Doing good in the digital world

Though digital interactions between people have become more commonplace and sophisticated, behavior in digital settings remains underresearched. A distinctive feature of the digital world is the ability to calibrate or withhold one’s identifier: a person can be identified by a string of letters, an avatar, their real name, or even nothing at all. Moreover, that digital identifiers allow a person to mask their physical identity also makes it difficult to attribute digital actions to a physical person, even when the actions are observed. I embed these two features in a laboratory experiment where subjects play a finitely repeated, linear public goods game. Treated subjects are identified in one of three ways—by their photograph, by a random number, or by a self-designed cartoon avatar—and their individual choices are revealed and either attributed to, or decoupled from, their identifier. In line with the previous literature, identifying subjects and increasing the precision of attribution increases contributions relative to a baseline condition without identifiers or revealed individual choices. Remarkably, the treatment effect is robust to less precise identifiers and attribution: contributions increase significantly even when subjects are identified by numbers and their individual contributions are revealed, but decoupled from, those numbers.