Feedback in Homogenous Groups

Public rank feedback increases productivity and learning effort. Often the comparison group of peers is the classroom. n digital environments, however, group compilation is up to the experimenter. In a field experiment we vary the probability that learning leads to a rank change by manipulating the group composition. 13,800 college-level learners are allocated to three treatment arms and feedback groups of ten. Every month groups are reset, and participants reassigned, but remain in their treatment arm. Participants in the control arm are assigned to feedback groups at random which leads to a high within-group skill heterogeneity. The treatment arm consists of feedback groups with students with a similar learning history and therefore exhibits low within-group skill heterogeneity. The third arm does not receive feedback. We expect more interaction with the learning platform in the treatment group and higher scores in a final standardized test a subgroup of learners will take.