10
USING VIRTUAL GAME
ENVIRONMENTS TO STUDY
GROUP BEHAVIOR
UNIVERSITY OF NORTH FLORIDA
VRIJE UNIVERSITEIT AMSTERDAM
PURDUE UNIVERSITY
Research on group behavior in virtual environments is needed, not only because people are spending more and more time in online virtual environments, but also because the distinction between real life and virtual life is blurring. Millions are entering online virtual worlds that simulate their everyday lives or are pure fantasy. It is clear that online games are ever growing in their popularity and the significance they play in everyday life. An incredible number of people have adopted fast-rising online virtual games like World of WarcraftTM (11.5 million),1Second LifeTM (13.3 million),2 and EVE OnlineTM (300, 000).3 Institutions are also flocking to virtual worlds. Universities are establishing their presence, even going to the extent of recreating their university in a virtual space (see Figure 10.1 for an example). Fortune 500 companies like IBM, Philips and ABN-AMRO are choosing to create virtual spaces4 in online virtual games. Some of the high-tech titans of these Fortune 500 companies (e.g., founders of corporations, project managers, vice presidents) admit their game playing helped them excel in their roles as bosses (Widman, 2008). As more and more individuals enter online worlds, research should also establish its place within the virtual worlds.
Virtual game environments create opportunities to study group behavior in a variety of different ways. A researcher studying groups could ask: how are virtual games part of everyday group experiences? How do virtual games fulfill the need to belong to a group? Virtual games can be used to observe group behavior that may be difficult to capture in a real-life setting. How do groups form and dissolve? How are meeting spots established for groups to meet? Virtual game environments can also be used to create virtual laboratories where previously unobservable behavior might be explored. For instance, after an individual is ignored and excluded (ostracized), how would he or she behave? Laboratory-based measures of behavior may constrain the participant's options and prevent a researcher from truly understanding the behavioral response to the ostracism. If an ostracized individual wants to behave aggressively, he or she may feel more comfortable aggressing against another avatar, but not another person. The goal of this chapter is to provide insights into the nuts and bolts of doing research that investigates any one of these topics.
FIGURE 10.1 Examples of structures that can be created on a plot of land in Second Life. These figures depict a replica of Vrije Universiteit Amsterdam created in Second Life.
The Role of Games
Games are often played as a form of entertainment, sometimes with friends or family. Alternately, games can be played for a variety of other purposes. For example, sports are played not only because they are enjoyable, but also because they are a form of exercise. The US Army uses games as a recruiting tool (www.americasarmy.com). Players get the experience of soldiering in the US Army through military elements such as training, technology, weapons, and by engaging in team-based, multiplayer, force-on-force operations. Virtual reality games are also used to help rehabilitate individuals with disabilities; for a review see McLaughlin (2006). Individuals might manipulate and move various objects around virtual environments to improve their motor abilities. In our case, we use games to study group behavior.
Games create the potential for individuals to interact, be mutually aware of each other, be interdependent, and anticipate interacting for a period of time, all defining criteria of a group (McGrath, 1984). We find games to be one of the best means to create an instant group and engage them in a task. Games were initially used by researchers studying group behavior to study social dilemmas (Hardin, 1968; Olson, 1965), such as cooperation versus competition in prisoners’ dilemma-type games (Axelrod, 1980, 1984; Luce & Raiffa, 1957). In social dilemma games, one chooses between a behavior that benefits the group and a behavior that benefits oneself (Baron, Kerr, & Miller, 1992). Prisoner dilemma games were popular because researchers could control a variety of aspects of a game (e.g., payoffs and punishment, ability to communicate) to induce behaviors they want to investigate.
Games continue to be used to engage participants in a group task. However, they have evolved from being played face-to-face to now being played in virtual settings. For example, researchers recently created a virtual version of the board game ClueTM (Jones, Carter-Sowell, Kelly, & Williams, 2009). Participants ostensibly interacted with other game players online (actually the computer) who gave the participant all the available clues to solving who killed the Professor, or elected to give the participant only some of the clues. Only receiving some of the clues excluded the participant from pertinent information, giving researchers the ability to investigate how a person feels when he or she is “out of the loop.” Additionally, a simple game of ball-toss was recreated in a virtual environment (e.g., Cyberball; Williams, Cheung, & Choi, 2000); for a virtual reality recreation, see Williams (2007). Online virtual game environments, referred to as Massively Multiplayer Online Role-playing Games (MMORPGs), are a new way researchers can study how groups form, complete a task, and dissolve. Virtual game environments have advanced to the extent that we use them to create virtual laboratories to study how members of a group interact. For the remainder of the chapter, we will discuss how to use virtual game environments to study group-based behavior.
Virtual Game Environments
A virtual environment can be defined as a three-dimensional, computer-generated simulated environment that is rendered in real time according to the behavior of the user (Loeffler & Anderson, 1994; see Figure 1 for an example). Games that utilize a virtual environment include a Massively Multiuser Online Games (MMOGs), which refer to an environment where computer players come together to engage in a game activity, such as in HaloTM. MMORPGs are large persistent virtual worlds that run independent of the user (Yee, 2006a). In these games, players take on the role of fictional characters, avatars, which move through worlds hosted by the game's publisher. These worlds are constantly evolving and changing. Players want to improve their character through completing tasks related to the game and often band together with other players creating long-term social groups. Some of the more popular MMORPGs include World of Warcraft, The Sims OnlineTM, Second Life, Lord of the RingsTM, and EverQuestTM.
In a MMORPG, players can live a virtual life by means of their avatar and’ dependent on the nature of the game, use their avatar to explore, compete, create content, collaborate, or socialize (see Figure 10.2 for an example avatar). An avatar is a graphical embodiment of a participant that conveys the participant's identity, presence, location, and activities to others (Benford, Greenhalgh, Rodden, & Pycock, 2001). Many MMORPGs offer an elaborate set of functions to edit avatar appearance. The default edit options allow for the development of human-like avatars, ranging from an accurate copy of the person operating the avatar, to the creation of an avatar that resembles the player's favorite movie character, as well as object-like avatars (e.g., cardboard box), and anthropomorphic animal-like creatures.
FIGURE 10.2 An example of avatars used in a Second Life experiment.
Why study group behavior in virtual worlds
“It's possible that one large category of human interactions in the future is going to be based on avatars.” (D. Williams quoted in Science; Miller, 2007, p. 1341).
Attempting to answer the question of why study online games now, D. Williams argued the decline in civic and shared spaces, along with a decline in real-world places to meet to converse with people has created the backdrop for the rise of social gaming (Wiliams, 2006). The need to belong (Baumeister & Leary, 1995) has not been reduced, but the number of social outlets has, prompting social ties to be made in virtual communities (Rheingold, 2000). For example, rather than meeting at the local drinking establishment, friends might find themselves logging in to World of Warcraft and, while planning group tactics, they might catch up on important events in their real lives (Williams, 2006). Playing games online is not just where individuals simply go to “play,” rather, games more and more are a reflection of naturally occurring phenomenon. As Ito put it, “Yes, it's just a game … the way the real world is a game,” (Newsweek, September, 2006).
We now have a whole new arena to capture behavior of groups. Online virtual environment games have allowed us to investigate questions that may not be easily answered. Contractor points out, “virtual worlds provide an outstanding exploratorium for us to gather data and test models,” (Miller, 2007, p. 1342). Contractor and his colleagues’ work demonstrates a primary benefit to online research: complex questions can be investigated in a reduced amount of time compared to real-world studies. He spent 3 years and approximately $1.5M conducting in-depth surveys with more than 30 working groups from places such as NASA, Boeing, and Charles Schwab. In only months, the research team conducted similar experiments looking at World of Warcraft, and found similar results as organization-based studies. Virtual online games such as The Sims Online or Guild WarsTM present an unprecedented means to observe large-scale phenomena (e.g., hundreds organizing for battle) or to find enough instances of rarely occurring phenomena to make conclusive observations.
In conjunction with the fact that more people are flocking to virtual environments for social purposes, we use virtual environments to overcome some of the limitations we experience when studying groups in a laboratory setting. There are at least three methodological problems virtual environments help us overcome: (1) the experimental control–mundane realism trade-off; (2) the lack of replication; and (3) unrepresentative sampling (Blascovich et al., 2002).
Experimental control versus mundane realism
Experimental research traditionally requires a trade-off between experimental control (i.e.., precise manipulation of independent variables) and mundane realism (i.e.., the extent to which the experiment is similar to situations encountered in daily life); the higher the mundane realism the lower the experimental control (Blascovich et al., 2002). We run studies in the laboratory, creating high experimental control, but we worry that the outcomes may no longer resemble how a group behaves in a more realistic environment. We use virtual environments to lessen this trade-off because they allow us to build elaborate and realistic experimental situations that still provide the experimental control we need to deduce what is causing an outcome.
Lack of replication
Replication is problematic because of the difficulties researchers studying groups experience in the exact application and implementation of procedures, methods, and settings used by other investigators. Sharing physically identical laboratories is impractical. However, by building laboratory settings in a virtual environment, we can export both the laboratory setting and the automated procedures to control the setting (e.g., scripts) to a digital file that can be imported in the environments used by other researchers. This process increases the portability and potential usage of existing experimental virtual laboratories. Additionally, we can invite other researchers directly into our virtual environment to conduct their research.
Nonrepresentative samples
We find one of the greatest benefits to using virtual game environments is the ability to have a random and representative participant sample, one greater in diversity than a college student sample. The use of internet-enabled virtual environments can level physical and geographical barriers to random and representative participant selection. Every person that has an internet connection can be invited to participate in an experiment that is hosted in a web-enabled virtual laboratory. We do not have to limit our scope of participants to the students in the classes we are teaching, but we can acquire a representative sample of participants, potentially by recruiting inhabitants of the virtual environments.
A typology of virtual game environments
We developed a typology to break down the different types of virtual game environments that can be used to study group behavior. We categorized virtual game environments by two dimensions: (a) whether they are based on a game or virtual life scenario; and (b) whether they are developed by designers or users. An individual playing any of the games within each of the typologies can either play alone or join others to form a group, such as warming up playing Rock BandTM alone before and joining others to form an impromptu band that goes on tour. Our typology may not neatly categorize every type of game that could be used to study group behavior, but it provides a starting point for deciding what game might work best to study a given behavior.
We first describe the distinction between game and virtual life scenario environments. They primarily differ in the way they give substance to the role-playing part of the games. The activities of players in game-scenario MMORPGs revolve around the rules of a built-in game scenario. Players have explicit goals, like “conquering land,” “beating the enemy,” “fighting monsters,” or “completing a quest,” and they behave in the context of story lines that are provided by the game (Yee, 2006b). Examples of game-scenario MMORPGs are World of Warcraft and EVE Online.
The primary purpose of virtual-life MMORPGs is to offer players the opportunity to live a virtual life independent of the rules of built-in game scenarios. In these environments people do the same things as they do in real life. Indeed, virtual game environments are a secondary form of life. Linden Lab, the creator and owner of Second Life, described Second Life as, “a world imagined, built and created by its Residents – people like you. Every minute, Residents assemble buildings, design new fashion lines and launch clubs and businesses.”5 There. comTM, Second life and Active Worlds TM worlds are examples of these types of virtual games. Compared to game-scenario MMORPGs, virtual-life MMORPGs offer more advanced features to create and develop tailor-made virtual environments because this is inherent to their purpose. We find these features very helpful for doing research.
In some cases designers developed the virtual game environments. Often these worlds are created with a specific intent. For example, Sony's HomeTM (accessible for Playstation 3 owners) allows users to go into a virtual environment to check out new media releases such as video games, movies, and music. Developments in these environments are based on additions that designers produce. Other virtual game environments are initially established by designers with the intent that users develop the virtual worlds. For example, in Second Life, users can purchase different land plots and develop them. Table 10.1 outlines the matrix of virtual game environments and gives an example of each type.
For the purposes of group behavior research, we will focus our discussion on virtual game environments that incorporate a game scenario (both designed and user developed) and user developed virtual life scenarios. These three types of virtual game environments have the greatest potential for investigating group behavior. All three provide a game element that causes groups to form quickly. In the case of a user-developed game, researchers can create virtual environments to study groups. These environments may be as formal as a virtual laboratory or informal, such as a park where avatars might want to hang out and observers can record their behavior.
Making observations of group behavior in virtual
game environments
In this chapter we will share our experiences with laboratory experiments that were executed in virtual environments. However, we want to point out that the use of virtual worlds to study group behavior does not have to be limited to executing experiments in virtual environment settings. Virtual worlds, like Second Life and Active Worlds, host a great variety of communities that frequently organize and engage in online group activities. These activities can be used to study natural group behavior, both as a participant observer who is engaged in group activities while simultaneously noting group behavior, or through observational analysis.
Table 10.1 Matrix of video game enviroments typology
| Designer developed | User developed | |
| Game scenario | Designers specifically create a virtual environment intended to enhance a game playing scenario. Examples: Rock BandTM, The Legend of ZeldaTM, Madden NFLTM, and WarcraftTM | Users develop the virtual environments by creating objects and modifying aspects of the environment. These objects become part of the ongoing game. Examples: World of Warcraft, Guild Wars, and Lord of The Rings |
| Virtual life scenario | Players cannot create the environment, but rather the designer can change the virtual environment as they see fit. The environment is designed so that players can live out their virtual lives by navigating through these worlds. One example is Sony's Home | Players can modify scenery and create objects, creating the worlds through wqhich players navigate. Players live out their virtual lives by doing a variety of activities and engaging with others as they would in everyday life. Examples: Second Life and There.com |
Some interesting observational questions that might be investigated include: how do group members know where to meet? What happens to group members that get excluded from the group? How do larger groups function differently than smaller groups? How do groups initiate new members (this might be interesting to explore as a participant observer)? Large-scale observational questions can also be investigated. How are rules created and enforced? How are social norms established? Where are major locations where group congregate (e.g., cities) established? As Castronova (2006) observed, virtual worlds are Petri dishes for the social sciences.
There also exist several challenging research questions concerning the likeness between group behavior in the real versus virtual world including: are real-life social network dynamics also applicable in virtual environments? Do online groups form in a similar way to real-life groups? Do people share knowledge in the same way when hosted in a virtual office, compared to the way they share knowledge in a real life setting? What is the influence on group dynamics if the CEO is represented by his or her avatar, and will not be recognized as such by the other members in the team? Do virtual teams also socialize at the virtual coffee machine? It is also interesting to explore how group behavior in a virtual setting can enlighten group behavior in a real-world setting. It is exciting to see that companies like IBM study game settings to investigate if the leadership strategies that emerge in virtual games are also applicable in real-life business contexts (Miller, 2007).
Designing a Virtual Game Environment to Study
Group Behavior
One way we study group behavior is to create a virtual environment where we can observe and measure a desired behavior. We created a virtual environment where participants could toss a ball around, called Cyberball. In our typology, this type of approach fits in the category of a designer-developed game scenario. We use this approach to study ostracism (being excluded and ignored) because it is optimal for efficiently conducting a variety of studies related to a single topic. Cyberball is an example of how a virtual game can be created to study numerous aspects of a specific phenomenon efficiently and effectively.
The development of cyberball
The first ostracism study involved confederate researchers playing ball-toss in a laboratory with participants (Williams, 1997). It was based on an experience Kip had at a park. After he returned an errantly thrown Frisbee, he was included in a game of toss only to have the players stop throwing it to him without an explanation why. To replicate this experience, participants sat in a triangle configuration with two other supposed participants (actually confederates) already seated across from each other (see Figure 10.3). At a certain point of the experiment, the confederate players took a ball out of a toy bin and tossed it around, initially including the participant. However, in one condition, the confederate players stopped throwing the ball to the participant, only to themselves (just like at the park). They additionally did not make eye contact or interact with the participant. Participants in this condition indeed felt ostracized.
Kip thought it was prudent to create a computer game that could address several limitations associated with face-to-face paradigm. The face-to-face ball-toss paradigm was cumbersome: only one participant could be run at a time, two confederates had to be present for every experiment session, and the confederates need to be trained on how to interact for each of the conditions. Confederates’ behaviors may have varied between sessions. Of greatest concern was how the confederates felt when being asked to ostracize another individual. It was initially distressing for confederates to ignore and exclude another individual. Potentially to reconcile the distress of being asked to unfairly ostracize another individual, the confederate researchers began to ostracize participants with a sadistic glee. The face-to-face ball-toss paradigm turned into Cyberball, a virtual game environment in which participants toss around a virtual ball (Figure 10.3).
FIGURE 10.3 Face-to-face ostracism and a virtual game of ball-toss, Cyberball.
Cyberball replicated the park experience and the face-to-face ball-tossing paradigm in a virtual game environment. By making the ball-tossing virtual, the need for confederates was eliminated and multiple participants could be run at a single time. During Cyberball (Williams & Jarvis, 2006; Williams et al., 2000), participants are instructed to visualize the ball-toss interaction, so as to avoid focusing on their ball-toss performance. The participant plays with two other players, similar to the face-to-face ball-toss, with both of these players being computer controlled. The participant is depicted by a hand at the bottom of the screen. Participants can play in either three- or four-player game.
Utilizing a designed virtual game environment
The features the designers included are critical to the utility of Cyberball and the ability to investigate ostracism's effects thoroughly. Initial ostracism studies investigated how ostracized participants felt compared to included participants (Williams et al., 2000). This included finding moderators of ostracism's impact, such as telling participants they were being ostracized by computer-controlled players compared to other players online (Zadro, Williams, & Richardson, 2004) or ostracized by a hated group (the Ku Klux Klan) compared to a more favored group (e.g., a mainstream political party to which one belongs; Gonsalkorale, & Williams, 2007). This last study involved manipulating pictures and names next to the confederate players. All of these questions utilized functions inherent in Cyberball or manipulations of instructions associated with the Cyberball game.
Small modifications to the virtual game environment can extend the ability to ask novel questions. Jim modified the characteristics of the Cyberball players, creating Cyberboys and Cybergirls, to study the impact of group membership on the recovery from an ostracism episode (Wirth & Williams, 2009). van Beest and Williams (2006) created €yberball (pronounced Euroball) where, in some conditions, participants lost money (Euros) when they received the ball, making it beneficial not to be included. A counter helped participants keep track of how many Euros they gained or lost. We evolved Cyberball by adding a new factor, control over how long it takes for the confederate players to decide to throw the ball (Wesselmann, Wirth, Pryor, Reeder, & Williams, 2009; Wirth, Wesselmann, Williams, Pryor, & Reeder, 2009). We find players that are burdensome to the group (i.e.., take a long time to decide to throw the ball) are ostracized. Because Cyberball logs who is thrown to and how long it takes to throw the ball, we were able to make Cyberball a dependent measure. This slight modification now allows us to use Cyberball to investigate a new ostracism question, when does ostracism occur?
What is needed to create a virtual game environment
To develop a virtual game environment from scratch, researchers either need to be a designer (someone knowledgeable about computer programming), be willing to learn how to program, work with a virtual game designer who is willing to create the program for little or no money as a way to get more experience with virtual game programming (thereby needing to be patient as he or she makes mistakes), or be willing to spend a lot of money (although, this is not the case for virtual games like Second Life). Even if you have a designer there will be difficulties as you figure out what to include in the program, how to do it, what is reasonable for the scope of the project, and be able to communicate clearly with the programmer about desired aspects of the game. The difficulty creating the program will be a function of how complicated you want to make the design. Cyberball is a relatively straightforward program, but it includes a number of ways to manipulate characteristics of the game and several dependent measures.
For others to be able to use a computer program successfully, it needs to be designed to be user friendly and have clear instructions. It is important to share a virtual game so that other researchers can validate and extend your results. Cyberball is a game that is available for anyone to download freely.6 It is also helpful to make the program one that can be modified to fit other researcher's applications. It was easy for Jim to create Cyberboys and Cybergirls, he simply changed the picture files that are used to show the Cyberball being tossed around. We were able to add the variable of how long a player took before throwing by adding lines of code to the Cyberball program. We worked with a computer programmer who was able to do it in a short amount of time for a minimal cost. Considering the costs that can be associated with creating an original virtual game, it may be worthwhile to first consider if using online worlds would achieve your goals. However, they have their own limitations.
Utilizing a User-developed Virtual Game Environment
to Study Group Behavior
One means of understanding group behavior is to observe how groups interact while playing a game. Applying our typology, this type of research would focus on user-developed game-scenario virtual environments. In these games, groups form to complete various tasks as part of the game scenario. For example, in World of Warcraft, players might form a guild to go on a quest to secure valuable minerals. There are generally minimal requirements for doing research in these environments: essentially, access to the game and, in the case of a participant observation study, the ability to develop the avatar player as the game progresses (e.g., acquire new weapons, buy accessories). Conducting research within a user-developed game-scenario virtual environment is ideal for studies that require a large number of participants, for making observations that are difficult to capture in real life, or for capturing processes that take long periods of time to develop (Bainbridge, 2007).
Researchers can use virtual games to study behaviors that are difficult to capture. For example, Martey and Stromer-Galley (2007) explored the “digital dollhouses” of The Sims Online. Specifically, they investigated social norms of avatars in houses created by players in The Sims Online. The researchers visited several houses using their avatar and minimally engaged with others as they made observations (participant observation). They observed a wide range of norms associated with hosting others at a house, being a visitor, group projects, public conversations, and avatar action and interaction.The Sims Online is designed to engage players in a variety of group interactions, earning more points for a group task than doing a task alone. Group tasks can be business-like, such as two to four Sims operating a pizza machine, or more leisurely, such as dance parties or weddings.
Group phenomena that are nearly impossible to investigate in a real-world context may be possible to investigate in online worlds. For example, Castronova (2006) was interested in studying coordination equilibrium, or a point at which players agree to meet at and create a market to exchange goods. He studied where markets for goods were established inside the games EverQuest and Dark Age of CamelotTM. Looking across similar worlds on several different servers, Castronova was able to test the theory of coordination. Similar to the prisoner dilemma games we discussed earlier, researchers use virtual environments to investigate how individuals can be induced to cooperate in producing public goods (Harrison & List, 2004). Online virtual games make is possible to observe large group behavior that otherwise would be nearly impossible to observe in a real-world setting.
Online worlds can potentially provide a large dataset to investigate questions where many observations are needed. They can also provide significant power to test for effects that may not be particularly large. For instance, Xiong, Poole, Williams, and Ahmad (2009) investigated the effects of group structure on group behavior and outcomes using data collected from Everquest2 TM. The researchers were able to demonstrate that structure and process at the group level can influence group behavior. They were also able to investigate direct and indirect effects due to the large sample size.
Researchers that are part of the Virtual Worlds Exploratorium offer additional examples of how online worlds can be used to study group behavior. These researchers forged a collaboration with Sony Online Entertainment to analyze data from the US-dedicated servers of Everquest2. This will allow them to investigate numerous dynamic group-based questions. Owing to vast social networks that are constantly changing, researchers (Xiong et al., 2009) are looking at who enters and leaves groups as they form, merge, and dissolve. They are also looking at how different levels of expertise can affect group outcomes. These researchers are capitalizing on the thousands of groups that can be observed.
Traditional field experiments can also be conducted in the virtual game environments. Eastwick and Gardner (2009) demonstrated that biases and processes in the real world carry over into the social world, despite the fantasy element of virtual worlds. They found the race of an avatar (light versus dark skinned) influenced the compliance with how willing individuals playing in Second Life would be to comply with a small request following a larger one (i.e.., door-in-the-face technique; Cialdini et al., 1975). These results are similar to those found in laboratory-based research: the door-in-the-face technique was less effective for dark-skinned versus light-skinned avatars. Further simulating real-world field experiments, researchers using online games have found avatars will leave physical space between each other and female players made more eye contact and stood closer to other avatars than males (Yee, Bailenson, Urbanek, Chang, & Merget, 2007). Many of the approaches used to study group behavior in user-developed virtual game environments capitalize on how the online world reflects real-life interpersonal interactions.
Investigating Group Behavior by Utilizing User-developed
Virtual Life Game Environments
The first two uses of virtual game environments involved researchers examining group behavior as it related to playing a game. This game may have been created to investigate a given group behavior, such as Cyberball, or group behavior may be observed as individuals participated in an online game, such as watching a group complete tasks in World of Warcraft. However, a third type of virtual game environment in our typology involves user-developed virtual life scenarios. These worlds are created to provide users a virtual second life. With this approach, rather than observing group behavior as a function of a game, researchers can create virtual scenarios, including a laboratory, within the virtual game environment. Second Life is an example of a virtual game environment we have adapted for the use of studying group behavior.
In user-developed virtual game environments, it is possible to create any type of scenario, including a virtual laboratory, in an undeveloped space. It is possible to own virtual estates in Second Life that can be used for any purpose the owner wishes to use it; in our case, we use them for research. The Second Life scripting language allows for the development of objects that can help to guide, control, and monitor key processes in experiments. Scripts in Second Life are a set of instructions that can be placed inside any object, or any object worn by or attached to an avatar. Scripts can make an object move, listen, talk, operate as a car or a plane, change size, shape, or color. A script can make an object react to your words as well as talk back to you. Scripts are written using an editor that is built into the Second Life user interface. Second Life offers a relative media rich environment including instant messaging, voice over IP, and avatar gestures.Second Life also supports the integration of streaming media (music and video). Controlling the availability of different cues allows for experiments that measure the influence of these cues on group behavior.
Frans, Alex, and Bart have already conducted group-based research in online virtual game environments, specifically, Second Life. They were interested in how virtual worlds influence group decision-making processes and their outcomes. They had four important reasons for conducting research in Second Life, including: (1) Second Life is a platform that hosts a variety of group processes; (2) based on prior experiences with Second Life, they knew that it was relatively easy to build an experimental setting and find participants familiar with Second Life (Verhagen, Feldberg, Van den Hooff, Meents, & Merikivi, 2011); (3) Second Life offers a great variety of features to modify avatar appearance easily; and last but not least, (4) at the time of the experiment, Second Life appeared to be a very popular virtual world attracting huge numbers of people as well as businesses all willing to explore the opportunities of virtual worlds.
The study Frans, Alex, and Bart conducted using Second Life involved participants, undergraduate students recruited from a class on business administration, who formed three-member teams representing a neighborhood council. The council had to make a decision among four different options on what would be built in a vacant space in the neighborhood they represented. The trio of researchers manipulated the degree the council was immersed in the virtual environment. Three levels of immersion were created for the experiment: (1) a text-chat only condition – participants in this condition could only make use of an instant messaging tool to communicate with each other, (2) a virtual decision room condition – participants in this condition were identified by an avatar and hosted in a small windowless meeting room containing only a table and chairs (see Figure 10.4); and (3) a virtual neighborhood condition – a digital representation of the neighborhood about which the team should make a decision (see Figure 10.5). In the virtual neighborhood condition, group members could move around the virtual space on which the group was making a decision. The findings of this experiment revealed that the participants felt more enveloped and included in the decision situation in the virtual laboratory they created for them in Second Life, than the participants who executed the decision task in a real-life setting and compared to those who used text-based communication only while not being in a three-dimensional (3D) virtual environment.
FIGURE 10.4 The virtual decision room.
FIGURE 10.5 Virtual neighborhood participants could navigate around while making a group decision on the use of the open space.
Requirements for creating an experimental environment
in an online virtual game
To build the experimental setting in Second Life, they first needed to acquire space to establish a virtual laboratory. They did this by buying two so-called “virtual regions” from Linden Lab. Regions can be bought through the Second Life web-site. Linden Lab (the designers of Second Life) offers different types of regions, qualified according to the numbers of avatars that can be hosted simultaneously on a single region, and the maximum number of shapes (called primitives or prims in Second Life) that can be used to create a virtual setting. Linden Lab's pricing model includes a setup fee, and a monthly maintenance fee. The setup fee for the type of island we bought is $700, and the maintenance fee amounts to $150 per month (a discount pricing model is applicable for educational institutions and nonprofit organizations). As long as the research conducted does not violate legal conditions, no special permission from Linden Lab is needed to use Second Life for scientific research. In the end, scientific research fits Linden Lab's vision of “a world imagined, built and created by its Residents.”
The virtual spaces they used were actually two “virtual islands,” both owned by their research group. We decided to use two different regions, each measuring 65, 000 square meters, in anticipation of three issues. The first issue was that the number of shapes (called primitives or prims in Second Life) that can be used to create a virtual setting is limited to 15, 000 per region. Each object (e.g., building, table, car) in Second Life is basically a set of linked prims (building blocks). For example, a table can be built by using five prims: one box prim for the tabletop, and four cylinder prims, one for each of the four legs of the table. The second issue they contended with was performance. Each region in Second Life is linked to a certain amount of computer processing power. Performance is dependent on such factors as the number of prims used, the number of avatars present, and the scripting load. By using two regions that are joined together, the same avatar, prim, and scripting load will be dealt with by doubled processing power. The third issue was interaction between avatars in differing conditions. If the teams in the two different conditions get into each other's space they could potential “hear” each other and the discussions become mixed.
Expertise in building and scripting
An additional requirement for conducting experiments in Second Life is building know-how and expertise. Researchers will need people who are able to learn the skills of building in Second Life. However, the learning curve for mastering the Second Life tools appears rather steep, even for people who do not have experience in 3D building and modeling. If you do want to make use of scripts, people with more advanced programming skills must be part of the team. Most scripting languages that are used to write scripts are comparable to programming languages like Java or C. For Fran, Alex, and Bart, they recruited students to assist them in developing the environment. One call for Second Life building expertise, posted in our digital learning environment, appeared to be enough to create a pool of dedicated Second Life building experts. The full experimental setting was created by two builders in a lead time of four weeks.
The virtual laboratories
Conditions 2 and 3 of the experiment were developed on two different virtual regions of Second Life. The virtual decision room (condition 2) was a small windowless meeting room. A conference table with three chairs, one for each participant, was positioned at the center of the virtual room (see Figure 10.4). The chairs and table were not developed by the research team, but instead were obtained for free on one of the Second Life locations that offer a multitude of “freebee” objects. Every avatar has a so-called “inventory entry” in which objects can be stored. Objects that are not copy protected (dependent on the settings of the object) can be transferred to other avatars or can be placed in a virtual space. It is also possible to buy objects, like furniture, clothes, or gadgets, in one of the many shops hosted in Second Life. A multimedia object, including a streaming video of the neighborhood the team had to decide on, was attached to one of the walls of the decision room. Each participant could individually watch this video by touching the play button in the media section of the user interface.
FIGURE 10.6 The development of the virtual neighborhood in Second Life was based on buildings in the real-life neighborhood.
The neighborhood (condition 3) was a representative reproduction of an existing urban neighborhood that consisted of a residential area, several office buildings, and small production and trade companies (see Figure 10.6). To create a setting that resembles reality as much as possible, the virtual neighborhood was an accurate copy of an existing neighborhood. The development of the virtual neighborhood was guided by pictures of the houses, buildings, and offices situated in the real-life neighborhood (see Figure 10.5). The vacant space in the center of the neighborhood in Figure 10.5 was the subject of the decision task. Avatars could explore the full virtual neighborhood, but could not leave it. Owing to the ability to control who enters and leaves the environment, no avatars other than the ones assigned by the research team were present.
Preparing the participants to use second life
Because most of the participants were not familiar with Second Life before the experiment, they had to be aware of so-called “novice effects.” If technology is needed in support of a task to be performed, task performance might be dependent on the level of expertise with the relevant technology of the person performing the task (Venkatesh & Davis, 1996). To avoid novice effects they organized a one-hour tutorial for the participants that were assigned to the virtual environment conditions in the week prior to the actual experiment. In this tutorial session participants learned the basics of the Second Life user interface as well as the basic skills of movement, navigation, and interaction. For example, they taught participants the different ways of moving around: walking, running, and flying, as well as how to communicate with each other through the built-in instant messaging features.
Recruiting participants in a virtual game
Next to recruitment of participants in real life, virtual worlds offer the opportunity to recruit participants “in world” too (“in world” means in the virtual world setting). Populated virtual worlds offer different ways for “in world” participant recruitment. On the individual level, a researcher can use his or her own avatar to invite others directly to participate in a research project. It is important to pick an online game where you fit in and can comfortably participate. Virtual worlds include built-in search features like “top ten most popular places,” “recommended events,” or “destination guides,” that can be used to select the right places for recruitment. Virtual worlds offer the opportunity to search for places and events. Researchers can use a keyword search to help select places and events that might be of specific interest for the researcher, or to compose a balanced sample of participants. A virtual place that has been tagged “shopping mall” by its owner, can be very interesting for a researcher exploring group decision-making in a commercial virtual settings. Crowded places offer both the opportunity to get in touch with other residents as well as being a venue for observational studies.
Participants can also be recruited for research at the community or group level. Most virtual worlds host groups or communities. In Second Life, for example, it is relatively easy to create a group and invite avatars to join the group you created. A researcher can also search through group profiles that are made available through a public directory. When a profile includes the name of the owner of the group, this owner can be asked to cooperate by asking the group members to participate through communication mechanisms like group notices. Group profiles can also be used to select special interest groups that might be of specific value for the researcher. Some virtual world communities even deploy activities in real life, have their own websites and maintain up-to-date mailing lists. These communities can be very valuable when it comes to participant recruitment. A combination of individual and group-level recruitment can be realized by asking an individual avatar to cooperate and ask all the members of the groups he or she had joined to participate in the research project too. Amy Bruckman (this volume) offers several tips for recruiting participants at the individual and group level.
Impressions of using Second Life for an experiment
What did Frans, Alex, and Bart learn from using Second Life to conduct group experiments? Second Life can be an appropriate platform to build virtual laboratories and support research projects without inducing huge investments. The opportunity to build and use scripts to control behavior was also valuable, as well as the built-in option to store the discussion logs. The opportunity to integrate multimedia objects relatively easily makes it possible to create lively and vivid virtual settings. Second Life offers a head start for creating an experimental laboratory environment, not only due to its nature of being a platform for building custom-made settings, but also because of the availability of a great variety of objects that can be bought at relatively low prices (or for free) to furnish your laboratory. During the execution of the experiment, Second Life appeared to be rather stable and did not cause serious technical or performance problems. Are they going to use Second Life for future projects too? The answer is “yes,” and they have already prepared a follow-up experiment that aims to address the research question: can 3D virtual environments contribute to individual and collective understanding of a decision problem?
Limitations to Using Virtual Game Environments
Researchers are limited in using online virtual game environments because they are dependent on the proprietors of the game. Using commercial virtual environments that run on computer servers that are controlled by third parties may result in “unscheduled” maintenance and other changes in service that could interfere with conducting an experiment. The quality of the game environment may also vary. The high-resolution user interface of virtual game environments like World of Warcraft, for example, produces more sophisticated images than the relatively low-resolution interface of Second Life, making the latter virtual environment less realistic. Although higher resolutions are not always preferred above lower resolutions (Serviss, 2005), research findings indicate that more realistic game environments (higher resolutions) positively influence engagement, immersion, and focus (Sabri, Ball, Fabian, Bhatia, & North, 2007). Researchers are also limited because they may not be able to adapt the virtual environments in ways that would be helpful, such as integrating media from an external system (e.g., opening an interactive web page in the virtual environment itself). Finally, by using a virtual environment that is established by an external commercial party, the user will always be dependent on the commercial viability of the supplier.
One should always be aware of the fact that new virtual worlds may come and go, existing for a short time before interest wanes and the virtual world no longer exists. To make our research projects less dependent on third parties we are exploring open source solutions for virtual worlds. One such a solution is the so-called OpenSimulatorTM (www.opensimulatior.org). OpenSimulator can be used to create a virtual environment similar to Second Life under your own control. It is even possible to use the Second Life client software to access virtual environments created in OpenSimultor. Content that has been created in Second Life can also be transferred to a virtual environment created in OpenSimulator. It is possible to make use of this open source solution by either creating your own virtual environment on your own computer servers, or to make use of a so-called OpenSimulator grid that hosts virtual worlds offered by third parties.
When you create your own virtual world based upon an open source solution, it will make you less dependent on third parties. In the end, you can always build and run your own environment by using the open source software. This independence, however, must be balanced against the effort, know-how, and investments needed to set up and run your own virtual environment. Another important issue to be considered in this context is the number of residents. If you set up your own virtual environment, initially this world is not inhabited. This can be a positive if you are planning for limited-scale laboratory experiments. However, if the set-up of your research requires large numbers of participants, the selection of an existing virtual world that is populated with huge numbers of residents might be a better choice.
Gathering data and conducting some experiments may be difficult in virtual game environments. It is difficult to gain access and work with data that a virtual game provider may share. A group of researchers worked closely with Sony to examine data structures of the servers, develop a data dictionary, and map the “in-game” actions, transactions, and interactions of game players (Williams, Contractor, Poole, Srivastava, & Cai, 2009). If access to information is gained, the data still needs to be organized as data logs do not identify groups, only characters, causing researchers to have to reconstruct groups as people in the same place, at the same time, and then link these data to character type, guild, and what experiences occur while in the group. Conducting field experiments in the virtual game can also be problematic as researchers do not want to do anything that is part of an experiment that could ruin the experience of a naïve game user. If a researcher is interested in how game players respond to unsociable behavior (e.g., aggression, abuse, rejection) or any behavior that could be irritating even to some, then conducting a study in an online world would not be the appropriate venue. Game players might complain and, moreover, game hosts and players may turn on researchers using these virtual environments.
Ethical concerns when using online virtual worlds
Risks associated with recruiting participants
There are some unique ethical challenges with recruiting participants online. The first of these begins with the online games themselves. Online games may have terms of service (TOS) that may state explicit rules on what studies would be allowed to be conducted. What is tricky is that TOS are selectively enforced. Researchers must also address ethical concerns with gatekeepers of certain areas of virtual games or the game designers themselves. The gatekeeper is someone who must be approached first before anyone else in the group can be addressed. Getting the gatekeeper's permission to conduct research is often needed. When recruiting participants it is important to depict yourself accurately. If you are a male, it could be considered unethical to depict yourself as a female to join a group of females for study purposes. Often researchers will put in their profile they are a researcher, or better still, they will create an avatar that depicts a researcher (potentially wearing a white lab coat). Getting informed consent may also be difficult. In some cases Institutional Review Boards (IRBs) may accept a click-to-accept consent form. Ultimately, it is best to consult with member of your institution's IRB before conducting extensive online research.
The ethical concerns for a virtual laboratory remain the same as the ethical concerns for traditional laboratory experiments. Ethical guidelines for this type of research are best guided by the American Psychological Association's (1992) ethical code of conduct. An additional ethical concern for both laboratory set-up in virtual environments and research conducted with naïve game players in virtual environments is related to using avatar representations of participants. If participants are put in a situation in which they feel uncomfortable during an experiment, it may be particularly difficult for an experimenter to identify a participant in distress based on the actions of an avatar. In a traditional laboratory experiment, the experimenter is able to interact face-to-face with participants and monitor social cues of distress. Participants may either hide their emotions or be unable to express them using their avatar, leaving an experimenter blind to any undue distress they cause participants.
To counter this ethical concern, additional safeguards might be established when using avatars. One might be putting a video camera on participants using an avatar in a study. Experimenters can watch the reactions of participants: re-establishing the ability to monitor social cues of distress (also recording nonverbal behavior). Additional directions at the beginning of an experiment might emphasize what participants should do if they feel distressed. Alternatively, participants can have their avatar leave the experiment if they become distressed, similar to if they were physically in an experiment and desired to leave. For those experiments where participants interact with naïve players, ethical guidelines indicate researchers cannot cause stress that is greater than everyday life experiences without obtaining the participants’ consent.
Researchers observing or interacting with naïve participants in online virtual environments must consider a host of ethical issues that are unique from experiments conducted in the laboratory. Kraut and colleagues (2004) discussed ethical issues associated with collecting data online in the form of experiments using questionnaires, analyzing chat room comments, investigating posting on bulletin boards, group forums, or email distribution lists.
We will discuss these issues as they relate to doing observational research or field-like research within virtual environments.
Privacy
Making sure game players remain anonymous after being observed or following an interaction with the experimenter is central to reducing risk. When game players cannot be identified directly or indirectly, then this observation of public behavior is exempt from federal regulations protecting human subjects. If chat-room content or the responses of players that were interacted with are going to be analyzed, follow the suggestions of Kraut et al. (2004) concerning the use of pseudonyms and alter quoted text so it cannot be traced back to the original author. Game players do not want to have the sense that “Big Brother” is watching over their game play. Federal regulations dictate that electronic communication between individuals is not protected if the means of communication are readily accessible to the general public. However, to the advantage of researchers, norms in online games may include observing the actions of others and reading content that is posted in a chat room (e.g., Nonnecke & Preece, 2000). Lastly, game players may feel that what happens in the online world stays in the online world. Unbeknown to game players, gaming websites record essentially every move the players make. While this information is recorded to aid programmers in enhancing the experience of the game, to researchers’ benefit, it can still be used for research purposes. Researchers can minimize harm by securing data acquired from game hosts and by de-identifying game players.
Risk associated with the experiment
The risk associated with any experiment conducted in an online world will be particular to the type of study being done. There are several initial restrictions on what type of studies can be conducted in online virtual game environments. First, investigators should not deceive participants in online virtual game environments research. In the case of Cyberball, Kip and Jim deceived participants in that they believed they were playing online with others, but because Cyberball is run in a laboratory setting, they were able to debrief participants fully. Recently, the American Psychological Association stated it would not publish research done online that involved deception due to concerns about participants being properly debriefed. Second, research using online worlds needs to be of minimal risk if an informed consent process is not used. Studies are considered minimal if the extent of harm or discomfort a researcher believes participants will experience is no greater than that during daily life.
Conclusion
The possibilities for researchers to study group behavior will only increase as virtual game environments expand in size, scope, and complexity. Virtual game environments can be created to examine a specific group behavior thoroughly. Even the simple virtual environment of Cyberball has led to researchers throughout the world downloading the program and running any number of studies. Group behaviors that a researcher could never imagine studying are now possible to explore in detail. By examining different worlds in a virtual game, by following groups over long periods of time, or even observing groups in action slaying a monster, researchers can learn about global group phenomenon and long-term group behavior. Using online game-scenario virtual worlds overcomes the limitations associated with a traditional laboratory. Regarding traditional laboratories, virtual laboratories, situated in virtual game environments, may revolutionize conducting research. Owing to virtual laboratories overcoming many limitations of group-based research, being easy to create, and requiring a minimal amount of resources, anyone can create a virtual laboratory and have access to a diverse group of participants. Virtual games will increasingly become a central feature of our everyday lives and an increasingly popular source for finding others with whom to interact. As this occurs, research will also increasingly become a greater part of the virtual online environments, opening up countless new ways to investigate how groups behave.
Notes
- 1 Based on statistics from December 2008. http://www.wired.com/gamelife/2008/12/world-of-warc-1
- 2 https://blogs.secondlife.com/community/features/blog/2009/04/16/the-second-lifeeconomy–first-quarter-2009-in-detail
- 3 Based on statistics from April 2008. http://secondlife.com/whatis/economy_stats.php; http://www.activeworlds.com/info/index.asp
- 4 IBM developed a virtual business center in Second Life to host virtual meetings and learning sessions (http://www.ibm.com/3dworlds/businesscenter/us/en/) and Philips Design set up an immersive virtual environment in Second Life to support participation and co-creation activities in new product design (http://www.design.philips.com/sites/philipsdesign/about/design/designnews/pressreleases/secondlife.page)
- 5 http://secondlife.com/whatis/#Welcome (accessed: November 26, 2009)
- 6 Cyberball is freely available for download. The website for downloading Cyberball is: http://psyclops.psych.purdue.edu/~kip/Announce/cyberball.htm
References
American Psychological Association (1992). Ethical principles of psychologists and code of conduct. American Psychologist, 47, 1597–1611.
Axelrod, R. (1980). More effective choice in the prisoner's dilemma. Journal of Conflict Resolution, 24, 379–403.
Axelrod, R. (1984). The evolution of cooperation. New York : Basic Books.
Bainbridge, W. S. (2007). The scientific research potential of virtual worlds. Science, 317, 472–476.
Baron, R. S., Kerr, N. L., & Miller, N. (1992). Social dilemmas. In R. S. Baron, N. L. Kerr, & N. Miller (Eds.), Group process, group decision, group action. Pacific Grove, CA: Brooks Cole Publishing Company.
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529.
Benford, S., Greenhalgh, C., Rodden, T., & Pycock, J. (2001). Collaborative virtual environments. Communications of the ACM, 44, 79–82.
Blascovich, J., Loomis, J., Beall, A. C., Swinth, K. R., Hoyt, C. L. & Bailenson, J. N. (2002). Immersive virtual environment technology as a methodological tool for social psychology. Psychological Inquiry, 13, 103–124.
Castronova, E. (2006). On the research value of large games: Natural experiments in Norrath and Camelot. Games and Culture, 1, 163–186.
Cialdini, R. B., Vincent, J. E., Lewis, S. K., Catalan, J., Wheeler, D., & Darby, B. L. (1975). Reciprocal concessions procedure for inducing compliance: The door-in-the-face technique. Journal of Personality and Social Psychology, 31, 206–215.
Eastwick, P. W., & Gardner, W. L. (2009). Is it a game? Evidence for social influence in the virtual world. Social Influence, 4, 18–32.
Gonsalkorale, K., & Williams, K. D. (2007). The KKK won't let me play: Ostracism even by a despised outgroup hurts. European Journal of Social Psychology, 37, 1176–1186.
Hardin, G. (1968). The tragedy of the commons. Science, 162, 1243–1248.
Harrison, G. W., & List, J. A. (2004). Field experiments. Journal of Economic Literature, 42, 1009–1055.
Jones, E. E., Carter-Sowell, A. R., Kelly, J. R., & Williams, K. D. (2009). “I'm out of the loop”: Ostracism through information exclusion. Group Processes and Intergroup Relations, 12, 157–174.
Kraut, R., Olson, J., Banaji, M., Bruckman, A., Cohen, J., & Couper, M. (2004). Psychological research online: Report of board of scientific affairs’ advisory group on the conduct of research on the Internet. American Psychologist, 59, 105–117.
Loeffler, T., & Anderson C. E. (1994). What is virtual reality? In T. Loeffler & C. E. Anderson (Eds.), The virtual reality casebook. New York : Van Nostrand Reinhold.
Luce, R. D., & Raiffa, H. (1957). Games and decisions. New York : Wiley.
McGrath, J. E. (1984). Groups: Interaction and performance. Prentice Hall, College Division.
McLaughlin, M. L. (2006). Simulating the sense of touch in virtual environments: Applications in the health sciences. In P. Messaris & L. Humphreys (Eds.), Digital media: transformations in human communication. Peter Lang Publishers.
Miller, G. (2007). The promise of parallel universes. Science, 317, 1341–1343.
Nonnecke, B., & Preece, J. (2000). Lurker demographics: Counting the silent. CHI 2000, ACM Conference on Human Factors in Computing Systems, CHI Letters, 4, 73–80.
Olson, M. (1965). The logic of collective action: Public goods and theory of groups. Cambridge, MA : Harvard University Press.
Rheingold, H. (2000).The virtual community: Homesteading on the electronic frontier. Cambridge, MA : MIT Press.
Sabri, J. A., Ball, R. G., Fabian, A., Bhatia, S., & North, C. (2007). High-resolution gaming: Interfaces, notifications, and the user experience. Interacting with Computers, 19, 151–166.
Serviss, B. (2005). Escaping the world: High and low resolution in gaming. IEEE MultiMedia, 12, 4–8.
van Beest, I., & Williams, K. D. (2006). When inclusion costs and ostracism pays, ostracism still hurts. Journal of Personality and Social Psychology, 91, 918–928.
Verhagen, T., Feldberg, F., Van den Hooff, B., Meents, S., & Merikivi, J. (2011). Satisfaction with virtual worlds: An integrated model of experiential value. Information & Management, 48, 201–207.
Venkatesh, V. & Davis, F. D. (1996). A model of the antecedents of perceived ease of use: Development and test. Decision Sciences, 27, 31.
Wesselmann, E. D., Wirth, J. H., Pryor, J. B., Reeder, G. D., & Williams, K. D. (2009). Observing ostracism in groups: With whom do we side? Presented at the Annual Meeting of the Midwestern Psychological Association, Chicago, IL.
Widman, J. (2008). Boss by day, gamer by night: Tech leaders’ favorite video games. Retrieved from http://www.computerworld.com/s/article/print/9124143/Boss_by_day_gamer_by_night_Tech_leaders_favorite_video_games?taxonomyId=0&taxonomyName=Default
Williams, D. (2006). Why game studies now? Gamers don't bowl alone. Games and Culture, 1, 13–16.
Williams, D., Contractor, N., Poole, M. S., Srivastava, J., & Cai, D. (2009). The virtual worlds exploratorium. Manuscript in preparation.
Williams, K. D. (1997). Social ostracism. In R. M. Kowalksi (Ed.), Aversive interpersonal behaviors (pp.133–170). New York : Plenum.
Williams, K. D. (2007). Ostracism. Annual Review of Psychology, 58, 425–452.
Williams, K. D., Cheung, C. K. T., & Choi, W. (2000). Cyberostracism: Effects of being ignored over the internet. Journal of Personality and Social Psychology, 79, 748–762.
Williams, K. D., & Jarvis, B. (2006). Cyberball: A program for use in research on interpersonal ostracism and acceptance. Behavior Research Methods, 38, 174–180.
Wirth, J. H., Wesselmann, E. D., Williams, K. D., Pryor, J. B., & Reeder, G. D. (2009). When and why do we ostracize? To punish burdensome group members. Poster presented at the Society for Personality and Social Psychology Conference, Tampa, FL.
Wirth, J. H., Williams, K. D. (2009).”They don't like our kind:” Consequences of being ostracized while possessing a stigmatizing group membership. Group Processes and Intergroup Relations, 12, 111–127.
Xiong, L., Poole, M. S., Williams, D., & Ahmad, M. (2009). The effects of group structure on group behavior and outcomes in an online gaming environment. Manuscript in preparation.
Yee, N. (2006a). Motivations for play in online games . CyberPsychology and Behavior, 9, 772–776.
Yee, N. (2006b). The demographics, motivations and derived experiences of users of massively-multiuser online graphical environments. Presence: Teleoperators and Virtual Environments, 15, 309–329.
Yee, N., Bailenson, J. N., Urbanek, M., Chang, F., & Merget, D. (2007). The unbearable likeness of being digital: The persistence of nonverbal social norms in online virtual environments. Journal of Cyberpsychology and Behavior, 10, 115–121.
Zadro, L., Williams, K. D., & Richardson, R. (2004). How low can you go? Ostracism by a computer lowers belonging, control, self-esteem, and meaningful existence. Journal of Experimental Social Psychology, 40, 560–567.