Collaborative Research: Instrumenting Behaviors and Attitudes in Virtual Worlds

University Of Illinois At Urbana-Champaign, Urbana IL

Investigators

Abstract

This project is a pioneering effort to collect data about Massively Multiplayer Online game players, their actions, attitudes, social networks, and economic activities. People increasingly conduct major parts of their lives in virtual worlds: online spaces where large numbers of users are able to congregate, interact, play, and socialize. Most of these worlds are built as competitive games, and are termed Massively Multiplayer Online games (MMOs). Best estimates indicate in excess of five million North Americans now interact in these worlds, with more than 20 million worldwide. Yet while the serious study of virtual worlds has begun in earnest, it is hamstrung by the absence of systematic data. No research project has been conducted with a representative sample, access to server-side data, or the cooperation of game administrators. However, this project has secured the cooperation of a large game developer, which has agreed to release its in-game data as well as help coordinate a survey instrument of its players. The resulting data set will thus be the first comprehensive study of virtual world behaviors, and will contain a rare combination of both unobtrusive behavioral data with standard attitudinal, demographic and psychographic measures. This data set must be properly collected, cleaned and organized in order to facilitate access by a wide range of researchers. Thus the initial effort will involve programming innovations, cleaning and hosting tasks, as well as the creation of new algorithms for several social science tests. The first three areas to be tested will be social networks, social capital, and economic activity. In each area, the dataset will be a testbed for examining theories in ways that were previously impossible, while also offering several programming and analysis challenges relating to dealing with a massive and comprehensive dataset. For example, social networks tested with multi-theoretical multilevel (MTML) models have generally been small due to the difficulty of collecting complete datasets on groups, yet the current project will offer thousands of complete groups of sizes ranging from two to 2,000, with complete behavioral data over a 16-month time window. For research into the social implications of online worlds, the data offer tests of the cyberbalkanization hypothesis, new media isolation hypotheses and explorations of virtual communities. The dataset also offers the ability to do something never yet done in economic analysis: generate a time series of an entire economy. The wealth of data will require the creation of indices, new metrics and new ways of interpretation. This project has no precedent in social science, both because virtual worlds have not yet been studied systematically, and because it combines attitudinal data with previously unattainable levels of unobtrusive behavioral data. The project allows for both methodological ground breaking as well as innovative theory testing. The research will require the creation of new techniques bridging the domains of both computer science and social science. These new methods will be valuable for future scientific research, for improving the games themselves, and for exploiting the educational potential of virtual worlds.