Presentation is loading. Please wait.

Presentation is loading. Please wait.

CSCW – Module 4 – Page 1 P. Dillenbourg & N. Nova Module 4 : The effects of Media Richness Experimental Studies.

Similar presentations


Presentation on theme: "CSCW – Module 4 – Page 1 P. Dillenbourg & N. Nova Module 4 : The effects of Media Richness Experimental Studies."— Presentation transcript:

1 CSCW – Module 4 – Page 1 P. Dillenbourg & N. Nova Module 4 : The effects of Media Richness Experimental Studies

2 CSCW – Module 4 – Page 2 P. Dillenbourg & N. Nova F2F Vidéo Audio Chat Mail The number of communication channels Media Richness

3 CSCW – Module 4 – Page 3 P. Dillenbourg & N. Nova SMS WAP Media richness   Usage More bandwith makes not always better products

4 CSCW – Module 4 – Page 4 P. Dillenbourg & N. Nova Interwoven Dialogue Turns in a Chat 88.5r1Hpage sherlock but what about the gun? 88.8PrivS'Hercule which motive jealousy? He would have killed hans no? 89.3PrivS'Hercule he stole it when the colonel was in the bar 90.3r1Hpage sherlock Giuzeppe wanted to avoid that one discovers that the painting was fake. HSSH turns (from Pair 11, translated ) 43.5BarHWhy does Heidi have a motive ? 43.6BarSHow do you propose we should go further? 43.9BarHShould we merge our note books? 44.1BarSShe said that she didn't like her (and Hans) HSHS turns (from Pair 12 )

5 CSCW – Module 4 – Page 5 P. Dillenbourg & N. Nova Index of complexity: regularity in turn taking Bootnap experiments, Dillenbourg & Traum, 1997 0 ABABAB 1 0.9 MOO dialogue ? F2F dialogue

6 CSCW – Module 4 – Page 6 P. Dillenbourg & N. Nova The imitation bias is the false belief that a medium is more effective if it is more similar to face-to-face interactions

7 CSCW – Module 4 – Page 7 P. Dillenbourg & N. Nova Review of experimental research on media richness QuestionDoes a richer media increase groupwork? HypothesisExpected answer ??? Independent variablewhat do you vary? (or Factors)??? Dependent variableshow do you measure effects ? ??? Controlled variablesthings you try to keep constant ? ??? Intermediate variablesRelate IndependentVarianles to DependentVariables ???

8 CSCW – Module 4 – Page 8 P. Dillenbourg & N. Nova Chapanis and colleagues (1972, 1975) They conducted a series of laboratory experiments testing whether nonverbal cues like head nods, gazes or expressions help speakers to evaluate listener’s understanding and attention. They studied different problem-solving tasks, by looking at task outcome measures such as time to solution and quality of solution. The tasks involved complex instruction giving and route planning. –In one task, subjects had to jointly construct a mechanical object where one person had the physical components and the other one the instructions. –In another task, one person was given a map and the other one a copy of the Yellow Pages: they were asked to find a map location satisfying a number of criteria, such as the nearest dentist to a given street address. The research compared two media conditions: audio-only communication, and high quality audio-video, where the video showed the head and shoulders of the remote participant. The studies revealed that adding visual information did not increase the efficiency of problem-solving, or produce higher quality problem solving Laure Carles, Geneva Interaction, Lab

9 CSCW – Module 4 – Page 9 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Olson, J. S., Olson, G. M., & Meader, D. K. (1995). They studied 36 existing groups of three professionals involved in a design task. Groups worked at a distance (three people in three separate rooms), connected by this shared editor and either audio or audio and video links. The video preserved spatial relations and enabled eye contact. Therefore there were three conditions : 1/ remote with audio-only and a shared editor 2/ remote with a shared editor, and audio and video links 3/ co-located with the shared editor. o Impact on outcome With video, work was a good in quality as F2F. With audio only, it was nearly as good. o Impact on process of design and coordination The activities of the groups were the same in the three conditions. Video groups spent less time than audio-groups stating and clarifying the issues. Working at a distance, groups spent more time managing their meeting and clarifying what they meant to each other than F2F. o Impact on participants’ satisfaction Groups rated the audio-only condition as having a lower quality and reported more difficulties communicating. Remote work with audio, video and shared editor was judged to be as high quality as F2F. The authors concluded that perceptions suffer without video, and that work is accomplished in a slightly different manner, but that the quality of the work suffered very little.

10 CSCW – Module 4 – Page 10 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Veinott, E.S, Olson, J., Olson, G.M., Fu, X. (1999) They explored the effects of video channel on task performance in a visual map task for which it is necessary to buid common grounds. In this task, one participant must lead another participant through a route on a map of a treasure island without showing him his own map. Video helped non-native speakers but not native-speakers to get the common ground on this shared map task.

11 CSCW – Module 4 – Page 11 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Anderson, A. H., O'Malley, C., Doherty-Sneddon, G., Langton, S., Newlands, A., Mullin, J., & Fleming, A. M., & Van der Velden, J. (1997) They explored the performance of users on a collaborative task, the Map Task, with three conditions : video-audio channel with eye contact (using videotunnels), video-audio without eye contact and audio-only. They proposed a full evaluation of these systems measuring both task outcome and communication processes. They expected task performance to be comparable across conditions but task dialogues to be significantly shorter in VMC with eye contact (like in F2F). They found no difference in levels of task performance across conditions. However, dialogues within the VMC with eye contact were significantly longer than dialogues from the other two conditions. The researchers noticed an unusual level of gaze behaviour (more than double than recorded in F2F). They concluded that VMC with eye contact may encourage participants to “overuse” the visual channel, which may be counterproductive. Eye-Contact was however related to regulation phases

12 CSCW – Module 4 – Page 12 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Fussell, Susan R., Kraut, Robert E. & Siegel, Jane (2000) In these tasks, people weared mobile collaborative systems and had to repair collaboratively a bicycle. This study adresses two main issues: the effects of communication media on task performance and on the strategies collaborators use to ground their utterances during repair dialogues. Three conditions: 1.Side-by-side: worker and helper located in the same room 2.Audio-video: worker and helper connected via audio-video links 3.Audio only: worker and helper connected by full duplex audio only links In addition, they compared collaborations in which helpers are experts in bicycle repair to those in which helpers are, like the workers, novices. Unskilled workers performed three repair tasks on a ten speed bicycle with the assistance of either an expert or novice helper. Worker and helper completed the task more quickly (around 25% reduction in work time) and accurately when they were colocated than when they were connected via audio only or audio- video. The authors conclude that these results demonstrate the value of a shared visual work space, but raise questions about the adequacy of current video communication technology for implementing it.

13 CSCW – Module 4 – Page 13 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Point 1. Video has no major impact on task performance Compared to high quality audio-only systems, the presence of video does not have major effects on task performance, unless that work is inherently visual. Point 2. Video conversations are still more formal than Face-to-Face (F2F) but people prefer them to audio only conversations. Compared to F2F No video-conferencing system managed to replicate F2F communication yet, high-quality Video-Mediated Communication is still characterized by highly formal conversational behaviours compared to F2F. Compared to high quality audio Still, video makes meetings more satisfying for the participants by easing the mechanics of conversation, helping them understand nuances in meaning and mostly enabling them to track the remote participants’ presence and attentional state. One of the most obvious finding is that people like to see each other when they interact, especially when they do not know each other well: regardless of any cognitive benefit video may provide, people like having it. However, there is strong evidence that reducing audio quality to incorporate video is highly disruptive of conversation processes.

14 CSCW – Module 4 – Page 14 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Cohen, 1982 The author compared F2F versus PicturePhone Meeting Service. Six to eight participants sit around a conference table in a room equipped with two monitors, one for the incoming display and one for the outgoing display, and two speakers. Each of three cameras is focused on two adjacent positions around the table. The cameras are voice-switched to display either the current speaker or the former speaker. The author noticed more speaker turns and interruptions and twice as many speaker exchanges in F2F than in PMS. Participants prefered F2F for handling discussion tasks. The author suggested that video teleconferences are less interactive, more orderly and more polite than F2F meetings.

15 CSCW – Module 4 – Page 15 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Sellen (1995) Investigated whether video offers process cues to support turn-taking in a series of laboratory studies of negotiation tasks, in which groups tried to reach consensus. There was little evidence to support the claim that high-quality video information improves conversation management and turn-taking, when compared with audio-only conversations. For objective conversation process measures such as pausing, overlapping speech and interruption management, there were no process differences between the video-audio system and audio-only systems. Furthermore, none of the video-audio systems replicated F2F conversational processes. The video-audio systems reduced the ability of listeners to spontaneously take the conversational floor, as measured by number of interruptions. Video-audio systems led speakers to use more formal techniques for handing over conversational initiative, such as naming as possible next speaker or using “tag” questions”, when compared with F2F interaction.

16 CSCW – Module 4 – Page 16 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Why is video less than face-to-face ? (1: obvious items) A lot happens outside the picture Gestures, body language Distances Contextual cues Back discussions External events … The camera inhibits behaviors that contribute to conversation management: Head turning Gaze awareness Eye contact …

17 CSCW – Module 4 – Page 17 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Sellen, A. J. (1992) 12 groups of 4 participants who did not know each other before were tested in three informal debates. These debates were on social issues, like the right to smoke in public, with participants playing the roles of two “pro” and two “con”, and lasted 15 minutes each. She compared patterns of spontaneous speech in F2F and two VMC systems for multi-party meetings: PIP (Picture In Picture): four small pictures of the participants are simultaneously displayed on the same monitor. Hydra units: each participant picture is displayed on a separate camera which position in the room respects the relative position of the participants in order to support directional gaze cues and selective listening. The experimental hypothesis was that Hydra would tend to produce conversational patterns more similar to F2F than a PIP approach, which fails to support selective gaze and listening, and is designed so that one participant also sees himself on the screen. Two hypotheses: 1/ F2F will result in the highest number of turns per session, and PIP in the fewest. 2/ The average duration of turns will be shortest in F2F, and longest in PIP. Contrary to expectations, there was no differences between the two video systems. Subjects rated the same room meeting as allowing them to better take control of the conversation than both video conditions. Nonetheless, this difficulty was not reflected in the management of turns among speakers (which did not differ across conditions either in their distribution, number or duration), but in the amount of simultaneous speech and pauses (higher in F2F). Subjects rated the same room meeting as being more interactive and as allowing them to better attend to one person at a time than both video conditions. 2/3 of the subjects said they preferred the Hydra system, which enables them to track the attention of others. But 1/3 of the subjects prefered the PIP system, which enables them to have a global look at all participants, including themselves.

18 CSCW – Module 4 – Page 18 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab We survive to video problems more easily than to audio problems. Audio is often the bug.

19 CSCW – Module 4 – Page 19 P. Dillenbourg & N. Nova Belvedere I see what you see I see you >

20 CSCW – Module 4 – Page 20 P. Dillenbourg & N. Nova A B C VIDEO + AUDIO A B C AUDIO Question to A: Were you nervous ? 1 2 3 4 5 Question to B: Was A nervous ? 1 2 3 4 5 Perceiving my partner’s emotions: Is video better than audio ?.62.70 René Glaus (TECFA), oct. 2002

21 CSCW – Module 4 – Page 21 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab Marshall and Novick (1995) Two-party teams worked on a geometric puzzle with and without video. When video was available, they used it to show the puzzle pieces, and this took preference over use of the camera for anything else. When no video was available, participants were forced to give lengthy descriptions, which proved to be inadequate despite their complexity. Gaver, W. & Sellen, A. & Heath, C. & Luff, P. (1993) Users were offered a choice of video views, either of work objects or collaborators, in design tasks. Work objects were chosen far more often than views of collaborators. Facial views were chosen 11% of the time. Mutual gaze occurred 2% of the time. People were more likely to choose an image of the object, spending 49% of their time with the object views. I see you <> I see what you see

22 CSCW – Module 4 – Page 22 P. Dillenbourg & N. Nova Laure Carles, Geneva Interaction, Lab VideoSpace prototype (Roussel, 2001). NTT’s ClearBoard system (Ishii & Koyabashi, 1992)

23 CSCW – Module 4 – Page 23 P. Dillenbourg & N. Nova MEDIA SPACES PRIVACY

24 CSCW – Module 4 – Page 24 P. Dillenbourg & N. Nova Why distributed teams do not work as well as co-present teams even if you give them all advanced media ?  Next Module Why is video less than face-to-face ? (2: Mediating Informal Communication)

25 CSCW – Module 4 – Page 25 P. Dillenbourg & N. Nova Conclusions Imitation bias:“The richer the better” or “The more face-to-face like the better” have not been confirmed by empirical results. It’s difficult to invent something new and simple. There is no robust advantage of video over audio. Video is not (yet) a killer application. (tell it to phone operators) Mediated communication remains more formal than face-to-face


Download ppt "CSCW – Module 4 – Page 1 P. Dillenbourg & N. Nova Module 4 : The effects of Media Richness Experimental Studies."

Similar presentations


Ads by Google