The Rolling-Dice Experiment

In a classic setup by Fischbacher et al., participants rolled a die in private and reported the outcome for a monetary reward — without any way to verify their honesty. Surprisingly, only 20 % maximized their gain by lying completely, while 39 % told the full truth.

Our version takes this experiment further: participants are recorded during the die roll, and the outcomes are verifiable. They are informed that their data will be used for research, but the experimenters themselves never see the videos — ensuring unbiased data collection.

Participants roll a standard six-sided die once. They earn 1 € times the number shown, except if they roll a 6 — which earns nothing. The full dataset includes 100 participants.

Note: This Rolling-Dice Experiment is a preliminary study for our upcoming GAME dataset. Due to privacy and ethical constraints, the raw video and audio recordings from this study cannot be published. Only anonymized outcome data (e.g. dice results) may be shared for illustrative purposes.