The degree of acquaintance among group members (friends or strangers) and the opportunity for group members to observe each other's actions (anonymous or non-anonymous) were manipulated in a 2 x 2 factorial design, to determine the amount of aversive stimulation that would be administered to a target person. When group members could not observe each other, friends and strangers administered aversive stimulation equally often. When others' actions were observable, group members administered the aversive stimulus more often. Furthermore, groups of friends administered it more often than strangers. Differential uncertainty and/or anxiety reduction via social support can explain these results. (Author)