Experimental Design Details
There are three main treatments – T1, T2, and T3. This is a one-shot game, and we use a between-subjects design. The subjects are randomly divided into two-person groups. Before the experiments start, subjects need to answer some practice questions to make sure they understand the instruction. After completing all the decisions in the experiment, subjects need to fill in a post-experimental survey.
The normative expectations (what subjects think others "ought to" or "should" contribute) are denoted as NE, and the empirical expectations (what subjects think others "will actually" contribute) are denoted as EE. Information is given on these two expectations, based on (at least) two previous sessions (with similar design but elicited the information of EE and NE; see the information below for an example), in a way such that EE and NE are not in conflict. For instance: "Subjects in a previous experiment (i) thought that most others "ought to" or "should" contribute 10 out of 20 units of their Stage 1 income towards the public good, and (ii) most subjects "actually contributed" more than 6 out of 20 units."
T1: Subjects do not receive information on NE or EE.
T2: Subjects do receive information on NE and EE right at the beginning.
T3: Subjects do receive information on NE and EE right at the beginning, but no punishment stage.
We will have 2 sub-treatments (h and l) in both T2 and T3 – T2h, T2l, T3h, and T3l. The difference between the sub-treatments (T2h vs T2l, T3h vs T3l) is that the subjects in the h sub-treatment (T2h and T3h) will be revealed of higher-level information of NE while the l sub-treatment (T2l and T3l) will be revealed of lower-level information of NE. Note that the higher-level and lower-level information of NE are elicited from T1.
Stage 1 (Public goods contribution stage)
Within each treatment, and after the information explained above is (or is not) revealed to the players, play proceeds as follows.
1. Before deciding on their contributions, players guess the contributions of their partners in an incentive-compatible manner (rewarded if the expected contribution is the closest to the actual contribution of the partner).
2. Players receive identical Stage 1 endowment and play a 2-person public good game and make contributions.
3. At the end of Stage 1, the contribution and income of each player are publicly revealed within their own group.
4. Players are asked to reveal the emotions that they are facing. We adapt Molnar et al. (2020) to ask from among a list of the following emotions: Frustration, Anger, Indignation, Shame, Elation, Satisfaction, Dissatisfaction. Subjects can tick as many emotions as they feel. For each chosen emotion they then indicate the strength of the emotional feeling from 1 (a little) to 7 (extremely).
Stage 2 (Punishment Stage)
1. Players receive identical Stage 2 endowment, and they only use the Stage 2 endowment to simultaneously choose the level of costly punishment to inflict on their partner (the player costs one unit to inflict three units of punishment on the partner). Before punishing, they need to guess the possible punishment level their partner may choose.
2. Players are asked to reveal the emotions that they are facing (same as the end of Stage 1).
3. The punishment and income of each player are publicly revealed within their own group.
Subjects know that only one stage from Stage 1 and Stage 2 will be randomly chosen to pay them after the experiment.