Transcript James Watt - University of Connecticut

Avatars, Anonymity and Social
Connection: Implications for Health
Games and Health Communication
James Watt
Professor Emeritus, Communication Sciences
University of Connecticut
[email protected]
11-Feb-2010
1
ABSTRACT
In this talk I will review the benefits and drawbacks of providing anonymity of differing
degrees in situations in which there is the possibility of personally or socially negative
consequences as a result of honest communication.
I will first discuss the positive and negative impact of providing anonymity in on-line
communication, then summarize some research into the characteristics of avatars and
animated agents with differing types and degrees of realism on social connection
between communicators who use these avatars as personal representations.
The results of two studies assessing the effects of different avatar characteristics on
social connection between interactants and on self-disclosure of personal information
will be presented.
These studies provide some solutions to minimizing negative communication
outcomes produced by anonymity. Implications for the design of health
communication technologies will be discussed.
Anonymity and Avatars
11-Feb-2010
2
Social Connection
• In mediated communication, various terms
are used
– Presence
– Copresence
– Social copresence
• All contain the idea of having access to the
other interactant’s internal psychological
state, similar to actually being present face-toface while communicating with the individual.
Anonymity and Avatars
11-Feb-2010
3
Positive outcomes of on-line social
connection in health communication
• Information exchange
– “Why don’t you check the AA site for suggestions
on how family members can cope?”
• Group identity formation and support
– “Everyone in this discussion is a cancer patient or
survivor. We’ll be here for you.”
• Non-threatening social pressure for healthy
behaviors
– “I walked for 45 minutes today. How did you do?”
Anonymity and Avatars
11-Feb-2010
4
Social connection in mediated health
communication
Valuable Outcome
Achieved by
Information gain
Good site design
Good moderation
Good game design
Group support
Good moderation
Inclusion of social facilitators
Ubiquitous communication ease
Social pressure to persist
Continual connection
Establishment of expectations
No negative communication
Anonymity and Avatars
11-Feb-2010
5
Negative outcomes of on-line social
connection
• Ridicule
– “You mean you can’t stay with an exercise program
more than 3 weeks? C’mon!”
• Teasing and bullying
– “If I was as fat as you, I’d have my refrigerator welded
shut.”
• Forfeiture of Privacy
– “We’d like to hire you, but you posted a note that said
you have recurrent bouts of disabling depression that
cause you to miss work.”
Anonymity and Avatars
11-Feb-2010
6
Anonymity in on-line groups and
health games
• The solution to most of the negative outcomes is to
provide anonymity of participants or game players.
– Not a “normal” way of communicating.
– May reduce the effectiveness of the communication
process by removing important information about the
communicator.
• The central research questions
– Can the positive outcomes can be retained if
communicators are allowed anonymity to deal with the
negatives?
– Are there ways to design communication technology to
make anonymous communication more effective?
Anonymity and Avatars
11-Feb-2010
7
Anonymity in on-line communication
• A good summary:
– Christopherson, K. M. (2007). The positive and negative
implications of anonymity in Internet social interactions: “On
the Internet, nobody knows you’re a dog”. Computers in
Human Behavior, 23(6), 3038-3056.
• Communication effects are focused on social
anonymity:
– “…the perception of others and/or one’s self as unidentifiable
because of a lack of cues to use to attribute an identity to that
individual.” (Christopherson, 2007, p. 3040)
• But removing cues may damage the
communication effectiveness.
Anonymity and Avatars
11-Feb-2010
8
Positive communication effects of
anonymity - Privacy
• Anonymity provides privacy:
– Ability to exert control of information about self
– Shown to have positive effects on psychological
well-being, some resulting from …
• Catharsis of anti-normative thoughts or behaviors
without negative social consequences
• Autonomy to explore new behaviors or identities
without negative consequences
Anonymity and Avatars
11-Feb-2010
9
Positive communication effects of
anonymity – status equalization
• “Equalization hypothesis”
– Dubrovsky, V. J., Kiesler, B. N., & Sethna, B. N. (1991). The
equalization phenomenon: status effect in computer-mediated
and face-to-face decision-making groups. Human–Computer
Interaction, 2(2),119–146.
– freedom from visual cues or other identification
removes status hierarchy, gender, and other social
status effects.
– Has received only very limited support, leading to …
– The “Watt Conjecture”:
• “On the Internet everyone knows you’re a dog.”
• Corollary:
– But they don’t necessarily know which dog you are.
Anonymity and Avatars
11-Feb-2010
10
Positive communication effects of
anonymity – group identity
• The Social Identity Model of Deindividuation
Effects (SIDE)
– (Spears, R., & Lea, M. (1992). Social influence and the influence of
the ‘‘social’’ in computer-mediated communication. In M. Lea
(Ed.), Contexts of computer-mediated communication (pp. 30–
65). London:Harvester-Wheatsheaf.
– Reinterpretation of Zimbardo’s Deindividuation theory
in the context of computer-mediated communication.
– SIDE theory predicts that when all members of the
group are anonymous to each other, group salience
will increase and members will identify more strongly
with the group.
• Important for many health communication objectives.
Anonymity and Avatars
11-Feb-2010
11
Negative communication effects of
anonymity-flaming
• Flaming and critical comments
– Shown to be more prevalent in anonymous than
interactant identified communication settings.
– Reinig, B. A., & Mejias, R. J. (2004). The Effects of National Culture and
Anonymity on Flaming and Criticalness in GSS-Supported Discussions
Small Group Research, 35(6), 698-723.
– Alonzo, M., & Aiken, M. (2004). Flaming in electronic communication.
Decision Support Systems, 36, 205-213.
– This likely results from the dark side of maintaining
privacy:
• Social norms are less powerful without the possibility of sanctions
to the individual.
– Also may be a possible side-effect of deindividuation.
• Group is seen as undifferentiated whole and reified into a simple
entity that has weak norm enforcement.
Anonymity and Avatars
11-Feb-2010
12
Negative communication effects of
anonymity-reduction of social
connection
• There is some evidence that anonymous
communications lead to less social
“connectedness” among participants
– Kang, S., Watt, J. H., & Ala, S. K. (2008) Social copresence in
anonymous social interactions using a mobile video telephone.
Proceedings of CHI 2008. New York: ACM Press.
• The presumed cause of the reduction is removal
of important communication cues like nonverbal
facial expressions, body movements, etc.
Anonymity and Avatars
11-Feb-2010
13
Media richness and social presence
• Media richness historically referred to the amount of information
(incorrectly termed “bandwidth”) provided by the mediating
channel, e.g. text vs. video.
– Daft, R. L. & Lengel, R. H. (1984). Information Richness: A new approach to managerial
behavior and organization design. Research in Organizational Behavior, 6.
– Only mixed support for hypothesis that richer media produce more
social presence.
• E.g. Walther’s Hyperpersonal Communication theory
– Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal,
and hyperpersonal interaction. Communication Research, 23(1), 3–43.
– Found more perceived psychological closeness for text than video in some
communication circumstances.
– Occurred primarily when participants were expecting future interaction.
• Reinterpretation of media richness:
– Focuses on “cues” provided by communicators rather than
technological affordance provided by the channel.
– Rich cues improve social presence.
Anonymity and Avatars
11-Feb-2010
14
Anonymity and Avatars
• Full visual anonymity removes cues that might be
important to effective communication.
– Nonverbal visual channel cues are particularly
important to accurate exchanges.
• CMC solution to provide anonymity while still
retaining some visual cues is to provide an avatar.
• An avatar can be any symbol that represents the
communicator.
– E.g. text screen name; abstract non-human object;
cartoon; photograph; full motion video, etc.
Anonymity and Avatars
11-Feb-2010
15
Avatar Realism
Avatar Realism
Visual Realism
Fidelity
Anonymity and Avatars
11-Feb-2010
Anthropomorphism
Behavioral Realism
Kinetic
Conformity
Social
Appropriateness
16
Impact of avatar realism
• More realistic avatars generally thought to
improve communication outcomes.
– Garau, M., Slater, M., Vinayagamoorthy, V., Brogni, A., Steed,
A., & Sasse, M. A. (2003). The impact of avatar realism and
eye gaze control on perceived quality of communication in a
shared immersive virtual environment. In Proceedings of the
SIGCHI conference on Human factors in computing systems
(pp. 529-536). New York: ACM.Some exceptions
– With exceptions
• Hyperpersonal communication
• Also, the “uncanny valley”
Anonymity and Avatars
11-Feb-2010
17
The “Uncanny Valley”
• From human-robot interaction research
– Mori, M. (1970). Bukimi no tani [The uncanny valley]. Energy,7(4),
33–35.
• Human feelings of comfort with technology decrease when
robot realism is high (close to a human) but not perfect.
• Applies to avatars, too, but only if unusual features are
included (i.e. lower anthropomorphism) or if visual and
behavioral realism are inconsistent.
– See: Seyama, J. i., & Nagayama, R. S. (2007). The Uncanny Valley:
Effect of Realism on the Impression of Artificial Human Faces.
Presence, 16(4), 337–351
• May decrease the effectiveness of avatars in
improving anonymous communication outcomes.
Anonymity and Avatars
11-Feb-2010
18
Two Studies of Avatars and
Anonymous Communication
• Study 1
– Effects of different types and levels of avatar
realism and anonymity on social copresence.
• Study 2
– Effects of different levels of visual anonymity on
interactants’ self-disclosure of emotionally-loaded
information.
Anonymity and Avatars
11-Feb-2010
19
Study 1 Setting
• Conducted in the Social and Behavioral
Research Laboratory at Rensselaer Polytechnic
Institute.
• Research team director, James Watt
– Primary researcher: Sinhwa Kang (Ph.D.
candidate)
– Technical support: Sasi Kanth Ala (Graduate
student)
Anonymity and Avatars
11-Feb-2010
20
Pretest of processed video in providing
anonymity
See Kang, S. (2007). The impact of
digital iconic realism on anonymous
interactants' mobile phone
communication. In CHI '07 extended
abstracts on human factors in
computing systems. New York: ACM
Press.
Anonymity and Avatars
11-Feb-2010
21
Study 1 Research Design
Factorial Between-Subject Experiment:
2 (Behavioral Realism) x 5 (Visual Type)
x 3 (Gender Dyad, M-M, F-F, M-F)
plus comparison group (No avatar).
N=18 per cell, total N= 198, college
students from two upstate NY
institutions.
Dependent Variable
Measurements (Self-Report)
Copresence
Social Richness of Medium
Interactant Satisfaction
Anonymity and Avatars
11-Feb-2010
22
Experimental Groups
NonAnonymous
Avatar
Anonymous Avatars
High Visual Realism Avatars
Unmodified
Video
Overall
Processed
(edge-detector
filtered) Video
Low Visual Realism Avatars
Outer-Masked
Processed
Video
High
Anthropomorphism Drawing
Low
Anthropomorphism Drawing
NonAvatar
(Audioonly)
Behavioral
Realism
HIGH
Dynamic
Avatar
(Video or
Animation)
LOW
Static Avatar
(Photo or
Drawing)
Anonymity and Avatars
11-Feb-2010
N=18
(3 gender dyads x
3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender dyads x
3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender dyads
x 3 replications)
N=18
(3 gender
dyads x 3
replications)
23
Mobil device software mockup
Experimenter’s
control panel on
right does not
appear to
experimental
subjects
Mock-up screen
resolution is
320x240 pixels
Anonymity and Avatars
11-Feb-2010
24
Medium fidelity, anthropomorphic,
kinetically realistic avatar
Anonymity and Avatars
11-Feb-2010
25
Low fidelity, low anthropomorphic,
kinetically realistic avatar
Anonymity and Avatars
11-Feb-2010
26
Anonymized video (medium fidelity,
anthropomorphic, high behavioral realism)
Anonymity and Avatars
11-Feb-2010
27
Experimental Procedure
• Subjects non-systematically assigned to
gender pair, gender pair randomly assigned to
experimental condition.
• Given a role playing task (“See if this person
might be compatible as an apartment-mate”).
– Requires self-disclosure from both participants.
– Encourages exploratory, non-task communication.
• Conversation continues for 10-15 minutes.
• Post-task social copresence questionnaire.
Anonymity and Avatars
11-Feb-2010
28
Characteristics of Social Copresence
Measures
• Dependent Measures
– Copresence incorporated more emotional
communication items and fewer task-oriented items
than past research.
• Important for social media success.
– Social Richness of Medium focused on appropriate
use perceptions.
• Important for fostering adoption of medium.
– Interactant Satisfaction is a summative measure of
perceived success of the interactions.
• Important for continued use of medium.
Anonymity and Avatars
11-Feb-2010
29
Sample social copresence self-report
items
Copresence (17 items, Cronbach’s alpha = .93)
S/he was intensely involved in our interaction.
S/he seemed to find our interaction stimulating.
I wanted to make the conversation more intimate.
I tried to create a sense of closeness between us.
Social Richness of Medium (5 items, Cronbach’s alpha = .86)
How likely is it that you would choose to use a mobile telephone interaction for a meeting in which you
wanted to persuade other person of something?
To what extent was this like a face-to-face meeting?
Interactant Satisfaction (15 items, Cronbach’s alpha = .90)
S/he responds appropriately to positive and negative emotions.
S/he handles others’ feelings and emotions sensitively and effectively.
Anonymity and Avatars
11-Feb-2010
30
Some Results-Avatar Behavioral
Realism
• Effects of higher behavioral realism
– Higher evaluation of Social Richness of Medium.
• A higher behavioral realism medium is seen as more
suitable for social interaction.
– No effect on Copresence or Interactant
Satisfaction.
• Perceptions of psychological access to other
communicator or satisfaction with the interaction are
not affected by dynamic visual cues.
Anonymity and Avatars
11-Feb-2010
31
Some Results-Avatar Visual Realism
• Effects of higher fidelity and anthropomorphism
– Higher Copresence
– Higher evaluation of Social Richness of Medium
– No impact on Interactant Satisfaction
• Visual quality and human appearance of the avatar
affect the perception of psychological access and
assessment of the medium, but not the results of
the communication.
• No avatar (audio-only) produced lower Social
Richness of Medium ratings.
Anonymity and Avatars
11-Feb-2010
32
Some Results-Avatar Anonymity
• High realism avatars that provide anonymity
do not differ from non-anonymous video on
any measure of social copresence.
• Low realism avatars produce less Copresence
and perceptions of Social Richness of Medium.
Anonymity and Avatars
11-Feb-2010
33
Study 1 Conclusions
• High realism avatars can be used to provide
anonymity without deterioration of important
Social Copresence variables, when compared
to non-anonymous high realism avatars.
• Lower realism anonymous avatars will impair
effective social communication.
Anonymity and Avatars
11-Feb-2010
34
Study 1 Publications
• Kang, S., Watt, J. H., & Ala, S.K. (2008). Communicators’
perceptions of social presence as a function of avatar
realism in small display mobile communication device.
In Proceedings of Hawaii International Conference on
System Sciences (pp. 147-156). New York: IEEE Press.
• Kang, S., Watt, J. H., & Ala, S. K. (2008). Social
Copresence in Anonymous Social Interactions Using a
Mobile Video Telephone. In Proceedings of CHI 2008
(pp. 1535-1544). New York: ACM Press.
Anonymity and Avatars
11-Feb-2010
35
Study 2 Setting
• Conducted in the Institute for Creative
Technology, University of Southern California
• Research director, Jonathan Gratch
– Primary researcher: Sinhwa Kang (Postdoctoral
Researcher)
– Research team member: James Watt
Anonymity and Avatars
11-Feb-2010
36
Study 2 research design
• Three group experiment with IV being ordinal
level of anonymity (none, medium, high).
• Dependent variables
– Amount of self-disclosure in response to questions
about personal matters.
– Copresence index from Study 1.
• N=108, 54 male and 54 female, from general
population of Los Angles, mean age = 37.5.
Anonymity and Avatars
11-Feb-2010
37
Experimental Manipulation
• Three levels of avatar anonymity
Unprocessed video.
No anonymity,
high behavioral
and high visual
realism.
Anonymity and Avatars
11-Feb-2010
Processed Video.
Anonymity with
high behavioral,
medium visual
realism.
Rapport Agents (Gender
matched to subject’s)
High anonymity with
medium behavioral and
medium visual realism.
38
Rapport Agent architecture
Anonymity and Avatars
11-Feb-2010
39
Examples of Rapport Agent behavior
Anonymity and Avatars
11-Feb-2010
40
Experimental procedure
• Subjects randomly assigned to experimental
condition.
• Told that both participants would be represented
by the same type of avatar.
• Assigned task:
– Find out as much as possible about the other person
by asking them 10 questions.
– Experimenter always asks same 10 questions.
– Copresence questionnaire administered before
subject asks questions.
Anonymity and Avatars
11-Feb-2010
41
Conversation task questions
• Questions asked in increasing intimacy elicitation order:
–
–
–
–
–
–
–
–
“How old are you?”
“What is your hometown?”
“What are your favorite things to do in your free time?”
“What characteristics of yourself are you most proud of?”
“What are some of the things you hate about yourself?”
“What do you dislike about your physical appearance?”
“What has been the biggest disappointment in your life?”
“What have you done in your life that you feel most guilty
about?”
– “What are some of the things that really hurt your feelings?”
– “What is your most common sexual fantasy?”
Anonymity and Avatars
11-Feb-2010
42
Measurement of self-disclosure
• Transcript of responses to avatar questions prepared.
• Two coders rate each utterance on self-disclosure as:
– 0 = no self-disclosure
– 1 = peripheral
• peripheral layer is concerned with biographic data, e.g. “I am 30 years old”
– 2 = intermediate
• intermediate layer deals with attitudes, values, prejudices, opinions,
aspirations, dreams, and desires, e.g. “I like to go shopping”
– 3 = core
• core layer is comprised of highly personal aspects related to basic values,
beliefs, needs, fears, self-concept, emotions, feelings, and things people are
ashamed of, e.g. “I’m so disappointed about my divorce.”
• Good intercoder reliability: Krippendorff’s alpha = .84
• Mean rating of both coders over full interaction used as subject’s
self-disclosure score.
Anonymity and Avatars
11-Feb-2010
43
Basic Path Analysis Results
Effect of
Other
Variables
Self-Disclosure
of Intimate
Information
.27
Anonymity
Level
-.26
.16
Emotional
Connection
Effect of
Other
Variables
Anonymity and Avatars
11-Feb-2010
44
Net impact of increasing anonymity
• Effect of increasing anonymity one standard
unit
– Direct Effect = +.27 std. unit of self-disclosure
– Indirect Effect = -.26 * .16 = -.04 std. unit of selfdisclosure
– NET EFFECT = .27 - .04 = .23 std. unit of selfdisclosure.
Anonymity and Avatars
11-Feb-2010
45
Study 2 conclusions
• Although increasing anonymity decreases
copresence (emotional connection) which in
turn decreases self-disclosure, this effect is
much smaller than the direct increase in selfdisclosure that anonymity produces.
Anonymity and Avatars
11-Feb-2010
46
Study 2 Publication
• Kang, S., Gratch, J., & Watt, J., (2010) Impact of
Anonymity on Social Copresence and Self-Disclosure in
Avatar Mediated Communication Settings. Working
paper (Under Review).
Anonymity and Avatars
11-Feb-2010
47
Implications for design of
communication systems for health
communication
• Anonymity, on balance, will improve self-disclosure in
emotionally sensitive communications.
• Avatars with high visual and behavioral realism will
reduce negative effects of anonymity on social
copresence and in turn, improve self-disclosure.
• Decreases in cost of digital communication technology
makes this feasible for social media for health
communication including health games that include a
social interaction component.
Anonymity and Avatars
11-Feb-2010
48
Avatars, Anonymity and Social
Connection: Implications for Health
Games and Health Communication
James Watt
Professor Emeritus, Communication Sciences
University of Connecticut
[email protected]
11-Feb-2010
49