Skip to main content


Social Comparisons and Contributions to Online Communities: A Field Experiment at MovieLens


With the increasing popularity of the Internet, information technology is changing the way we interact, entertain, communicate and consume. In online communities, groups of people meet to share information, discuss mutual interests, play games and carry out business. However, despite the popularity of online communities, many such communities fail due to non-participation and under-contribution. For example, many people use Wikipedia, but few contribute new entries.

Under-contribution is a problem even in active and successful online communities. For example, in MovieLens ( ), an online movie recommendation website that invites users to rate movies and, in return, makes personalized recommendations and predictions for movies the user has not already rated, under-contribution is common. More than 22-percent of the movies listed on the site have fewer than 40 ratings, so few that the software cannot make accurate predictions about which users would like these movies. Similarly, Eureka, a Xerox Corporation online information sharing system, which enables its 20,000 worldwide customer service engineers to share repair tips, also suffers from under-contribution. While many service engineers download machine repair tips from Eureka, only an estimated 20-percent have submitted a validated tip to the system.

Many online communities are populated with peripheral users, who observe the community and use the content created by others without contributing to the community content or discussions. In 2000, the P2P file sharing site Gnutella saw 25-percent of users share 98-percent of the content while 66-percent of users share nothing (Eytan Adar and Bernardo Huberman 2000). By 2005, 85-percent of users shared nothing (Daniel Hughes, Geoff Coulson, and James Walkerdine 2005). Thus, a key challenge to the online community designer is to motivate the peripheral participants to become active contributors, and the core participants to sustain and improve their contributions.

In Chen, Harper, Konstan and Li (American Economic Review 2010), two economists team up with two computer scientists to mine social science theories to solve the socio-technical problem. They conducted a randomized field experiment to explore the use of social information and social comparison theory to increase contributions to Movielens. After sending out personalized email newsletters to nearly 400 MovieLens users which compare each user?s contributions with the average user, they find that, after receiving behavioral information about the median user’s total number of movie ratings, users below the median demonstrate a 530-percent increase in the number of monthly movie ratings, while those above the median decrease their ratings by 62-percent. When given outcome information about the average user’s net benefit score, above-average users mainly engage in activities that help others, such as updating the movie database. These findings suggest that effective personalized social information can increase the level of public goods provision.

Address Goals

This research applies social science theory to solve a socio-technical problem prevalent in many online communities. It contributes to the new emerging field of user-generated content as well as putting social science theories to test in the online setting. By using personalized social information, we achieve a 530% increase of movie ratings for the laggards, and an overall increase in quality.

The two doctoral students trained on this project, Harper and Li, have successfully defended their thesis, and started their new jobs as entrepreneur for a technology firm (Harper) and assistant professor of Economics at UT-Dallas (Li).