Altruism, reputation and noise in linear public goods experiments

We report an experiment using a design that permits the direct measurement of individual decision rules in voluntary contribution games. We estimate the distribution of altruism in our subjects and find that observed 'overcontribution' is attributable to a combination of random variation in behavior and a few altruistic players. We also employ Andreoni's partners/strangers design to measure reputation effects. The only difference observed is that the strangers treatment produces slightly more random variation in behavior. Our results explain some anomalies about contribution rates, and support past findings that reputation-building plays a minor role in such experiments.