|
Experimental Design |
|
¡¡ |
|
41 participants were assigned to undergo all three conditions (chat only, indirect avatar gaze, direct avatar gaze). The order in which the participants experienced the conditions was assigned randomly in order to satisfy the requirements of a repeated measure ANOVA. Order randomization is needed in order to distribute the effects of potential sources of error that stem from the use of a repeated measure design. Thus constant error is converted into a less significant random error (Dyer). Each participant performed a 3 minute negotiation role-play with the experimenter. This was done thrice under each different condition. Through the task, participants played the role of a customer with a complaint, while the experimenter represented a customer service representative. Avatar gender was always matched in order to achieve same sex dyads. This was done by manipulating the experimenter¡¯s avatar. |
|
Participants were Second Life¢ç players compensated with 250 Linden dollars (approximately 1 U.S. dollar) for their participation in the study. Participants were recruited through Second Life groups such as ¡°Science Friday¡± and ¡°SL Researchers.¡± Some participants were also recruited by word of mouth from previous participants. The 41 participants consisted of 18 male avatars, and 23 female avatars. The data from the first 7 participants were designated as pilot sessions. As such, the data collected from these participants were not used. Pilot sessions were necessary in order to train the experimenter for consistent implementation of procedure for all participants. Pilot sessions were also necessary so that the experimenter developed a consistent style of speech for the role-play. Thus, data from only 34 participants were used for analysis. There are 15 male avatars and 19 female avatars that make up this section. The Three Conditions (Independent Variable) Before describing the specific actions used to facilitate each condition, it is important to note that certain SL™ avatar behaviours were maintained for the use of this study. One such behaviour is the ¡°typing behaviour¡± exhibited when a user is typing. For this action, the avatar pretends to type on an imaginary keyboard, while lowering its gaze towards this keyboard. The other behaviour that was maintained is random eye blinks. 1. Indirect Avatar Gaze Condition: The experimenter¡¯s avatar is seated on the ¡°Experimenter Chair 1.¡± Indirect avatar gaze is maintained at Target Prism 1. This is done by pointing the mouse-look mode cross hair directly on the target prism. No gaze adjustments are made while in this condition. Based on the coordinates, this gaze condition looks about 30 degrees away from the participant. 2. Direct Avatar Gaze Condition: The experimenter¡¯s avatar is seated on the ¡°Experimenter Chair 1.¡± Direct avatar gaze is first initiated by pointing the mouse-look mode cross hair directly on the participant¡¯s face. When not typing, the experimenter¡¯s mouse-look mode cross hair was randomly moved from the participant¡¯s face, to either one of ¡°Target Prism 2¡± and ¡°Target Prism 3.¡± This was done at two second intervals, and this time was chosen based on previous research (Garau). Based on the coordinates, gaze conditions for both target prisms results into a gaze that is 6 degrees away from the participant. Triple nodding behaviour was also exhibited by the experimenter¡¯s avatar whenever the participant inputted a chat message. Direct avatar gaze was not maintained 100% of the time, since this may be interpreted as hostile behaviour by the participant (Argyle). 3. Chat Only Condition: The experimenter¡¯s avatar is seated on the ¡°Experimenter Chair 2.¡± Mouse-look mode is maintained by the experimenter. Both the experimenter and the participant have no view of each other due to a blind that is strategically placed between them. As such, the screen only shows chat history. Quality of Communication (Dependent Variable) Quality of communication, the dependent variable of interest, was defined similarly as a previous research paper (Garau). In this paper, quality of communication was divided into four categories. 1. ¡°Face to Face: The extent to which the conversation was experienced as being like a real face-to-face conversation.¡± 2. ¡°Involvement: The extent to which the participants experienced involvement in the conversation.¡± 3. ¡°Co-presence: The extent of co-presence between the participants – that is, the sense of being with and interacting with another person rather than with a computer interface.¡± 4. ¡°Partner evaluation: The extent to which the conversational subjects positively evaluated their partner, and the extent to which the conversation was enjoyed.¡± These definitional categories were quantitatively measured by utilizing a post-experiment questionnaire. The questionnaire was measured on a 7-point Likert scale, where 1 was defined as ¡°strongly disagree¡± and 7 was defined as ¡°strongly agree.¡± The questions were also based on previous questionnaires designed to ¡°elicit subjective responses to mediated communication¡± (Garau). Each question was also grouped under the four categories. The definition for quality of communication, and the questionnaire were adopted from previous research because they have proven to be valid and reliable. ¡°Validity deals with whether the questionnaire can provide the information that it claims to provide and reliability deals with whether it provides accurate and consistent data¡± (Dyer). ***The following is a
Windows Media Video
that shows the laboratory apparatus. Please allow about 3 minutes for it
to load. |
|
¡¡ |