Our field experiment
enables this gap to be narrowed, while
still controlling for relevant variables. Second,
the panel structure of the data set allows an
analysis of how people with heterogeneous prosocial
preferences, identified from past behavior,
react to social comparison.