Experimental Design Details
The experiment consists of 4 stages. Subjects' behaviors in Stage 2 and 3 affect their payoffs. At the end of the experiment, the computer randomly decides whether Stage 2 or Stage 3 will be used to determine each subject's payoff.
In Stage 1, I assign subjects in both treatments to different groups based on their preferences for paintings. The procedure in this stage mainly follows Chen and Li’s design (2006) with a few changes to guarantee that the number of subjects in each group is the same.
Six subjects are assigned to two groups based on their reported preference on five pairs of paintings . In each pair, there is one painting made by Paul Klee and one painting made by Wassily Kandinsky. Each subject independently chooses which painting she prefers in each pair without being told the artist of each painting. After all subjects make their decisions, the computer sorts the six subjects based on how many Klee’s paintings they prefer (if there is a tie among multiple subjects, then these subjects’ orders are determined randomly). The first three subjects, who prefer the most Klee’s paintings, are classified into Group Klee. The second three subjects, who prefer relatively less Klee’s paintings, are classified into Group Kandinsky. Subjects in Group Klee are told that all of their group members relatively prefer Klee’s paintings, compared with other subjects. Subjects in Group Kandinsky are privately told that all of their group members relatively prefer Kandinsky’s paintings . Each subject does not receive information about any other subject’s group membership. Groups remain the same for the entire experiment.
In Stage 2, subjects in both treatments participate in a two-round competition in which they answer questions from an established IQ test. All the IQ test questions are selected from Raven’s Standard Progressive Matrices Plus. In the first round of the game, all subjects are assigned into pairs in which the two subjects are from different groups (i.e. one subject is from the Klee group and the other is from the Kandinsky group). Each pair participate in a competition in which they are asked to solve as many IQ test questions as possible within five minutes. At the end of the first round, the subject who correctly solved more problems within each pair wins the first round of the game and receives an $2.50 bonus, while the subject who loses the first round receives $0.
In the second round, each subject is again given five minutes to solve as many IQ test questions as possible. At the end of the second round, the computer calculates the total number of correct answers the three members get in each group. The group with more total correct answers wins the second round and each of the three group members receives an $2.50 bonus, while each of the three group members in the group who loses the second round receives $0.
After the second round, subjects in the Baseline treatment and the TreatInfo treatment receive different information about their performance in Stage 2. Each subject in the Baseline treatment only sees a screen which tells him/her that Stage 2 is finished but receives no information about whether she wins Round 1 or whether her group wins Round 2. They are not told the results until the end of the experiment. Each subject in the TreatInfo treatment, however, is told whether she wins the first round and whether her group win the second round. Each subject does not receive any information about the number of questions she correctly solves or any other subject’s game results throughout the whole experiment.
In Stage 3, I elicit subjects’ engagement in “we-thinking” in both treatments. Each subject is asked to play a dictator game in which she decides how to allocate six tokens between her personal account and her group account. Each token allocated to her personal account is worth $1.00, while each token allocated to her group account is worth $1.50. At the end of this stage, the computer randomly selects one subject in each group to determine the payoffs in her group. If a subject’s decision is randomly selected to determine payment, she gets all the money she allocates to her personal account, and the money she allocates to her group account is evenly shared by the 3 members of her group. In other words, this subject’s payoff equals to $1.00 * the number of tokens she allocates to her personal account + $1.50 / 3 * the number of tokens she allocates to her group account. The two other subjects in her group, whose decisions are not selected, receive payoffs of $1.50 / 3 * the number of tokens she (i.e. the subject in their group whose decision is selected) allocates to her group account.
In Stage 4, subjects in both treatments are asked to answer several 7-point Likert scale questions about to what extent they are attached to their groups and to what extent they feel good about their individual and group performance in Stage 2 (i.e. their self-reported group-esteem and self-esteem).
After subjects in both treatments finish Stage 4, they are shown the last screen which displays all the game results in Stage 2 (i.e. whether they win the first round and whether their groups win the second round), Stage 3 (i.e. whether their decisions are randomly selected and the allocation results), whether their payoffs in Stage 2 or Stage 3 is selected for payment and their final earnings from the experiment.