Uses of Multiple Characters in Online Games and Their Implications for Social Network Methods Alex Leavitt Annenberg School for Communication & Journalism, University of Southern California aleavitt@usc.edu Joshua Clark Annenberg School for Communication & Journalism, University of Southern California joshuaac@usc.edu Dennis Wixon School of Cinematic Arts, Interactive Media Division, University of Southern California dwixon@cinema.usc.edu ABSTRACT In most sociotechnical systems, individuals are tracked through user accounts. This paper explores the various ways in which people create and use multiple user representations, specifically in online games. Using 8 years of population data from a popular multiplayer online game, EVE Online, we examine how multiple character creation and use occurs at scale and how operationalization of individuals between accounts and characters impacts methods. We suggest that conceptualizing participants in online games based on the assumption that one character equals one individual can lead to incorrect analyses regarding the demographics or behaviors of the population. Additionally, social network analysis suggests that a character-centric, rather than account-level, viewpoint can change the results of statistical relationships with network metrics such as eigenvector centrality. Author Keywords Multiple profile maintenance; online games; social network analysis; user representations ACM Classification Keywords H.5.3. Group and Organization Interfaces–Web-based Interaction INTRODUCTION In most sociotechnical systems, an individual must create a user account to contribute. Frequently, we see the technology industry – and even occasionally academia – accept that trace data tied to one account represents the holistic experience of one individual. Especially as more and more systems track people’s metadata, engineers and researchers continue to adopt user profiling and other aggregated empirical techniques to more accurately reconstruct individual’s experiences. However, some individuals that participate in online systems decide to create multiple representations of themselves by generating more than one user account. In the literature, this practice is known as multiple profile maintenance, and we call the implementation of user identification in sociotechnical systems “technical user representation.” These representations are then used for a variety of reasons, though usually each is set aside for a distinct purpose. For instance, one person may create a more-private personal Twitter account to share updates with friends while maintaining another public, work-specific account. While some research has been conducted on why people separate these traces across multiple accounts, little research has empirically examined how people separate these accounts. Studying how people use multiple technical representations is important for understanding identity practices, but these uses also have important implications for social science methodologies and theories. If people create multiple representations of themselves, researchers should be aware of how these practices might impact their interpretation of trace data. Games constitute one approach to understanding the behaviors of separating experiences across different technical user representations. Players can create multiple avatars and interact with the game environment or other players in varied ways, generating different sets of trace data across these virtual characters. Frequently, games catalogue each avatar under one account per individual, uniting the characters under a common login, even though they represent distinct personae in the game world (especially since the player often cannot use more than one character at the same time). Our motivation is to understand how the two different representations of a person within a game world (account and character) impact our understanding of behaviors in sociotechnical systems. In this paper, we focus on two research questions, the first informing the second: RQ1: How do players use multiple characters, and do players create similar experiences (and trace data) across multiple characters? RQ2: When players use multiple characters, how do these uses affect analyses of trace data and interpretation of behaviors? In this paper, we use the case study of one massively multiplayer online game (MMOG) to study multiple character use. First, we survey the literature on multiple account use; in particular, we analyze the current game literature’s approach to multiple character use. We then examine the entire population of players using a dataset comprising over 8 years of server logs since the game’s Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page.