Responsibility Attributed Under Institutional Variation: A Robustness-Oriented Replication in Sequential Group Decisions

Reliable cumulative knowledge requires that behavioral findings remain stable not only across statistical replications but also under theoretically meaningful institutional variations. This paper investigates the robustness of responsibility attribution in sequential group decisions, 10 using the experimental paradigm of Bartling et al. (2015) for the broader meta-scientific question of whether established experimental results hold under controlled design changes. We systematically extend the sequential dictator game with punishment along two institutional dimensions: (1) the presence of a default option (equal vs. unequal allocation preselected) and (2) the group-formation mechanism (random assignment vs. assignment based on university 15 entrance grades). This design allows us to assess whether the original finding – that pivotal decision-makers receive the highest punishment – is robust to institutional variation. We cannot reproduce the dominance of pivotality as the primary driver of punishment. Instead, being the first decision-maker to choose the unequal allocation (initiation) emerges as the stronger punishment motive in our extended setting. Default structures affect punishment directed at 20 pivotal decision-makers, but these effects disappear once punishment motives are jointly controlled. The group-formation mechanism does not significantly influence punishment behavior. These findings demonstrate that responsibility attribution is more context-sensitive than the original study suggests, and that seemingly minor procedural variations can meaningfully alter observed punishment patterns. Therefore, our study illustrates the value of robustness-oriented 25 replication as a complement to exact replication for assessing the reliability of experimental evidence.