An Experiment on Information Avoidance with Respect to Intention

2020-02-11

2020-02-28

• frequency of player 1 choosing the high investment
• average donation by player 2 (conditional on high outcome and high investment and low outcome and high investment)
• compare average donation by player 2 between different investment decisions conditional on outcome (only in full info treatment)
• compare average donation by player 2 between different outcomes conditional on investment decisions (only in full info treatment)
• compare frequency of player 2 choosing to learn player 1’s decision between decisions where outcome is low and decisions where outcome is high
• We also elicit beliefs about choices of player 1 and risk preferences and will use these in a regression-based approach to understand decisions of player 2.

• average donation by player 2 (between treatments)
• compare the average informed (self-selected vs. exogenous) donation by player 2 conditional on high outcome and high investment
• compare the average informed (self-selected vs. exogenous) donation by player 2 conditional on low outcome and high investment
• compare average donation by player 2 given a high outcome conditional on player 1's investment when informed vs. uninformed (only in hidden information treatment)

We seek to investigate how decision makers take the intention of another decision maker and the outcome of her choice into account when choosing a reciprocating action. We vary if decision makers observe their partner’s action or only the outcome of the decision. We ask if decision makers are willing to reveal the action of their partner –his intention –or if they prefer to remain uninformed. If decision makers feel obliged to reward good intentions, they might prefer to remain ignorant about their partner’s action and exploit their willful ignorance by acting selfishly. We further explore how decision makers incorporate their partner’s action and the outcome of her actions into their own decisions when both are observable.

General outline of the experiment
In the experiment, subjects take two roles, Player 1 and 2, which they will keep throughout the experiment.Players interact in anonymous groups always consisting of one player of each type. Groups are formed randomly at the beginning of each round. In each round, Player 1 receives an endowment of 30 points from which he can invest a fraction in a project. The return of the project is split in equal shares between both players. Player 1 decides between two risky investment options. If the investment succeeds, the return is high (50 points) and if the investment fails, the return is low (10 points). One of the options has a higher success probability than the other (75% vs. 25%), but at the same time it is more expensive for Player 1(25 vs. 5 points). After Player 1 has made her decision, Player 2 decides how much of his endowment of 30 points to give to Player 1. This interaction is repeated with newly formed pairs in the next round. At the end, one of the rounds is randomly chosen for payment. In a between-subjects design, we vary the availability of information on the first player’s investment choice. In the full information treatment, the second player observes both Player 1’s investment decision and the outcome of the investment, whereas in the hidden information treatment, only the outcome of the investment is directly observable. Player 2 has the option to reveal Player 1’s decision, but he can also decide on the split of his endowment without knowing which option Player 1 has chosen. In the full information treatment, Player 2 is asked for his choice for a given decision of Player 1 using the strategy method. At the end of the experiment, we elicit the beliefs with respect to Player 1’s behavior as well as the subjects’ risk preferences. Only one of these tasks is relevant for payment. Which one it is will be determined randomly.

Questionnaire
Sessions will be closed with a questionnaire on demographics (age, gender, field of study, nationality) and a few questions with respect to the subjects’ political and honesty preferences. Beyond that, no personal data is collected,such that subjects' anonymity will be maintained throughout the experiment.

Subjects from the subject pool sign up for sessions without knowing which treatment is run. In this sense, allocation to treatment is random.

experimental sessions

Yes

12 experimental sessions in Berlin and 8 experimental sessions in Potsdam

264 subjects in Berlin and 160 subjects in Potsdam

Berlin: 3 sessions for full information and 9 sessions for hidden information
Potsdam: 2 sessions for full information and 6 sessions for hidden information
In case of an unexpectedly high number of no-shows, we will run additional sessions in Potsdam and/or Berlin to reach the planned number of observations.