows.
A recent study investigated the social behavior of wealthy individuals in Europe, who have more than 1 million euros in their bank account. These wealthy individuals were given 100 euros by the researcher, and were asked how much of the 100 euros they wanted to share with another person. Whatever they decided to give away was truly given to another person, and they kept the remaining amount. That is, the sharing decisions were implemented for real.
Importantly, the wealthy individuals making the sharing decision knew that the recipient paired with them was also wealthy, having more than 1 million euros in his/her bank account [was a low-income person who earned less than 12,500 euros per year before taxes].
All interactions were strictly anonymous and money payments and transfers were arranged by the researchers.
Many studies have looked at how people share an amount of money given to them in an experiment with an anonymous other person. Most of these studies involved neither particularly wealthy nor particularly low-income participants. The finding from these studies was that people on average give away about 28 percent of the given money to the other person.
Condition: empirical expectations, millionaire recipients
Do you think that the wealthy individuals in the study on average gave more or less than the typical share of 28 percent to the other person, when sharing with another wealthy person?
Condition: empirical expectations, low-income recipients
Do you think that the wealthy individuals on average gave more or less than the typical share of 28 percent to the other person, when sharing with a low-income person?
Condition: normative expectations, millionaire recipients
We are interested in what you consider an appropriate behavior by the wealthy individuals in this situation.
Should the wealthy individuals give more or less than the typical share of 28 percent to the other person, when sharing with another wealthy person?
Condition: normative expectations, low-income recipients
We are interested in what you consider an appropriate behavior by the wealthy individuals in this situation.
Should the wealthy individuals give more or less than the typical share of 28 percent to the other person, when sharing with a low-income person?
Next, conditional on their first answer, the participants answered a question regarding the precise amount they have in mind:
Condition: empirical expectations
You think the wealthy individuals gave more {less} to the other person than typically observed. How much of 100 euros do you think they gave on average to the other wealthy person [the low-income person]? Respondent types in number between 28 and 100 {between 0 and 28}.
Condition: normative expectations
You think the wealthy individuals should give more to the other person than typically observed. How much of the 100 euros do you think they should give to the other wealthy person [the low-income person]? Respondent types in number between 28 and 100 {between 0 and 28}.