CORDIS - EU research results

Understanding Collaboration in 3D Virtual Environments

Final Report Summary - UNCOVER (Understanding Collaboration in 3D Virtual Environments)

The UNCOVER research project aimed to advance our understanding of group interaction processes in avatar-based three-dimensional collaborative virtual environments (3D CVEs), and why virtual teams succeed or fail in using these novel collaboration technologies effectively.

Five research objectives were formulated regarding

(1) a task-technology fit (i.e. what types of tasks 3D CVEs are suitable for),
(2) a person-technology fit (i.e. individual differences in preferences and use of 3D CVEs),
(3) perceived presence (i.e. identification with one's avatar and its behavioral manifestations),
(4) interpersonal perception (i.e. the role of avatar appearance and group diversity), and
(5) e-leadership (i.e. determinants of emergent leadership in virtual teams).

Three large-scale international field studies were conducted in the context of a global education project called the ShanghAI Lectures. More than 200 students from 10-15 universities word-wide participated in the ShanghAI Lectures every year. They collaborated on international group exercises as avatars in a 3D CVE, which provided a unique opportunity to study virtual team collaboration in a cross-cultural context. An experimental laboratory study was conducted in addition to the international field studies for a systematic evaluation of the benefits and shortcomings of 3D CVEs.

The UNCOVER project was unique regarding its mixed-method approach combining survey data and behavioral observation, which allowed for both quantitative and qualitative analyses. We developed tracking systems that made it possible to continuously track in-world behavior (including avatar communication and navigation) in an unobtrusive way. We further explored the use of software bots (i.e. computer-controlled avatars) as a novel method for automated data collection within a virtual world.

Our original research question of which medium is most suitable for which type of task had to be rephrased as: How do collaboration tasks and the 3D environment need to be designed in order to create an added value for using virtual worlds as a collaboration tool? Using 2D objects in a 3D CVE requires a high design effort but does not necessarily create an advantage for virtual team collaboration. The tasks should be designed in a way that makes use of its distinct features, that is, dynamic interactions between avatars, 3D objects and environment. 3D interaction modes have been found to be particularly beneficial for brainstorming and decision making tasks, while mere information sharing did not require a 3D space and avatar embodiment.

We validated a self-presence questionnaire that measures the level of participants' psychological connection to their avatar. Self-presence was positively correlated with group identification and cohesion, but was not correlated with group performance. We found differential impacts of self-presence on work satisfaction depending on the extent to which an avatar is customized or personalized. These findings indicate that avatar design choices need to be considered in order to create optimal conditions for virtual team collaboration in 3D CVEs.

Moreover, individual preferences seem to play an important role regarding the usability and acceptance of virtual worlds as collaboration tools. The most prevalent factor determining 3D CVE design preferences was participants' cultural background. Europeans preferred an individualistic appearance of avatars, while Asians were in favor of displaying group-related attributes, which reflects the values of individualistic and collectivistic cultures. Asians stressed the importance of nonverbal communication (NVC) in virtual team collaboration, and requested more advanced options to perform nonverbal behaviors through avatars. NVC was of less importance to Europeans who stated that verbal communication was sufficient for effective collaboration. This reflects the preferred communication styles of high and low-context cultures in virtual environments. We conclude that sociocultural values are transferred into the virtual world and manifested in different design preferences. These findings call for a culturally-sensitive design of virtual environments used for global virtual team collaboration.

Similar results have been found regarding the persistence of cross-cultural differences in the use of personal (virtual) space. Avatars controlled by Asian participants maintained larger interaction distances than European avatars, which is consistent with observations in physical environments. These transfer effects of sociocultural conventions from the real into the virtual world are interpreted as a behavioral indicator of perceived presence. Participants indeed appear to experience a feeling of 'being there' (even in such non-immersive desktop-based virtual worlds), as they respond to actions inside a virtual world as if they would in an equivalent physical environment. This has important consequences not only for the use of virtual worlds as collaboration tools. It also provides evidence for the validity of using virtual worlds as a tool for intercultural communication research.

One of our main goals has been to uncover predictors for the success of global virtual teamwork. Neither the strength of emergent leadership nor the distribution of leadership roles among team members was indicative of a team's performance. The composition of team members regarding their nationality (i.e. level of cultural diversity in a team) also did not influence their performance. Instead, we found that the team members' average social sensitivity was a strong predictor of group performance. We obtained similar results for multicultural teams, in which intercultural sensitivity significantly predicted performance. In both social and intercultural sensitivity, 'interaction attentiveness' plays an important role, that is, participants' ability to pay attention to the subtle (nonverbal) cues in interpersonal interactions.

Humor has been found to be another key factor that influences virtual team performance. Humor moderates the relation between conflict experienced during collaboration in a 3D CVE and the team's performance. We distinguish between relationship conflict (i.e. tensions between the team members that are of socio-relational nature) and task conflict (i.e. conflict due to task-related issues). Relationship conflict was negatively related while task conflict was positively related to team performance. Humor seemed to add to the negative effect of relationship conflict. Team performance was highest when task conflict and humor were both high, and relationship conflict was low. Team performance suffered when both humor and relationship conflict were high.

In summary, the UNCOVER research has not only advanced our theoretical understanding of social interactions in virtual environments but also has practical implications for the design and deployment of virtual worlds as collaboration tools. The findings of the UNCOVER research project provide the basis for future research on how avatar technologies and virtual environments can be used to bridge between cultures and enhance intergroup relations.