Avatar Analytics

Mary Ellen Gordon

One thing I enjoy about research is seeing the extent to which perspectives on a topic vary among research participants, who often vary from one another in many ways besides their attitudes regarding the topic under investigation. I suspect many other researchers share my interest in trying to understand the range of opinions that surround issues but few are exposed to research participants who vary quite as much as mine: opposite me in an interview or focus group is often an avatar. I also spend a lot of time analysing survey data about avatars.

What is an Avatar?
The movie Avatar increased familiarity with the concept but just to be sure you know what I’m talking about, avatars are digital “stand-ins” for real life people in virtual worlds. By virtual worlds I mean not just anything that happens online, but a graphic environment similar to what you might see in an online game.

Avatars act as representatives of real life people in virtual worlds in the same way the race car and the top hat do in Monopoly. Avatars may look like whatever you want whether that’s like you, Barbie, a rabbit, or a coffee cup. They move about within the virtual environment but their decisions and actions are controlled by the real life person “driving” them via their computer. People do all sorts of things in virtual worlds with the help of their avatars. They go on quests, they listen to live music, they attend courses at real universities, they buy (real and virtual) things, and fortunately for me, they participate in research.

Research in Virtual Worlds
Market Truths, has been doing research in virtual worlds since 2006 and we operate a research panel in Second Life which is the largest social (non game-based) virtual world for English speaking adults. Some of the research we do relates directly to virtual worlds but other times we use virtual worlds as a venue for conducting research in the same way other researchers might use the phone, a shopping mall, or a focus group facility.

Conducting research with the virtual world population and using virtual worlds as a venue for research requires us to address issues that are familiar to most researchers but also present some unique opportunities and challenges. Working through what initially seemed like unique aspects of virtual worlds has also informed the way I view other forms of research that have nothing to do with virtual worlds.

Avatars and their Humans
When people began using the Web for research, some questioned whether it was possible to get “real” information when you couldn’t see or hear the person supplying that information. This sort of thinking is illustrated by the old cartoon with the caption: “On the Internet, nobody knows you’re a dog.” That type of reaction is perhaps even more prevalent with virtual world research because the avatar being interviewed might actually be a dog. By now most people are comfortable with “normal” Internet research and I believe that in time, the same will be true for research conducted in virtual worlds.

A question I get asked a lot is whether virtual world research participants respond on behalf of their avatars or themselves, but as with any research, careful wording of questions can remove ambiguity and ensure we get the information required. Besides being careful to specify whether we are asking about the avatar or the real life person and behaviour in the real or the virtual world, we face an added layer of complexity in that people may have more than one avatar in a given virtual world or spread across virtual worlds. We need to be cognisant of that when we ask questions and also keep in mind that the physical form taken by avatars is much more malleable than for humans. It’s possible to change size, shape, gender, or even species of avatars with a click of a mouse.

In practice, the dominant choice of avatar, at least in Second Life, is a human whose gender matches the real human; however, the fact that a given person can have multiple avatars and those avatars can easily change their physical form does present some sampling challenges. In particular, since our unit of analysis is typically the real life person and we are generally looking for independent observations, we had to develop systems to prevent the same real life person from participating multiple times using different avatars. When you think about it though, this is really just a different form of a challenge in all online research since it’s just as common to have multiple e-mail address as it is to have multiple avatars.

Another question I get asked a lot is whether the virtual world population is sufficiently representative to be useful for research pertaining to anything other than virtual worlds. Everyone seems to have a theory about how virtual world participants differ from “normal” people but in reality, virtual world users vary a lot from the stereotypes and in some cases defy them. For example, data we analysed recently, gathered from a sample that was demographically representative of the U.S. general population, showed that virtual world users are more likely to be employed than non users are and they also have a lower average body mass index. Virtual world users range in age from pre-schoolers to people in their seventies. More adult males than females participate in virtual worlds overall but the gap is narrowing and reverses for some social worlds.

Psychographics of virtual world users vary from world to world but, for example, the psychographics of Second Life users differ from those of the general U.S. population only in a few ways that can be explained by the nature of the world. (Highly competitive people are under-represented in Second Life and people who are interested in making social connections are over-represented. These percentages are likely to be reversed in game-based worlds such as World of Warcraft.) The psychographic differences (within the US population) that persist across worlds after controlling for other variables are religious affiliation and political ideology. Secular people are more likely to be familiar with virtual worlds than those who are religious, and self-described conservatives have less favourable perceptions of virtual worlds than those who describe themselves as moderates or liberals.

Perhaps the most noteworthy way in which virtual world users are unrepresentative of the general population (at least in the U.S., which is the only place we’ve done these general population comparisons to date) is in their media habits. People who participate in virtual worlds also tend to be heavier than average users of other forms of digital media and lighter than average users of traditional media. In particular, time spent in virtual worlds tends to substitute for time spent watching TV.

Real Work in Virtual Worlds?
Even if by now I’ve convinced you that “normal” people participate in virtual worlds and that it’s possible to get information from them about real life products and services in that environment, you may still be wondering: why bother? There are a variety of reasons. One is that virtual worlds offer the “anytime / anywhere” convenience of other forms of online research with the feeling of presence more typically associated with face-to-faceresearch. Most people find that having their avatar sit across the table from someone else’s feels more like a direct connection to that person than interacting online in other ways does. Virtual worlds share with other forms of online interaction the ability to track what people say and do in a great level of detail but they also have other capabilities that go beyond that. For example, research participants can manipulate objects in 3D in real time or easily move their avatars from one setting to another or within a given setting as needed.

The benefits just described make virtual worlds attractive not only for research, but for other “real world” endeavours too. For example, hundreds of colleges and universities from around the world use Second Life and its open source cousin, OpenSim, to conduct classes, educational activities, and research. Many real life businesses have also made use of virtual worlds for promotion, public relations, customer support, recruitment, training, meetings, and events. Well known brands that have undertaken activities in virtual worlds include:

Lessons Learned

In “Exodus to the Virtual World”, Edward Castronova argues that experiences playing online games will change how people perceive the real world and expectations they have in real life. My experiences doing research in virtual worlds bears this out.

For example, the importance of culture is often mentioned; however, in virtual world research there are at least two layers of culture to be aware of. One is the culture of research participants’ real life countries (our panel includes people from over eighty countries) and the other is the culture of the particular virtual world and / or sub-cultures within that. Failure to account for either type of culture may threaten the validity of research results, and awareness of the potential for that has made me more attuned to the possibility of cultural differences even when working with non virtual world samples.

Though they can be used for serious purposes, most people are attracted to virtual worlds because they are fun and interesting. To get people to spend some of their virtual world time participating in research we need to think about how we can make that fun and interesting. It turns out that making research fun enhances the quantity, quality, and depth of insights obtained. That seems obvious but it’s not something I thought much about before doing research in virtual worlds, and judging by some of the surveys that arrive in my inbox it’s not something to which other researchers give a lot of thought.

Perhaps because our research participants are highly engaged or because of the nature of the virtual world environment or culture, we get a lot of feedback on the research process as well as about the topic we are researching. We hear whether research participants had fun, but also things like whether or not the process felt authentic or if there were things they wanted to talk about that we didn’t ask about. While I haven’t completely solved how best to do it, this has me thinking more and more about how research can be more of a two-way conversation yet still produce rigorous, robust, results.

Last but not least, answering questions about virtual worlds has left me wondering why people don’t ask some of the same questions about more familiar contexts. For example, why do so many people perceive time spent in an interactive virtual world as wasted but don’t think the same about time spent passively watching TV? Or turning back to research, it’s fine to question the representativeness of virtual world samples but hasn’t everyone noticed that it is becoming increasingly difficult to get representative samples anywhere as media audiences fragment, many people no longer have land lines, and research refusal rates increase (due, perhaps, to an overabundance of poor quality research, highlighting the importance of taking care in research design regardless of venue)?

Things like audience fragmentation and social media are changing the nature of marketing and developments such as DIY surveys and Web analytics are changing the nature of research. Like it or not, that means researchers need to change too. I see research in virtual worlds as one of a number of ongoing experiments into Research 3.0 or whatever you want to call it. I don’t think anyone can say for sure what marketing or research will look like in another ten or twenty years but I believe that experimenting with new ways of doing research will help our profession continue to evolve.

Mary Ellen Gordon is managing director of Market Truths in New Zealand who specialise  in researching within and about virtual worlds.