Inequality with unequal opportunity

In all treatments, workers complete a simple task (a multiple-choice quiz), and their initial payment rewards are determined by the specific rules of the treatment, described in detail below. The spectator is provided with a summary which describes the situation faced by workers and the initial payment reward result, and then decides on the final payment for the pair of workers. Workers are informed during their part of the task that their task performance will only determine their reward temporarily, but that they must wait for a spectator to make the final reward allocation decision between them and their paired worker. Regardless of the spectator’s decision, each worker will receive at least $2 for completing their part of the task, and thus each worker will be compensated reasonably for their time spent (approximately 10 to 15 minutes).

We implement four different treatments that allow us to study how random opportunity (both ex-ante and ex-post) affects people’s acceptance of inequality between the two workers. There are two additional treatments in which spectators can decide whether or not to perform a number checking task to reveal the conditions under which workers received their initial reward allocations.

2022-01-29

2022-06-22

The main outcome variable is the amount redistributed spectators to the workers in different treatments, r_i. The different redistributive amounts will lead to different implemented inequality levels, that is, the absolute value of the difference in payment to the two workers divided by total payment j_i=(|6-2r_i |)/6.

In all treatments, workers complete a simple task (a multiple-choice quiz), and their initial payment rewards are determined by the specific rules of the treatment, described in detail below. The spectator is provided with a summary which describes the situation faced by workers and the initial payment reward result, and then decides on the final payment for the pair of workers. Workers are informed during their part of the task that their task performance will only determine their reward temporarily, but that they must wait for a spectator to make the final reward allocation decision between them and their paired worker. Regardless of the spectator’s decision, each worker will receive at least $2 for completing their part of the task, and thus each worker will be compensated reasonably for their time spent (approximately 10 to 15 minutes).

We implement four different treatments that allow us to study how random opportunity (both ex-ante and ex-post) affects people’s acceptance of inequality between the two workers. There are two additional treatments in which spectators can decide whether or not to perform a number checking task to reveal the conditions under which workers received their initial reward allocations. Our study has a between-subject design, that is, to compare spectators’ redistributive behavior in different treatments, thus each recruited participant only participates in *one* of the treatments described below.

• Treatment 1 (Luck): Both workers first study identical reading materials (one to two short paragraphs), then complete the same set of quiz pertaining to the same topic as the materials. However, they are informed in advance that the initial assignment of earnings between the two workers will be determined by a lottery (one of them will be randomly selected to receive the entire initial reward, $6).
• Treatment 2 (Merit): Both workers first study identical reading materials (one to two short paragraphs), then complete the same set of quiz pertaining to the same topic as the reading material. The initial assignment of earnings will be determined by merit (i.e. whichever worker has a higher number of correct answers in the quiz will receive the entire initial reward, $6).
• Treatment 3 (Random-Opportunity): Two workers, unbeknownst to themselves, receive different reading materials; one is highly relevant to the knowledge quiz while the other is of lower relevance (both reading materials are of similar length, one to two short paragraphs). They then complete the same set of quiz pertaining to the same topic as the highly relevant reading material. The spectators will be informed about which worker receives the highly relevant reading material, and that the worker with the higher number of correct answers in the quiz receives the entire initial reward, $6 (we expect this worker will almost always be the one who receives the highly relevant materials).
• Treatment 4 (Random-Employment): Both workers first study identical reading material (one to two short paragraphs). One randomly determined worker will have the chance to answer 15 quiz questions, while the other worker has the chance to answer 4 quiz questions. The spectators will be informed which worker has received the version with 15 quiz questions and that the worker with the higher number of correct answers in the quiz receives the entire initial reward, $6 we expect this worker will almost always be the one who receives received the version with 15 quiz questions).
• Treatment 5 (Info-rOpportunity): Two workers, unbeknownst to themselves, receive different reading material; one is highly relevant to the knowledge quiz while the other is of lower relevance (both tutorials are of similar length, one to two short paragraphs). They then complete the same set of quiz pertaining to the same topic as the highly relevant reading material. Spectators know the assignment of earnings has been determined by the number of correct answers in the quiz, but they need to complete a number checking task to unveil whether the two workers actually received the same reading materials, or reading material of different relevance levels.
• Treatment 6 (Info-rEmployment): Both workers first study identical reading material (one to two short paragraphs). One randomly determined worker will have the chance to answer 15 quiz questions, while the other worker has the chance to answer 4 quiz questions. Spectators know the assignment of earnings has been determined by the number of correct answers in the quiz, but they need to complete a number checking task to reveal whether the two workers had access to the same number of quiz questions, or different numbers of quiz questions.

Randomization done by a computer in an online survey

Individual

No

About 2100 respondents

About 2100 respondents

We plan to recruit 210 spectators plus 140 workers for each treatment. The sample ought to be representative of the US general population in terms of gender, age, education, and region.