Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Economists are increasingly studying the market for charitable giving, including behavioral aspects. In the "identifiable victim effect" people donate more to help identified individuals than groups. However, individuals from socially low-ranking groups may not prompt similar charity. The continuing depth of caste-based prejudice in contemporary India is widely debated. We report three randomized experiments conducted with computer-using Indian participants. First, we conducted a survey experiment with detailed information about participants. Second, we experimentally varied 56,000 real advertisements for charitable donations shown online. Third, we conducted an Internet choice experiment in which participants allocated real money. In all three studies an identification treatment is crossed with population group membership of the recipient. We indicate population group membership of identified recipients subtly, using names that connote caste; participants' understanding of these names is verified with a separate experimental sample. In all studies, we find an identifiable victim effect for generic Indian and high-caste recipients, which is absent or even reversed for low-caste recipients. Caste still matters among the Internet-using, English-speaking population of our participants.