Transcript Social Dynamics Can Be Distorted in Video

Social Dynamics Can Be Distorted in
Video-Mediated Communication
Wei Huang, Judy Olson, Gary Olson
Collaboratory for Research on Electronic Work (CREW)
School of Information, University of Michigan
1
Overview


Problems
Review of Aspects of Social Dynamics






Physical context
Proximity
Height
Experiment
Results
Discussion & Implications
2
Problems

Motivation – video technologies are not fully
accepted and satisfied, WHY?



Enthusiastic about video technologies
“Despite Advances, Video Still a One-Way Channel”
(American Banker; Dec 2000, Quinn)
Mixed results on VMC research


Video + audio add little audio-only communication on
cognitive tasks (Chapanis, 1975, 1972, etc)
Video helps when
 People have very little common ground (Veinott et al, 1998)
 Tasks involve negotiation (Short et al, 1976)
3
Theoretical perspective

Whittaker (1999, 1997) non-verbal communication
that visual channel supports



Cognitive cues to determine the other’s understanding;
 head nods, visual attention
Turn-taking cues to support conversation mgmt process;
 head turning, posture, eye gaze
Social or affective cues that reveal the other’s emotional
state and interpersonal attitudes;
 facial expression, posture, eye gaze
4
Theoretical grounding:
Interpersonal communication

First impressions formed based on outward
appearance cues


age, gender, race, ethnicity, body shape, height, dress etc.
This includes a rapid categorization process and an
activation of social schemas (Jones, 1990)



We make assumptions
Impressions formed later are reinforced or modified
Initial impressions form baseline comparison which help to
make proper inferences and causal attributions, which
affects the dynamics of interpersonal communication
(Heider, 1958)
5
Distortion in VMC

Video systems don’t provide the full array of visual
cues as in FtF interactions (Fussell et al, 1995)


Physical context is blocked out
 traditional “talking heads” VC systems only show the
shoulder and head and fill the image on the screen
Titled perspective and direct eye-contact impossible
 Desktop VC systems have to place the camera on either
top, bottom, the side of the monitor
6
This study

Extends Whittaker’s non-verbal communication


suggests that visual channel provides appearance and
physical context cues with which people form impressions
of others, which affect people’s behavior.
Focuses on three factors



Physical context cues on video
Interpersonal distance on video
Apparent height
7
Physical context

In FtF interaction



includes setting of human communication, objects
and people
Cues from the physical context help people
perceive precisely the size of an object (Walsh et
al, 1998).
VMC

Physical context is often blocked out (Mantei et al,
(1991)
8
Proxemics

Hall (1966) - a person’s structuring and
perception of space
9
Proximity in FtF interaction

Equilibrium of interpersonal distance (Argyle et al,
1960s)

Interpersonal distance is often influenced by age,
gender, race and culture



M-M > M-F > F-F (Rosegrant, 1973)
Arabian males keep very close distance (Hall, 1966)
Same ages < different ages (Pedersen, 1973)
10
Proximity in VMC

Different from that in FtF (Heath, et al, 1991)



Threat and occupation of space is attenuated
Accessibility of nonverbal cues varies according to camera
zoom
VMC prototypes


MAJIC: Too far away: “Can you hear me?” instead of
“Hello” (Okada et al, 1994)
Clearboard: Too close, not appropriate for strangers or
people at different level (Ishii et al, 1993).
11
MAJIC
12
Clearboard
13
Height in FtF interaction

Tall people are believed to enjoy higher status,
dominance and power (Ellis, 1994).

Height is positively related to social esteem, leader
emergence, performance, income. (Judge et al, 2004, to
be appeared in JAP)

Height - a predictor of social dominance and academic
success (Hensley, 1993)




Asst Prof. 1.24” taller,
Assoc Prof. 1.50” taller,
Full Prof. 1.97” taller,
Chair 2.14” taller than average Americans of the same age and
gender
14
Height/camera angle in video

Perception of height can be manipulated by
camera angles

TV/Film production


Low angle = Superior
High angle = Inferior
(Giannetti, 1973, 1999)
15
16
17
18
Purpose of this study

Examine apparent height effect in video-mediated
communication, emphasize




Negotiation: Group decision-making
Interaction and influence measures
Impressions of the others
We do that by


Manipulating the camera angle and monitor placement
Manipulating the camera zoom.
19
Hypothesis

Apparent height effect

The apparently tall person (that is created by artificially
placing the camera lower) is more influential in the
decision-making than is the apparently short person
(created by artificially placing the camera higher).

The visibility of the physical context (that is manipulated by
zooming in or out the camera) helps people to make
judgments of the other’s height, and thus the more the
context on the screen, the stronger the apparent height
effect.
20
Experiment Design

2 x 2 (between-dyad)

Monitor distance – one side: far (6 ft) vs. close (2 ft)
Camera zoom: in or out

Dyad: camera angle at high or low (+/- 28o)


Group decision-making: Arctic Survival Task




Crash landing in inhospitable Artic area
Expert ranking
Individual ranking of 15 items
Group ranking of 15 items
21
Experiment Design Cont.

Measures

Individual task influence
 the difference between one’s individual ranking and the
group ranking. So the lower the score, the higher the
influence

Perceived influence: Self-reported influence in post-test
survey

Impressions of self & the other - Dominance Scale
(Burgoon et al, 1992)
22
Experiment

Subject




White Americans
Male
18 – 35 years old
196 subjects
Distance
Zoom
Out
Far
Close
28 pairs
24 pairs
In
19 pairs
27 pairs
23
Distance
1 2
3 4
Zoom
Far
Out
12
In
34
Close
24
5 6
7 8
Distance
Zoom
Far
Close
Out
5 6
In
78
25
Task Performance
Group Expertise
74
66
58
far
50
close
42
34
26
Out
In
Zoom out or in
26
Individual Influence
Is ‘tall” more influential?
74
64
54
44
34
24
14
4
tall
short
far
close
far
out
close
in
Zoom x Distance
27
Perceived Influence %
Is ‘tall’ perceived more influential?
65
60
55
50
tall
45
short
40
35
30
far
close
far
Out
close
In
Zoom x Distance
28
What impressions were formed?
SELF
PARTNER
4.8
4.6
4.4
4.2
4
far
close
Out
far
Zoom x Distance
close
In
Rating of Dominance
Rating of Dominance
Apparently Short's Rating
SELF
Apparently Tall's Rating
5
4.8
4.6
4.4
4.2
4
3.8
PARTNER
far
close
Out
far
Zoom x Distance
close
In
29
Summary of Findings
Distance of subject from monitor
Camera zoom
Out
In
Far
Close
Tall > Short
Tall = short
Tall = short
Tall = short

When monitor was placed at far distance and camera was
zoomed out, there produced a significant height effect.
Hypothesis is partially supported.

Impressions - Both apparently tall and short persons rated
themselves more influential than they rated their partners.
30
Discussion

Finding


We did find that apparently tall person was and was
perceived more influential in the group decision-making
task only when more physical context was presented and
monitor and camera was placed farther away and higher.
Apparent height effect


Monitor & camera have to be placed farther away & higher
– apparently tall person has to be very tall.
Task-oriented communication reduced the use of visual
channel.
31
Implications and future work

May impact more directly on tasks like distant
interview over video where task itself involves
interpersonal perception and judgment

In the future, plan to conduct conversation/language
analysis and nonverbal communication analysis

Other factors such as screen size, lighting, camera
horizontal angle, room objects are potentially
important for distance collaboration.
32
Thanks!
33