Intervention(s)
In the experiment, participants, referred to as “spectators,” decided whether to redistribute income between two individuals, termed “workers.” These workers consisted of college students. In addition, one of the student’s parents was also recruited to participate in the study. The students engaged in the experiment within a college laboratory without internet access. At the same time, their parents participated concurrently from a separate location. Communication between students and their parents was prohibited to ensure the independence of their actions. We verify the family relationships by asking parents to provide their child’s college ID number along with their own official document, which we subsequently verified against the student’s enrollment records.
Spectators were randomly assigned to one of 6 treatments: the luck treatment (L), the merit treatment (M), the luck-parent treatment (LP), the merit-parent treatment (MP), the luck-parent-no-choice treatment (LPN), and the merit-parent-no-choice treatment (MP). In the L and M treatments, workers’ incomes were determined by a random lottery or by their performance, yielding 0 or 6 points. In the LP treatment,
parents’ incomes were assigned similarly by lottery, with possible earnings of 0 or 15 points. Parents could forego 6 points so their child could receive 6 points. Thus, in the LP treatment, the final income for workers depended on the transfer decisions made by winning parents, who knew their decisions would remain anonymous to their children. The MP treatment followed the same setup as LP, except that the
parent’s income was based on their performance in an assignment. In LPN and MPN, parents' income is forced to be transferred without their choice.
Spectators were made aware of the conditions under which the workers and their parents operated. For instance, they knew that workers’ incomes would be determined based on predefined payment rules, which remained unknown to the workers before and after the assignment. This setup ensured that, across treatments, workers were perceived identically, except for the rules determining earnings. We employed a contingent response method to assess the spectators’ fairness preference, requiring them to make redistribution decisions among the workers for all earning outcomes of paired workers. Our analysis, however, focuses primarily on scenarios where income distribution was unequal, with one worker receiving 6 points and another receiving none