Lying in Reciprocal Appraisals

There are two interventions, hypothesized to lead to varying levels of dishonesty. The main intervention (captured by the formal model of reciprocity) concerns the timing of play (simultaneous or reciprocal). The secondary intervention concerns the 'grouping', namely whether participants are placed together in groups that assess each other or not.

2023-09-01

2023-12-01

The primary outcome will be the distribution of reports of the rolls of the dice, which will be compared to the expected distribution stemming from a series of unbiased rolls of the dice. This distribution of reports will be also compared across experimental conditions. Differences in the distribution will also be measured focusing on means and the percentages attributed to high reports (4s and 5s). The primary hypotheses concern comparisons between Control Condition 1 and Treatment 1, corresponding to the predictions of our formal model.

The main hypothesis of our study concerns differences across the reported distributions of participants in Control Condition 1 and the first movers in Treatment 1, as per the predictions of our formal model.

Measures of the distribution of the reports of the dice rolls (as in the discussion of primary outcomes above), but focusing on the ancillary conditions of our experiments, namely Control Condition 2 and Treatment 2.

Our ancillary hypotheses are that belonging to a group matters for reporting behavior (reporting behavior across the two control conditions differs) and that 'generalized reciprocity' matters in that reporting behavior of 'B players' in Treatment 2 will depend on the reports they received from 'A players'.

The study is based on Fischbacher and Föllmi-Heusi's (2013) die-rolling paradigm. Participants are instructed to roll a dice and report the result, which determines the payments of other participants. There will be four experimental conditions, all of which shall be one-shot. The experiments will be conducted online using a widely used platform.

The four experimental conditions are as follows:

Simultaneous Non-reciprocal: in a large group, everyone observes a dice-roll and reports a number (based on that roll) that determines someone else’s payoff. Participants do not know who this “someone else” is. They do know that this “someone else” is not the person who reports for them.

Simultaneous and Reciprocal: Subjects are randomly paired into groups of two. In a group with subject A and subject B, A observes a dice-roll and reports a number that determines B’s payoff, and in the meantime, B observes another dice-roll and reports a number that determines A’s payoff. Participants know this matching. Both of them report for the other person before observing the number reported for them, i.e., they move simultaneously.

Sequential and Reciprocal: Subjects are randomly paired into groups of two. In a group with subject A and subject B, A is the first mover, observing a dice-roll and reporting a number that determines B’s payoff. The second mover B observes A’s report, as well as a dice-roll and then reports a number determining A's payoffs.

Sequential and Non-Reciprocal (Generalized-Reciprocal): Subjects are randomly paired into groups of three, with subjects A, B and C. A is the first mover, observing a dice-roll and reporting a number that determines B’s payoff. Second mover B observes A’s report, as well as a dice-roll and then reports a number for C.

Randomization performed by online participant pool service provider (Prolific).

Individual participant

No

NA

1050 online participants

Control Condition 1(Simultaneous Non-reciprocal): 150 individuals
Control Condition 2 (Simultaneous and Reciprocal): 150 individuals
Treatment 1 (Sequential and Reciprocal): 300 individuals
Treatment 2 (Sequential and Non-Reciprocal): 450 individuals

SEE ATTACHED FILE