simulations and virtual worldsVirtual worlds in educational research |
CHAPTER 19 |
This chapter introduces computer simulations and virtual worlds and discusses:
simulations and virtual worlds
theoretical bases of simulations and virtual worlds
applications of virtual worlds
a worked example of virtual world research
opportunities and limitations
issues and problems in virtual world research
using a virtual world and simulations in educational research
ethical issues in virtual world research
online tools for data collection from virtual worlds
A computer simulation is a representation of a real-world system. Simulations have two main components: a system of interrelated features in which the researcher is interested and that lends itself to be modelled or simulated, and a model of that system that is often a mathematical analogue (Wilcox, 1997).
In deterministic simulations all the mathematical relationships between the components of a system are known whereas in stochastic simulations, typically the main types used in educational research, at least one variable is random (Wilcox, 1997). A simulation is a model of the real world in which the relevant factors in the research can be included and manipulated. A model may operationalize a theory and convert it into a computer program (see Gilbert and Troitzsch, 2005: 3), making explicit its assumptions.
Gilbert and Troitzsch (2005: 6) suggest that the prime purposes of computer simulations are for discovery, proof and experiment. Beyond simply prediction, computer simulations enable an understanding and explanation to be gained of how processes operate and unfold over time, and the results of these. This explodes the value of prediction as a test of a theory; rather it argues that the test of a theory should be its explanatory and hermeneutic power, rather than its predictive value. Indeed computer simulations may be useful in developing rather than testing theories.
Virtual worlds are computer-based simulated environments in which the possibilities for developing and testing theories are enhanced by the agency of the participants, who join online communities and interact with each other.1 These worlds are also often persistent, in that they may continue to exist and develop with possible consequences for an individual user, even when that user is not present online. Virtual worlds commonly feature multiple users interacting with each other and the environment in a dynamic way. They share with many computer and video games (including games using virtual worlds) the deployment of realistic three-dimensional environments but, unlike them, their rationale does not have to lie in scenarios of ‘winning’ and ‘losing’. Nor do they need to have predefined objectives or rules of play, unlike, say, a computerized chess game played by participants or virtual worlds that exist to facilitate ‘questing’ or the ‘levelling’ of characters. The virtual worlds we are most concerned with here are created by the participants and the world emerges from the interaction of the participants. Participants create an ‘avatar’ for themselves, which represents them in the virtual world – this representation may take whatever form the participants wish (human, animal, object or creation of any sort) and may be changed at any time. Avatars are usually three-dimensional and interact with each other through communication and messaging, both private and public, both synchronous and diachronous.
In such virtual worlds individuals can project their own views and values on topics through their avatar and receive the feedback of others in the system. It is the participants who create their world and who affect each other. Such a projection technique enables authentic and honest views to be expressed and, where desired by the participants, to be modified. Topics can be both given and created – the environment is emergent – and they are sometimes for this reason described as ‘synthetic worlds’ (Castronova, 2005).
Pure (closed) computer simulations are characterized by:
modelling and imitating the behaviour of systems and their major attributes;
enabling researchers to see ‘what happens if the system is allowed to run its course or if variables are manipulated (e.g. to enable prediction);
a mathematical formula that models key features of the reality;
mathematical relationships that are assumed to be repeating in controlled, bounded and clearly defined situations, sometimes giving rise to unanticipated outcomes (Tymms, 1996);
feedback and multiple iteration procedures for understanding the emergence of phenomena and behaviours;
complex phenomena and behaviours derived from the repeated interplay of initial conditions/variables;
deterministic laws (the repeated calculation of a formula) sometimes leading to unpredictable outcomes.
By enabling the researcher to control and manipulate the variables and components, these simulations are useful in addressing ‘what if questions, e.g. ‘What happens if I change this parameter or that parameter?’ ‘What if the person behaves in such-and-such a way?’ ‘What happens if I change such-and-such a feature of the environment?’ The relevant elements are put into the simulation and are then manipulated – set to different parameters – to see what the outcomes are.
Bailey (2007) suggests that simulations have advantages such as:
economy (they are cheaper to run than the real-life situation);
visibility (they can make a phenomenon more accessible and clear to the researcher);
control (the researcher has more control over the simulation than in the real-life situation);
safety (researchers can work on situations that may be too dangerous, sensitive, ethically questionable or difficult in real-life natural situations).
In the field of education what is being suggested is that environments for learning, whilst being complex, nonlinear, dynamical systems, can be understood in terms of the working out of simple mathematical modelling. This may be at the level of analogy only (see Morrison, 2002a), but, as Tymms (1996: 130) remarks, if the analogue fits the reality then researchers have a powerful tool for understanding and prediction in terms of the interplay of key variables or initial conditions and a set of simple rules.
Computer simulations are powerful in that, as well as enabling researchers to predict the future (e.g. in economic forecasting), they also enable them to understand and explore a phenomenon. They can act as a substitute for human expertise, sometimes enabling non-experts to conduct research that, prior to the advent of computers, would have been the exclusive preserve of experts. Gilbert and Troitzsch (2005: 5) cite the example of geologists, chemists and doctors and also suggest that computer simulations are useful for training purposes (e.g. pilots) and, indeed, for entertainment. However, they underline (p. 5) the prime importance of computer simulations as being discovery and formalization of theory (i.e. clarity, coherence, operationalization, inclusion of elements and completeness of a theory).
However Bailey (2007) notes several reservations about computer simulations:
artificiality (they mimic life, rather than being the real thing);
cost (computer simulations can be expensive);
training of participants (simulations often require considerable training);
quantitative problems (they may require programming expertise).
To concerns that computer simulations subvert human agency and freedom and produce limited outcomes constrained within and determined by the assumptions on which they are built, we can respond that simulations can reveal behaviours that occur ‘behind the backs’ of social actors to illuminate social facts and patterns (Durkheim, 1956) and can therefore tell us what we do not know (Simon, 1996). The charge that they artificially represent the world and are a reductio ad absurdum is an argument for refining simulations rather than abandoning them.
Other concerns are:
The complexity and chaos theory underpinning many mathematical simulations might explain diverse, variable outcomes (as in school effectiveness research), but may not help us intervene to promote improvement (e.g. in schools) so explanation here is retrospective rather than prospective (Morrison, 2002a); however, researchers may be able to manipulate parameters and variables to see what happens when they do this.
How we ascertain the key initial conditions set into the simulation (i.e. construct validity) and how do simulations from these lead to prescriptions for practice.
How acceptable is it to regard systems as the recurring iteration and reiteration of the same formula/model?
Simulations work out and assume only the interplay of initial conditions, thereby neglecting the introduction of additional factors ‘on the way’, i.e. the process is too deterministic (that said, there are computer simulations in which the computer ‘learns’ during the simulation).
Manipulating human variables is technicist.
There is more to behaviour than the repeated iteration of the same mathematical model.
There will always be a world of difference between the real world and the simulated world other than at an unhelpfully simplistic level.
The agentic, moral and reflexive behaviour of humans is not as simple as what happens to simpler phenomena that have been studied in computer simulations (e.g. birds and ants, or the dynamics of sand dunes).
As with other numerical approaches, simulations might combine refinement of process with crudity of concept (Ruddock, 1981: 49).
If reality operates ‘behind the backs’ of players, where does responsibility (free will) for agentic actions lie?
Reducing the world to numbers is quite simply wrong-headed; the world is too complicated for numbers.
These reservations at conceptual and practical levels do not argue against simulations but for their development and refinement. Simulations promise much and in areas of the sciences apart from education have already yielded much of value.
Virtual worlds can offer closer approximations to real-life situations than closed simulations, including the handing over of control to the human agent projections (avatars) which enables researchers to explore hitherto less accessible or inaccessible territory without losing the benefits offered by simulations. Hence, to explore issues using virtual worlds may risk both exposing areas of conflict and ‘dangerous knowledge’ (Giroux, 1983) in areas such as community and religious identity, statehood, values and politics, majorities and minorities and ethnicity and identity. The creation of a safe environment is one means of addressing this. And how can this be done? One response is through projection techniques. Whilst some of the pedigree for this lies in personal construct theory, developments in information technology have enabled newer methods to be adopted.
First, it is through simulation (Cohen et al., 2007: 245-51). Simulations and virtual worlds offer the opportunity to explore situations, values and behaviours in a safe environment (though this is not guaranteed: for example cases are reported of student suicide as a result of cyber-bullying through social networking). Second, it is through agent-based modelling in computers.
The use of virtual worlds can overcome to a significant degree many of the limitations of, and reservations about, pure simulations by providing features that can be exploited to offer unique educational and research opportunities difficult to replicate in conventional contexts. Their resemblances to visually realistic, imaginatively designed three-dimensional game worlds makes them useful in stimulating participants’ imaginations, where the sense of being in a real place may be important as, for example, in creating greater engagement so as to convey difficult moral dilemmas which engage and retain the user’s attention. Creating a sense of life as a fictional experience may also be valuable in characterization exploration and development at one remove (i.e. via the avatar). Such facilities makes virtual worlds attractive places to develop otherwise inaccessible or impossible environments to explore human interaction and human agency, values and perceptions, historical locations and experiences, future imagined scenarios or uninhabitable settings such as inside highly ‘radioactive’ spaces.
These features often make virtual worlds more effective training environments than simulations, where the depiction of a hostile environment (say in firefighter training for casualty location) can be enhanced to offer more realism because not all parameters are controlled or known in advance. Similarly, otherwise impossible experiences can be offered in virtual worlds to capitalize on human agency, especially when participants are acting anonymously (where neither the user nor the individuals being engaged by them know each other’s identity), such as the experience of being immobile or disabled, or being a member of a different social or ethnic group.
The low risk and ‘repeatability’ of experience associated with virtual worlds offers advantages over models and mathematical simulations, not just by virtue of the safety they offer in environments characterized by essential unpredictability but also because, despite their creation of a heightened sense of realism over other approaches, virtual worlds afford the user discardable experiences which can carry relatively little personal, experiential or emotional cost (e.g. when training armed forces in decision making in pressured situations; training medics to treat ‘real’ casualties; or advancing views, proposals or identity depictions in hostile situations) (Waller and Hunt, 1998).
Virtual worlds also offer the participant control or even superhuman abilities such as the ability to fly. Together with the configurable avatar these features make virtual worlds good technologies for studying and affording rehabilitation experiences such as enabling those who have undergone traumatic experiences such as domestic violence to explore how to deal with situations in future, or offering opportunities for participants to make otherwise unavailable choices regarding actions, gender or personality. Their potential for exploiting role/real playing and the blending of real-life activity with virtual activity offers the researcher wide scope for exploring innumerable scenarios.
Working in artificial/simulated/virtual worlds such as Second Life, using projection techniques with networking and social interaction, feedback and iteration involves the creation and development of virtual worlds that are safe environments, in which the participants themselves are responsible for their creation and development. These are environments in which participants can project, share and comment on their own and others’ views, leading to the exposure and exploration of further, often sensitive, issues for investigation and possible resolution and to the resolution of potential conflict and disagreement and hence to the development and enlargement of understanding. Through the creation of safe, virtual worlds and communication in them, participants can be encouraged to be open, honest and authentic in disclosing their views, their values and their beliefs about given and created situations and issues, particularly about sensitive issues. A comparison of simulations and virtual worlds is made in Table 19.1.
The use of simulations and especially those identified as virtual worlds offer significant opportunities here for both practitioners and researchers. Exploring virtual worlds means that the researcher has to be open to a range of methodologies but has at the same time to re-examine the assumptions people may make when seeking to apply them to virtual environments. These environments offer higher levels of interaction and embodiment with participants than simulations constructed using purely mathematical models, and they offer unique features for data capture, including the ability to record sound, chat and text as well as images and video, all in real time.
In such scenarios we may need to modify, or at least revise, what we understand by ethnographic or qualitative approaches such as participant observation, focus groups and interviewing. We will need to examine the assumptions we make when deploying our methodological tools to verify that they are still applicable in these different communities, in which alternative values and different articulations of reality are to be found, in which participants may alter their displayed embodiment at will and in which we may simultaneously conduct our research with individuals in different locations around the world. In this way by exposing the researcher and practitioner to new constructions, expressions and transformations of identity, reality and community, virtual worlds offer unique opportunities to re-examine the nature of community and self in the virtual and the real world and also offer a unique opportunity to rethink, refine and improve existing pedagogy and research methodology and instrumentation.
TABLE 19.1 FEATURES AND AFFORDANCES OF SIMULATIONS AND VIRTUAL WORLDS
Features and affordances |
|
Simulations |
Virtual worlds |
Modelling/imitating |
Realizing/acting |
‘What if modelling of known variables |
Few known variables |
An underlying mathematical construct |
Minimal underlying constructs |
Modelled and interpreted reality |
Catching and manipulating the fine grain of reality |
Bounded, defined parameters |
Unpredictable outcomes sometimes |
Iteration and feedback to reveal emergent phenomena |
Human agency as the driver of emergence |
Repeated interplay of set initial conditions |
Any set initial conditions rapidly abandoned |
Unbounded and undefined parameters |
Unpredictable outcomes common |
Limited simultaneous users |
Multiple simultaneous users |
Transience |
Persistence |
For Laplace and Newton, the universe was rationalistic and deterministic, where effects were functions of causes and where predictability, causality, patterning, universality, linearity, continuity, stability and objectivity all contributed to a view of the universe as an ordered and internally harmonistic mechanism in a complex equilibrium. This was a rational, closed and deterministic system susceptible to comparatively straightforward scientific discovery and laws. Since the 1960s this view has been increasingly challenged with the rise of theories of chaos and complexity (see Chapter 1). Central to these theories are several principles (see also Chapter 1) (Gleick, 1987; Morrison, 1998,2002a):
small-scale changes in initial conditions can produce large and unpredictable changes in outcomes;
very similar conditions can produce very dissimilar outcomes (e.g. using simple mathematical equations (Stewart, 1990));
regularity, conformity and linear relationships between elements break down to irregularity, diversity and non-linear relationships between elements;
even if the mathematical equations are very simple, the behaviour of the system that they are modelling may not be simple;
effects are not straightforward continuous functions of causes;
the universe is largely unpredictable;
if something works once there is no guarantee that it will work in the same way a second time;
determinism is replaced by indeterminism; deterministic, linear and stable systems are replaced by ‘dynamical’, changing, evolving systems and nonlinear explanations of phenomena;
continuity is replaced by discontinuity, turbulence and irreversible transformation;
grand, universal, all-encompassing theories and large-scale explanations provide inadequate accounts of localized and specific phenomena;
long-term prediction is impossible.
More recently, theories of chaos have been extended to complexity theory (Waldrop, 1992; Lewin, 1993) which argues that, in many respects, the real world, though highly complex, is built on comparatively simple rules that give rise to such complexity (see also Gilbert and Troitzsch, 2005: 10) and the foundations of computer simulations including those we have identified as virtual worlds lie in complexity theory, providing a response to the charge that they oversimplify the real world.2
Determinism has given way to complexity theory and the ascendance of the latter is realized in virtual worlds where the inherent organic, unpredictable and irreplicable nature of the universe and human experience is mediated and celebrated. To the extent that virtual worlds are able to facilitate the creation of realities which hold true for participants and ‘make sense’ to them, these environments replace ‘real’ reality and meaning with symbols and signs to create a perceived reality (a simulacrum) which in some ways replaces human experience. For the educational researcher, virtual worlds offer opportunities to study identity, society and experience by means which do not replace the ‘real’ but afford means to better understand it by exploring it in alternate possibilities.
The use of media, either interactive or non-interactive, to externalize the self and create an impression of presence in an environment that we are not physically part of relies on the ‘willing suspension of disbelief (Coleridge, 1817) to create an ‘illusory shift in point of view’ (Dennett, 1978: 312) as well as the use of our own knowledge, imagination and enthusiasm in the experience (Zhao, 2003). Our sense of being present in the real world appears to be an essential component of consciousness but is not something we normally think about unless prompted by a displacement of our self-perception, for example through a dream, literature or cinema experience. Contemporary technology provides increasingly detailed and convincing simulations of reality and a sense of physical and/or social presence is now a commonplace feature of broadcast, cinema and virtual environments. Co-presence is the intersection of physical and social presence, where individuals have a sense of being together in a communal, shared environment.
A sense of presence is likely to be highly individual and conditional upon the sensory information presented within and the user’s level of control over the environment (Sheridan, 1992; Ijsselsteijn et al., 2000; Sadowski and Stanney, 2002). An immersive experience results when presence in an environment is augmented by its apparent overall fidelity to physical reality (Slater and Steed, 2000), although this does not have to be egocentric or computer mediated; it appears when ‘being there’ is augmented by a total response of ‘making sense there’ (Schuemie et al, 2001; Riva et al, 2003).
Agent-based modelling simulates or facilitates the actions and interactions of agentic (i.e. freely choosing, autonomous, intentional, purposeful) individuals in a network of other agentic actors, which leads to a range of effects on a whole system (Bonabeau, 2002). This is a key distinction between modelling environments using simulations and those found in many virtual worlds. The background for this lies in the work of Holland (1992), Ulam (1992) and Kauffman (1993)3 in developing computation-based and grid-based models to represent the actions of bounded rational agents in a system and to see how individual actions both pattern themselves and pattern the network. See, for example, the work of Reynolds4 (1987) on the behaviour of flocking creatures (‘boids’), where he identified three simple initial conditions (rules) which modelled behaviour:
keep a minimum distance from other objects (including other boids);
keep to the same speed as other boids;
try to move towards the centre of the flock.
It is important to note here that complex, patterned behaviour emerges from simple rules, and that the patterns cannot be understood entirely in terms of the initial conditions that gave rise to them, they cannot be decomposed or atomized into the initial conditions -the new whole is greater than the sum of the parts. It is more than an aggregate model. A new situation emerges from the interaction of agents and behaviours, giving rise to Durkheim’s social facts: ways of acting that can exert an external constraint over the individual or a society, and which have an existence of their own which is independent of their individual manifestations (Durkheim, 1982: 59).
What is being argued here is that macro-level behaviour emerges from micro-level agents; that is, complex, unpredictable virtual realities and norms emerge from the interactions of agents (individuals) within virtual worlds, just as they do between real people in real worlds. Agent-based modelling is therefore useful in exploring how individual behaviour and changes in individual behaviour contribute to new emergent patterns of behaviour in a system. These make virtual worlds useful technologies for research into human perception and interaction. Agent-based models place emergence through self-organization at their heart and in essence, therefore, a central feature of complexity theory underpins the field (see Chapter 1).
Whilst some agent-based models and forms of artificial life simulations have been criticized for oversimplifying the complexity that they are trying to model, and for being too mechanistic (e.g. in cellular automata)5, nevertheless it is a powerful emergent methodology for understanding larger scale, patterned behaviour, and it has been applied to fields as varied as: biology (Holland, 1992; Kauffman, 1993; Langton, 1984), political science (Axelrod, 1997); economics, cognition and neural networks (Sun, 2008); seasonal migration, pollution, sexual reproduction, wars, traffic jams, logistics, supply chains and transmission of disease (Epstein, 1996). The establishment of the European Social Simulation Association (www.essa.eu.org/) attests to the significance of such developments. Such models are less deterministic and more stochastic and emergent, faithful to central tenets of complexity theory.
The popularity of social networking sites indicates the significance of communication as a strong feature of contemporary society, where such communication leads to the emergence of key patterns, systems of thoughts, values and ideas. Communication via avatars offers a natural extension of such opportunities to circumvent traditional power games in face-to-face communication and are examples of the ‘privatisation of sociability’ (Castells, 2009: 389) that brings with it a greater likelihood that individuals will feel free to express themselves.
Such a model of communicative action accords with central tenets of the ‘ideal speech situation’ from Hab-ermas (1979, 1982, 1984, 1987). The principles of Habermas’s ‘ideal speech situation’ echo strongly through the view of exploration of issues through virtual worlds, for example (see the discussion of critical theory in Chapter 2) (Morrison, 1995a: 102):
freedom to enter a discourse;
freedom to check questionable claims;
freedom to evaluate explanations;
freedom to modify a given conceptual framework;
freedom to assess justifications;
freedom to alter norms;
freedom to reflect on the nature of political will;
mutual understanding between participants;
equal opportunity to select and employ speech acts;
recognition of the legitimacy of each subject to participate in the dialogue as an autonomous and equal partner;
equal opportunity for discussion;
the consensus resulting from discussion derives from the force of the better argument alone, and not from the positional or political power of the participants;
all motives except for the cooperative search for truth are excluded;
the speech-act validity claims of truth, legitimacy, sincerity and comprehensibility are all embodied.
Virtual worlds can provide environments that reflect ‘real-life’ activities and facilitate greater social inclusion, especially in situations exploring sensitive or contentious issues, where the anonymity and ‘ideal speech situations’ offered by the use of an avatar can facilitate participant engagement.
Virtual worlds also allow researchers to collect more participant data than are often possible in the traditional classroom, through capturing text, chat and user-presentations from the ‘inworld’ environment, and by videoing activity.6
By providing a scaffolded vocabulary, researchers can also more easily collect data from individuals with communication difficulties, where the (currently) reduced amount of data involved in communication in these environments (lack of subtle facial or body-language communication, for example) and the slower pace of communication can be advantageous (Raven-scroft and McAllister, 2006).
The communication in virtual worlds via text, chat, voice and signing can all be captured for later analysis via recording and transcription and, because data can be time-stamped, it is possible to compare the outcomes from analysis from these different communication channels for data triangulation (Martin and Vallance, 2008; Vallance and Wiz, 2008). These data can be converted to numbers for quantitative analysis or can be used to provide a richer understanding through interpretive phenomenological analysis, which is a form of qualitative analysis that is useful when we are interested in describing how people negotiate, understand and make sense of the world. Both quantitative and qualitative data need to be used in light of the perspective we wish to adopt, the kind of information we seek to collect and the assumptions that underpin our approach as well as what we wish to achieve (prediction or interpretation, for example):
Quantitative |
Qualitative |
An outsider’s view (etic) |
An insider’s view (emic) |
Researcher objective |
Researcher subjective |
Objective reality (facts) |
Subjective reality (reality socially constructed) |
Numerical deductive |
Naturalistic inductive |
approach |
approach |
Hypothesis-oriented |
Process-oriented |
Prediction |
Interpretation |
Reliable/generalizable |
Contextual |
Classification and summar |
Detail and richness |
Larger samples |
Smaller samples |
Faster |
Slower |
Virtual worlds are useful for the study of human interaction, especially in dynamic, fluid, uncertain or contested contexts, for exploring complex behaviour variables and for monitoring developments over time. By their nature virtual worlds – especially the perceptually realistic ones that are increasingly common – offer the researcher an opportunity to exploit the participant’s sense of immersion in the created world (a sense of being there) and also their sense of a shared experience with other individuals – a sense of presence and co-presence in a virtual world which responds to individual actions (a sense of being there). Both immersion and co-presence have been identified as important facilitators of user engagement in a time when media consumers demand more and deeper experiences (Riva et al., 2003; Boellstorff, 2008) through artefacts exploiting ‘the fluid boundaries between mechanism and flesh’ (Turkle, 2000: 555).
Virtual worlds are valuable for studying interactions between individuals especially when we wish to explore contexts where sensitive or dangerous knowledge is important and where it is important for subjects to feel safe to participate and respond candidly. Virtual worlds can offer ways of exploring behaviours and attitudes towards others in such contexts, such as when studying contested opinions or beliefs, or individual perceptions (including those of oneself) in relation to others and in exploring how these might change over time as a result of interaction with them. This makes virtual worlds useful places for researching the development of understanding, of perception, of processes where negotiated meaning is important and of the development of consensus and the dynamics of discord.
This section presents a worked example of using a virtual world as a safe environment in which researchers and participants explore sensitive issues in educational research and education itself. It reports how researchers can study how young people see themselves in relation to the society in which they live. It is based on work in progress at the time of writing, by Martin (2010), in exploring virtual worlds as safe environments for the study of sensitive issues, in this case of citizenship-related topics. Such a study might explore tensions between the internal image that individual young people have of themselves (their constructed identity) and the civic responsibilities and duties they are expected to discharge publicly, the loyalties which they are expected to defend and the behaviours which they are encouraged to display. A study of these elements of their ‘citizenship’ identity can explore: (a) individuals’ perceptions of, and attitudes towards, others; (b) their relationship to the dominant values, laws and legal systems in their country; (c) the social expectations of other individuals and groups; and (d) their experience of discrimination or disempowerment. Many countries have significant groups in their population who have different cultural, religious and social values and associated backgrounds. These pluralistic societies are often concerned about social fragmentation and political (dis)engagement and, in some cases, the marginalization and radicalization of groups within them. A study would therefore be very likely to touch upon sensitive issues, explore areas of ‘dangerous knowledge’ (that knowledge which is unsettling and which could provoke social unrest), and highlight social and individual tensions.
Using a virtual world in this context, as with any other approach, would involve developing appropriate research questions, deciding on the data that would be needed to answer them and determining how the required data could best be collected and analysed.
Suppose our research interest is in exploring what kind of society young people in secondary education wish to become part of, where the context is an economically developed, pluralistic democracy. Our research question might be ‘What do young people think it should mean to be a citizen in our country?’, assuming that its basis should be tolerance and the empowerment of individuals. We might also be interested in exploring how citizenship education might be developed. To address our research question we would need to know what the sampled young people associated with citizenship and how they defined ‘tolerance’ and ‘empowerment’. What values might they invoke and how could we discover how their definitions and understandings of these sometimes sensitive and contested issues emerged and operated?
In this study we might be interested in exploring the internal world of individuals and the way in which their values and feelings are externalized, projected, expressed and used, and how these drive behaviour. We might also wish to study how these can be shared and discussed in a safe environment. We therefore decide that we need not only to measure and ask our sample about the values and feelings they associate with a tolerant, empowered citizenship, but also that we wish to observe how these are deployed in practice, during discussion and negotiation with others, especially in situations where the values and acceptable behaviours might be contested. We might be interested in tracking emergent values and beliefs within discussions about citizenship.
Hence we decide to provide a number of inworld scenarios featuring ‘moral dilemmas’ of citizenship for our participants to discuss, so that we can observe how groups explore and discuss these, how different values and assumptions are brought into play and the relative weights that are accorded to each. Observing real-world group interaction in such circumstances would be unlikely, we conclude, to be ethically acceptable or practically useful because of the significant risk during discussion of self (or group) censorship, the influence of individual sensitivity to group expectations, individual’s anxieties about exposure when offering contentious views or arguments and the associated risk of conflict or the fear of it. We decide that even if ethical concerns about placing subjects in environments in which they may fear personal threat or harm could be overcome, our data would be subject to too many uncontrolled variables for them to be useful. Given these difficulties, we can better conduct our experiment in a virtual environment, and address and minimize many of the problems that could be foreseen with real-world groups. In a virtual environment the participants would be able to project their feelings, values, arguments and evidence anonymously via their avatar and interact with others in live (synchronous) debate to explore issues that arise, but with fewer anxieties about potential conflict, and greater confidence in sharing thoughts openly.
To operationalize our study and collect our data we therefore select an appropriate population for our study – here we focus on young people in secondary education. For practical reasons we cannot involve every such young person in our study, so we wish to select a fair sample. If the sample is ‘fair’ (i.e. representative of the population) we will be able to generalize (apply) our findings reliably back to the wider population from which our sample was drawn. This is not straightforward because, in order to be representative of the population, the sample should include as many features as possible from the wider population, in the same proportions as they are found in that population. In our study we would need to include males and females in the right proportion, and, within these groups, also the range of ages, cultural backgrounds, religious beliefs and everything else we might consider to be important – each in the correct proportions. To prevent this from becoming unmanageable, we recognize that, unless we are going to have a very large sample, we will have to be measured in the claims made from the study. We will not be able to generalize fully (unless everyone in our defined population is included) and so we should narrow our study – say to secondary school pupils in a particular age-group or year of study – and be explicit about the population to which we are trying to generalize. For us this means restricting the range of features covered in the sample and being clear that our conclusions will apply less and less widely as the contexts to which our findings are being applied become increasingly dissimilar to those sampled.
Our participants will complete a questionnaire which collects basic demographic information for use as the independent variables in our analysis. It also measures appropriate citizenship-related attitudes. A questionnaire is developed from focus group discussions with individuals representative of our population. Such discussions identify relevant key themes and items associated with demographic factors, identity, citizenship, tolerance and empowerment. For each theme (i.e. ‘tolerance’ or ‘the rule of law’) we design a number of Likert-style questions to sample individual views and feelings, and then ask the focus group(s) to assess each of these items for relevance, clarity and ‘fit’ to each theme. After any necessary modifications, our questionnaire is completed by participants pre- and post-experiment.
Then we construct a virtual environment similar to Second Life but host this securely on our own computers and allow entry only via a secure password system so as to address concerns about access. In our environment we have an ‘inventory’ of relevant items developed from focus group discussions. Avatars access the inventory to select and use items, but every time they must attach a value and associated definition. Values and definitions are stored in a ‘Values Dictionary’, which is augmented over time as more avatars are created and discussion of scenarios proceeds.
Within our experimental environment we ask participants to create an avatar, and attach an associated biography/back story. Participants are asked to make their avatar depict how they see themselves as a citizen, using clothing and artefacts drawn from the inventory. Avatars are provided with a dwelling which participants are asked to furnish and decorate with inventory items in order to create an ‘installation’ depicting who they are and what is important to them as a citizen. These data will be used to understand the items and attached values drawn upon and how they are defined in relationship to citizenship identity.
Groups of avatars are subsequently offered a series of contentious scenarios, each designed to highlight known areas of difference, tension or controversy within debates about citizenship. Participants’ avatars are invited to discuss these collectively and attempt to reach a consensus. Discussions are monitored discreetly to ensure civilized and courteous debates, but otherwise they are uncensored. Each avatar response must be accompanied by one or more attached values and the selection of an ‘emoticon’ that visually conveys to other avatars and their drivers the feelings attached to the response. Discussion continues until a majority view or impasse is reached.
Following each session, participants complete an online diary entry, unseen by other participants, where they talk about their feelings and views on the session. They are asked to reflect upon anything of significance they wish to identify, on suggestions for other scenarios for inclusion, on whether their experience would be likely to change their real-world behaviour, on how their experience within the experiment was different from the way in which they were normally taught about such topics in school, together with the advantages and disadvantages of each approach. These data are used to explore developing views and feelings as different scenarios unfold (e.g. views on how citizenship education might develop), and to discover topics for exploration during the experiment and exit interviews.
At the close of all the scenario discussions, a still image of each avatar is placed alongside its installation/dwelling together with its biography and attached values; each participant is then asked to review these and write a confidential note in their diary for each, in which they respond to these examples of citizenship identity. These data are used to gather reflections of others’ constructed citizenship identity for thematic and correlational analysis.
Participants also create a ‘Legacy Document’ in which they leave a statement for future users, in which is recorded what they have learned from taking part in the experiment and in which they offer advice that they wish to give to new participants in the experiment. These data are used to improve subsequent experiences of participants and as an indicator of personal change.
Participants then complete again the initial questionnaire and undertake a real-world face-to-face exit interview in which they are asked to reflect on their experiences within the virtual world and to respond to the anonymous notes left by other avatars about their installation/dwelling and their created avatar. These data are used as participant confirmation for the extracted inferences and conclusions, and to enrich the data set.
We may analyse our data via correlation, factor analysis or by applying item response theory or thematic analysis, to discover relationships between the independent variables (demographics and initial views and values) and any changes in values, definitions and perceptions of what participants think it should mean to be a tolerant, empowered citizen in our country (dependent variables). Statistical and thematic analysis will provide us with measures of internal validity for our findings. We will also use one or more control groups to which we will compare our findings from the experimental group, to help ensure that outputs from our analysis are not unduly skewed by the experience of the experiment and its features which are unrelated to the study focus or exposure to the environment we have created, and are not just outcomes that arise simply because individuals know they are being studied.7
We may wish to compare findings from several of our sources so that, by the triangulation of our data, we may strengthen internal validity8 and reliability in our conclusions. Demonstrating external validity9 is more difficult but we could provide increasing support for this by repeated use of our experiment with more samples from our population.
For further material on this project and its developments, readers can refer to the project website at http://web.me.com/stewartmartin2/Stewart_Martin/Citizen-ship.html">http://web.me.com/stewartmartin2/Stewart_Martin/Citizen-ship.html (see also Martin, 2010).
Simulation methods provide a means of alleviating a number of problems inherent in laboratory experiments. At the same time, they permit the retention of some of their virtues. Simulations, notes Palys (1978), share with the laboratory experiment the characteristic that the experimenter has complete manipulative control over every aspect of the situation, such as found in flight simulators, surgical simulators, training in dangerous environments and decision-making training, in which virtual world/reality applications are very varied and the experiences and skills involved can be simple, single, physical or abstract.
At the same time, the subjects’ humanity is left intact in that they are given a realistic situation in which to act in whatever way they think appropriate. The inclusion of the time dimension is another important contribution of the simulation, allowing the subject to take an active role in interacting with the environment, and enabling the experimenter to observe a social system in action, with its feedback loops, multidirectional causal connections and so forth. Finally, Palys (1978) observes, the high involvement normally associated with participation in simulations shows that the self-consciousness usually associated with the laboratory experiment is more easily dissipated. Hence, in summary, the advantages of simulations and the use of virtual worlds are:
experiential and active learning, so encouraging motivation and engagement;
visualization, managing complex environments;
access to impossible/difficult environments;
flexibility, can be programmed to offer a wide range of situations/stimuli;
monitoring, sessions can be recorded, examined -good for evaluation and assessment.
The advantages of virtual worlds over pure simulations lies in exploring contexts in which the experimenter neither desires nor is able to exert manipulative control over every aspect of the situation and where the researcher himself/herself desires the environment itself to be shaped by participants. Virtual worlds offer environments which enhance the agentic control and immersion of participants and therefore provide quantitatively and qualitatively different affordances for the researcher. Virtual worlds are highly suited to mixed methods approaches (with their inherent advantages and disadvantages), the development of research activity design, the exploitation of game theory and complexity/chaos theory and the general development of research methodology (Broadribb et al., 2009). Worlds such as Second Life and Club Penguin are examples of collaborative environments using ‘inclusive’ research practices in which researchers and subjects collaborate on equal terms and therefore offer opportunities to develop scenarios of ‘ideal speech’ (Rybas and Gajjala, 2007; Sheehy, 2010).
By their nature, virtual worlds lend themselves to projects which do not require participants or researchers to be physically located near to each other. Collaborative affordances are therefore just as important for the research team as for the participants, and being able to work together on analysis can be important for the research team. Synchronous and asynchronous researcher collaboration can be facilitated by technologies such as ‘Google docs’ or ‘Dropbox’ where individuals can share documents, spreadsheets and databases with colleagues and work synchronously on them to develop understandings, investigative approaches and dynamic trend-analysis, including motion-graphs such as those featured in Gapminder10 software.
Challenges can emerge when creating common research protocols which must function in institutions which apply different procurement constraints, use dissimilar and sometimes incompatible IT infrastructures and with policies with very different embedded institutional and cultural assumptions about the nature of academic roles and responsibilities (including those towards, and for, students and research participants). International research into the effective use of immersive virtual environments which, by its very nature, offers the opportunity to annihilate conventional frames of space and time, also involves successfully navigating a combination of real-world and inworld assumptions, limitations and obstacles whilst exploiting their many opportunities and resources for exploration, innovation and progress.
Research difficulties in using virtual environments also relate to having sufficient expertise in their use within both researchers and participants, and training in this may be a prerequisite. Having computers with appropriate specifications is important (commercial virtual worlds have advice on their websites), as is having sufficient bandwidth and safe passage through institutional network firewalls.
Participant enthusiasm is common but cannot be assumed and some studies have found to their cost that participants who are initially enthusiastic find that time constraints, the relative complexity of virtual worlds and bandwidth demands (which can create operational slowness) proved the biggest disincentive for continuing engagement (see the case study in lifelong learning conducted by Jarmon et al., 2009). Despite the visual and conceptual allure of virtual worlds, user acceptance remains one of the most significant challenges to be overcome (Fetscherin and Lattemann, 2007).
Evaluating virtual worlds will naturally focus on whether they achieve what is hoped for, but researchers should also seek unexpected outcomes and changes to practice or perceptions as a consequence of individual experiences of and within them (Lewis and Allan, 2005). Less often considered, but equally important, should be comparisons with existing practice, such as asking about ways in which the use of the virtual world has produced outcomes that could not equally well have been achieved by other or more conventional means.
The most effective use of a technology in a given context (for example in research or for teaching and learning), whether paper-based, physical, electronic or virtual, is always contingent upon exploiting its uniqueness – what it can do or provide which other means or technologies cannot. Using a technology to simply replicate existing practice or to copy what other technologies do is an example of cultural reproduction or ‘first order change’ which is unlikely to produce a fundamental or sustained difference (Cuban, 1986, 2003; Fullan, 2007). It also does not represent genuine innovation and is likely to produce overly elaborate, less effective and efficient, and often more expensive, outcomes than the alternatives. So we may ask how educators are using virtual worlds and whether they are doing so as a result of novelty, personal enthusiasm, to aid student motivation (which may decline with familiarity) or because they have identified some unique affordance within this technology.
In addition to the usual decisions to be made when designing a research project, if researchers conclude that a virtual world or simulation will be both necessary and useful for their project then they will additionally have to set up and manage the environment they wish to use. Depending on their experience and proficiency they may decide to involve someone with technical expertise to help them choose whether to make use of an existing commercial product11 or to create a purpose-built environment.12 The former is more straightforward but the options available for customization may be more limited, so it is important for researchers to be clear in advance which features they require for their study, what data they wish to collect and how. A customized platform may be the ideal but often it will be better to use a commercial product first, starting simply and adding complexity and customization as needed and proficiency grows.
Virtual worlds offer both new opportunities and new obstacles for research. Whilst researchers can use many familiar research tools, they also need to have familiarity with the virtual world’s technology and may find arranging the research venue more complicated and disseminating outcomes to the inworld community problematic, especially if using handheld devices.13
User engagement and scheduling of synchronous activity can also be difficult, due to time zone differences, technology differences and/or incompatibilities between user sites, and in terms of participant availability and technical skill.
When considering the design of educational research projects in virtual worlds, a number of factors should be taken into account (Moschini, 2010):
Type of project: exclusively inworld, blended, distance-learning or comparative (i.e. where similar activities will be conducted inworld and in the physical world for comparison).
Carefully define the activity to be investigated (e.g. focus on skill acquisition, collaborative work or communication).
Carefully formulate the research question and the appropriate theoretical background.
Venue: a ‘closed’ private environment or an ‘open’ space (such as in the public areas in Second Life).
Participants: do they know (or will they get to know) each other in the physical world or is the project an exclusively inworld experience?
Methodology: will evaluation use only inworld tools or will a blend of online and traditional research methods be used?
Ethical issues: is the project complying with the virtual world’s ‘rules of behaviour’ and with the researcher’s institutional guidelines for online research?
Data analysis: the tools to be used.
Dissemination: identification of the best channels for dissemination to both the inworld and wider academic community.
Deciding on the appropriate methodology for using virtual worlds or simulations in educational research involves considering offline and online tools (Fielding et al., 2008; Markham and Baym, 2008) and, as a result, mixed methods/mixed worlds approaches are common (Johnson et al., 2007; Martin et al., 2010), although these can exacerbate some problems, especially of maintaining user engagement and support (see the case study in Livingstone and Bloomfield, 2010).14
Exploring virtual worlds is itself a sensitive issue and it bears the hallmarks of sensitive research, i.e. that research which ‘potentially poses a substantial threat to those who are involved or have been involved in it’ (Lee, 1993: 4), or when those studied view the research as somehow undesirable (Van Meter, 2000), a feature which Morrison (2006) notes particularly for totalitarian regimes. Sensitivity is dealt with in Chapter 9, and readers should refer to the list of sources of insensitivity contained in the opening pages of that chapter.
The main ethical issues here concern vulnerability, individual risk and informed consent, especially when dealing with: online identities (when individuals may (re)present themselves as machines, real or fantasy creatures, plants, objects, etc.); the nature of communication (is it public or private); security; confidentiality and privacy and the inworld standards and rules.
Virtual worlds present both new opportunities and challenges when collecting data, even where the data are then to be processed by conventional means. Lewis and Allen (2005) emphasize that collecting data from an online or virtual community requires the researcher to be clear from the start about what data are needed, so that they can be gathered throughout the project, preventing the need for a major collection task at the end. Reflective diaries are useful sources of information, thoughts, feelings and ideas, and social network and group communication software can provide tracking tools. These collection techniques can be supplemented by online survey tools such as the Bristol Online Survey, Survey Gizmo, Survey Monkey or Zoomerang.15
Surveys and questionnaires are relatively simple data collection devices but can be time-consuming to design and always require careful forethought and planning in terms of exactly what analysis is to be done with each piece of information gathered. Deciding in advance exactly how each piece of gathered information will be used in subsequent analysis helps ensure that all (and only) information that is required is collected.
Virtual worlds offer possibilities for collecting data over time in ways that may be more difficult by other means. As such they offer opportunities to conduct research in ways and on topics that have proved problematic in the past, particularly those where extended interactions are important. One of the challenges for the researcher is therefore how to interrogate data that are chronological or sequential in nature. Dynamic data depiction and analysis tools are becoming ever more important in such areas. When working with dynamic data (data where values change over time in ways not yet fully understood), it can be difficult for the researcher to identify the correct questions to ask, let alone identify which are the appropriate variables to analyse; here traditional static graphs, charts and tables may be insufficiently informative. One example of a valuable tool for researchers to use in working with dynamic data can be found in Gapminder software.16 Gapminder was developed to explore trends over time, and it is useful for contextualizing data where chronology is important but where the relationships between variables over time are little understood.
In many virtual worlds inworld events and objects may also be connected to other online technologies such as blogs, wikis, questionnaires, rating-systems, databases, etc. and increasingly to inworld tools such as questionnaires.17
This chapter has argued that emergent, contentious or sensitive topics can be usefully explored using safe virtual environments applying an IT-facilitated form of projection technique, and this has been explicated in terms of computer-based simulation and agent-based modelling. These, it has been argued, are rooted in complexity theory which argues that new systems and patterns emerge through self-organization, through the effects of the parts/individuals on the emergent whole and on the parts/individuals themselves, and on the impossibility of understanding the whole as simply a sum of its parts. The process of emergence, from micro to macro and from macro to micro, requires adaptability, open systems, learning, feedback, communication and connectedness in order that individuals and systems develop their own identity (autopoiesis).
Further, this chapter has argued for the importance of including developments in behaviour and understanding behaviour such as understanding of networks and social networking, agentic behaviour and agent-based modelling and models, the micro-macro link and emergence through self-organization. How can such an understanding be addressed in research terms, whilst respecting the sensitivity of the research topic and embracing the safe environment of a computer-generated world? One way is through the creation of a virtual world in which agents create their own world and interact with each other in the creation of that world. This is akin to the stochastic simulation mentioned at the start of the chapter, which is emergent, unpredictable, but real, in that real people create the world and, within it, project their real opinions, values, identity and views.
Companion WebsiteThe companion website to the book includes PowerPoint slides for this chapter, which list the structure of the chapter and then provide a summary of the key points in each of its sections. This resource can be found online at www.routledge.com/textbooks/cohen7e.