Social Interaction Design for Social Media: The Case of Groupthink

Transcription

}w
!"#$%&'()+,-./012345<yA|
M ASARYK U NIVERSITY
FACULTY OF I NFORMATICS
Social Interaction Design for Social
Media: The Case of Groupthink and
Aggression
A
DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
D OCTOR OF P HILOSOPHY
(I NFORMATICS )
Michail Tsikerdekis
Brno, October 2012
Advisor:
RNDr. Jiří Zlatuška, CSc.
Department of Computer Science
D ECLARATION
I hereby declare that this dissertation is my own original work and has not previously been submitted to any other institutions of higher education. I further declare that all sources cited or quoted are indicated and acknowledged by means
of a comprehensive list of references.
ii
A CKNOWLEDGEMENT
I would like to thank several people that made this dissertation possible.
First and foremost, my advisor Jiří Zlatuška for his help and support with
my work and studies throughout my doctoral journey. His direction, ideas, and
encouragement were essential for the development of this interdisciplinary research.
Further, I would like to thank all faculty, staff and students at Masaryk University and especially the Faculty of Informatics that have assisted me and being
influential throughout my PhD candidacy. I extend a sincere thank you especially
to Bernadette Nadya Jaworsky whose research insights and assistance as well
as teaching support have been influential throughout my journey, Eva Hladká
whose help and support has given me a greater understanding of teaching and
Ada Nazarejová whose assistance with the faculty’s procedures and culture has
proven invaluable.
I would like to thank every single one of the respondents in my research studies. Without all of these volunteers the data would have never been collected. I
would also like to thank my mother, father and brother as well as my friends.
Without your love and support throughout my life and especially my dissertation process I would not be where I am today. Lastly, I would like to thank my
other half Ani, you have always stood by my side through the highs and lows
and always believed in me even when I did not believe in myself. Thank you for
your never ending love and support.
iii
A BSTRACT
Social media today play an increasingly important role in computer science, the
information technologies industry and society at large, changing people’s everyday communication and interaction. The domain of social media encompasses
a variety of services, such as social networking services, collaborative projects,
microblogging services and even virtual social worlds and virtual game worlds.
There are long established principles, guidelines, and heuristics that apply to social media design and are part of the foundations of human-computer interaction (HCI). For example, in interaction design two set of goals guide the design
of systems, usability goals and user experience goals. However, current design
and development frameworks are still ill-equipped for the ever-changing online
world. Ironically, they fail to take into account the social dimension of social media software. Cracks in the social fabric of a community operating under social
media software may have devastating effects, not only to the evolution of the
community but also to the longevity of the social media service. As such, social
media cannot be developed in isolation, without taking into consideration the
social experiences of users. Psychological and sociological principles should become part of the design process of modern social media. My research contributes
to this endeavor by focusing on the design and engineering of social experiences
on social media services. In my dissertation, I propose that an additional layer
be added to the usability and user experience goals. The new layer includes social experience goals, which are further classified as desirable, undesirable and
neutral. I produced a new definition for social interaction design that incorporates social experience goals. Building upon previously developed frameworks
and models for interaction design, I demonstrated how social interaction design
applies to activities such as needfinding, developing alternative designs using
prototyping and modeling, developing interactive versions of design and evaluating designs. I presented the benefits of using such framework by focusing
on two showcase phenomena deeply rooted in social behaviors - aggression and
groupthink. The aim was to demonstrate that social media design and development could be driven by goals that aim to increase collaboration and decrease
conflict in a community. I analyzed the effects of different features found in social
media today in respect to aggression and groupthink, and found positive evidence to suggest that social interaction, behavior, attitudes and phenomena can
be affected by social media design. By examining two vastly diverse social experience goals using quantitative as well as qualitative research methods currently
used in HCI, I demonstrated the usefulness of social interaction design in various classifications of social media services such as collaborative projects, social
networking sites and even virtual game worlds. In short, I argue that social experience could be engineered through software using the framework I provide for
social interaction design.
iv
C ONTENTS
1
2
3
4
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1 Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.1 Human-Computer Interaction . . . . . . . . . . . . . .
1.1.2 Software Engineering . . . . . . . . . . . . . . . . . . .
1.1.3 Interaction Design, User-centered and Affective Design
1.1.4 Social Interaction Design . . . . . . . . . . . . . . . . .
1.1.5 Social media . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Examples of Social Media Issues . . . . . . . . . . . . . . . . .
1.3 Aim of This Study . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4 Significance . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Web 2.0 and Social Media History . . . . . . . . . . . . . . . . . . .
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2 Web 1.0, 2.0 and 3.0 . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Social Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.1 Blogs . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.2 Collaborative Projects . . . . . . . . . . . . . . . . . . .
2.3.3 Social Networking Sites . . . . . . . . . . . . . . . . . .
2.3.4 Content Communities . . . . . . . . . . . . . . . . . . .
2.3.5 Virtual Social Worlds . . . . . . . . . . . . . . . . . . . .
2.3.6 Virtual Game Worlds . . . . . . . . . . . . . . . . . . . .
2.3.7 Microblogging . . . . . . . . . . . . . . . . . . . . . . .
2.4 Additional Considerations . . . . . . . . . . . . . . . . . . . . .
2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Groupthink Theory and Application . . . . . . . . . . . . . . . . .
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Defining Groupthink . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Examples of Social Media Endangered by Groupthink . . . . .
3.4 The Beginnings of Groupthink and Research . . . . . . . . . .
3.4.1 The Groupthink Model . . . . . . . . . . . . . . . . . .
3.4.2 Historical Examples of Groupthink . . . . . . . . . . .
3.4.3 A Model of Controversy . . . . . . . . . . . . . . . . . .
3.4.4 Preventing Groupthink . . . . . . . . . . . . . . . . . .
3.5 Groupthink and Online Groups . . . . . . . . . . . . . . . . . .
3.5.1 Researching Online Groupthink . . . . . . . . . . . . .
3.6 Software Features Studied for Groupthink . . . . . . . . . . . .
Researching the Individual Features for Groupthink... . . . . . . .
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Pro/con Lists . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2.1 Method and Initial Results . . . . . . . . . . . . . . . .
4.2.2 Analysis for Argument Production . . . . . . . . . . . .
4.2.3 Bayesian Analysis for Readability . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
3
3
5
6
8
9
10
11
14
15
15
16
19
20
21
22
24
25
25
26
27
29
30
30
30
31
32
32
34
36
37
38
39
40
42
42
42
43
43
47
v
5
6
4.2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3 Alternatives Randomization in Voting Interfaces . . . . . . . . . . .
4.3.1 Literature on Response Order and Behavior . . . . . . . . .
4.3.2 Assessing the Appropriate Response Order . . . . . . . . . .
4.3.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4 Leaders in Social Media Affecting Opinions . . . . . . . . . . . . . .
4.4.1 On Leaders and Groupthink . . . . . . . . . . . . . . . . . .
4.4.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5 Anonymity States and their Effect on Non-Conformity and Groupthink . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5.1 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Online Aggression Theory and Application . . . . . . . . . . . . . . . .
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2 On Aggression in General . . . . . . . . . . . . . . . . . . . . . . . .
5.2.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.2 Types and Categories of Aggression . . . . . . . . . . . . . .
5.2.3 The Nature of Aggression . . . . . . . . . . . . . . . . . . . .
5.3 Research on the Causes for Human Aggression . . . . . . . . . . . .
5.3.1 Theories on Aggression . . . . . . . . . . . . . . . . . . . . .
5.3.2 General Aggression Model . . . . . . . . . . . . . . . . . . .
5.3.3 Additional Related Phenomena . . . . . . . . . . . . . . . . .
5.4 Research on Online Aggression . . . . . . . . . . . . . . . . . . . . .
5.4.1 Media Violence Research . . . . . . . . . . . . . . . . . . . .
5.4.2 Online Disinhibition Effect . . . . . . . . . . . . . . . . . . .
5.4.3 Benign & Toxic Online Disinhibition Effects . . . . . . . . . .
5.5 Preventing Online Aggression . . . . . . . . . . . . . . . . . . . . . .
5.5.1 Software Features Studies for Aggression . . . . . . . . . . .
Researching the Individual Features for Aggression... . . . . . . . . . .
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 Anonymity States and Aggression . . . . . . . . . . . . . . . . . . .
6.2.1 Background Theory . . . . . . . . . . . . . . . . . . . . . . .
6.2.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2.3 Findings and Results . . . . . . . . . . . . . . . . . . . . . . .
6.2.4 New Hypotheses for the Additional Analysis . . . . . . . . .
6.2.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Quality of Error Messages and Suggestion of Alternatives . . . . . .
6.3.1 Error Messages and Frustration . . . . . . . . . . . . . . . . .
6.3.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.4 Message Submission Errors and Recovery . . . . . . . . . . . . . . .
6.4.1 Messages, Communication and Aggression . . . . . . . . . .
6.4.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
50
51
52
54
54
55
56
57
60
61
62
62
68
68
68
68
69
70
70
71
72
73
74
75
76
79
79
81
82
82
82
83
83
84
84
88
89
89
90
92
95
96
96
98
vi
7
8
6.4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Social Interaction Design Framework . . . . . . . . . . . . . . . . .
7.1 A New Look on Interaction Design . . . . . . . . . . . . . . . . . . .
7.1.1 Defining Social Interaction Design within Interaction Design
7.2 Needfinding: Identifying Social Experience Goals . . . . . . . . . .
7.2.1 Example: Social Experience Goals in UrCity . . . . . . . . .
7.3 Prototyping, Conceptual Models and Physical Models . . . . . . . .
7.3.1 Example: Prototyping and Modeling for UrCity . . . . . . .
7.4 Evaluating Social Interaction Designs . . . . . . . . . . . . . . . . .
7.4.1 Example: Evaluation for UrCity’s Features of Interest . . . .
7.5 Additional Aspects for Consideration . . . . . . . . . . . . . . . . .
7.5.1 Social Cognition . . . . . . . . . . . . . . . . . . . . . . . . .
7.5.2 Distributed Cognition Framework . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8.1 Reflecting on the Dissertation . . . . . . . . . . . . . . . . . . . . . .
8.2 Looking into the Future . . . . . . . . . . . . . . . . . . . . . . . . .
99
101
103
103
104
105
108
108
110
110
111
113
113
113
115
115
116
vii
L IST OF F IGURES
2.1
2.2
2.3
2.4
2.5
2.6
4.1
4.2
4.3
4.4
4.5
4.6
4.7
Classifications for social media 20
Google’s official blog. 21
Wikipedia’s encyclopedia page on pseudonymity. 22
IBM’s fan page on social networking site Facebook. 23
Masaryk’s University promotional video on content community,
YouTube. 24
Twitter page for the Faculty of Social Studies, Masaryk
University. 27
Descriptive results for pro/con lists and textual collaborative
interfaces in terms of total argument production and unique
arguments. 44
The prior and posterior distributions of effect size δ, when comparing
total arguments between Debatepedia and Wikipedia. The top graph
illustrates the unrestricted t-test for H1 , the middle graph illustrates
the order-restricted test stating that Wikipedia produced more
arguments than Debatepedia (H2 ), and the bottom graph illustrates
the t-test for the opposite order-restriction (H3 ). The dots mark the
height of the prior and posterior distributions at δ = 0. 46
The prior and posterior distributions of effect size δ, when
comparing uniqueness of arguments between Debatepedia and
Wikipedia. The top graph illustrates the unrestricted t-test (H1 ), the
middle graph illustrates the order-restricted test stating that
Wikipedia produced more unique arguments than Debatepedia (H2 ),
and the bottom graph illustrates the t-test for the opposite
order-restriction (H3 ). The dots mark the height of the prior and
posterior distributions at δ = 0. 47
Descriptive results for pro/con lists and textual collaborative
interfaces in terms of readability measures FRE, Flesch-Kincaid and
SMOG. 48
The prior and posterior distributions of effect size δ, when
comparing readability between Debatepedia and Wikipedia. Each
sub-graph illustrates the unrestricted t-tests (H1 ), the order-restricted
test stating that Wikipedia scored higher in readability than
Debatepedia (H2 ), and the Bayesian t-test for the opposite
order-restriction (H3 ). The dots mark the height of the prior and
posterior distributions at δ = 0. 49
The first version of the voting process with the leader supporting the
first solution. 57
The second version of the voting process with the leader supporting
the third solution. 58
viii
4.8 Demographic information for the leadership survey. 59
4.9 Simple graph for the votes for each solution per treatment. 60
4.10 Research model for research on anonymity states and groupthink
with results. Note: * p < .05, ** p < .01, *** p < .001 63
4.11 Likelihood of not conforming for each anonymity state. 64
4.12 Likelihood of not conforming and perception of anonymity. 65
4.13 Line chart displaying the gradually ascend of the perception of
anonymity as progressing to higher anonymity states. 66
4.14 Likelihood of not conforming for all three scenarios based on the
self-reported level of importance. The second graph is significantly
being impacted by this dependency. 67
6.1
3D clustered graph depicting the responses that individuals gave
under each anonymity state and their reported strength of opinion
on a topic. Responses under the state of pseudonymity (here
depicted with a light brown color) show a substantial effect 85
6.2 Responses based on anonymity states for each one of the three
scenarios. Each left column represents complete anonymity, each
middle column represents using real names and each right column
represents pseudonymity. 86
6.3 Model for level of aggression in responses under pseudonymity
based on situational context and strength of opinion about a
topic. 89
6.4 The two versions of 404 errors that respondents encountered. 91
6.5 Questions included in the survey investigating the effectiveness of
error messages. 92
6.6 Graphs depicting demographic information for the sample that took
part in the AG1 survey 93
6.7 Results for the three aspects of the two versions of error messages
evaluated by respondents. 94
6.8 Survey questions for the message submission error and recovery
survey. 98
6.9 Bar chart depicting the ten-point scale for importance of a message
and time availability. 100
6.10 The error submission model based on the results. 102
7.1
Social Interaction Design Goals adapted from the Interaction Design
goals developed by (Rogers, Sharp, & Preece, 2011) 107
ix
L IST OF A BBREVIATIONS
HCI
Human-Computer Interaction
SE
Software Engineering
ACM SIGCHI Association of Computing Machinery Special Interest Group on
Human-Computer Interaction
GAM
General Aggression Model
NHST Null Hypothesis Statistical Testing
SNS
Social Networking Sites
IxD
Interaction Design
SIxD
Social Interaction Design
UCD
User-Centered Design
GDSS Group Decision Support Systems
IT
Information Technology
Ajax
Asynchronous JavaScript and XML
REST Representational State Transfer
CMC
Computer-Mediated Communication
x
C HAPTER 1
I NTRODUCTION
Social media are a driving force of today’s world. The sheer economic potential
has mesmerized the business world (Libert, 2010). The opportunity for a young
generation to actively contribute and share content has affected individuals, their
families and society as a whole that can now share information free from restrictions of the physical world (Palfrey & Gasser, 2010). Software designers and developers are pushing the abilities of their field to the limit and attempt to break
new grounds and deliver services in an extremely competitive Internet world
that resembles the movie industry. However, unlike Hollywood, this is not a traditional industry.
Sharing and interaction is the moving force for social media (Kaplan & Haenlein, 2010). The iterative process of a project’s life-cycle goes beyond revising the
final product; if there can ever be a final product in social media. This is a young,
rapidly growing and ever-changing field (Poynter, 2010). Software engineers and
users are interacting reciprocally and passionately demand new breakthroughs.
This information society is not just hungry for digesting and sharing information
but also for new technologies that will allow for new levels of interaction between
users. Today’s forecast shows that “the rate of change and development of new
features of social media will continue for the foreseeable future” (Poynter, 2010,
p. 256). This is a world where social media services can grow rapidly through
word of mouth alone (Boyd & Ellison, 2007).
The role of Human-Computer Interaction (HCI) in the development of new
software and technologies such as social media has been the center of discussion
among computer scientists (Canny, 2006; Cogburn, 2003; Wang, Carley, Zeng,
& Mao, 2007). However, evidently the field is struggling to keep up with the
technologies being developed. Social media form a diverse field of various environments varying from Social Networking Sites (SNS) to video games (Kaplan &
Haenlein, 2010, 2011). The sheer variability of software and interfaces along with
rapid growth seems to have led to a disruption of communication between HCI
scientists and developers. Examples such as the downfall of Friendster (Boyd &
Ellison, 2007), examples of social media privacy issues (Fogel & Nehmad, 2009;
Gross, Acquisti, & III, 2005; Nippert-Eng, 2010) and other unwanted incidents
(Atkinson, 2008; Hinduja & Patchin, 2010; Wallace, 2001) paint a really grim
future for social media services and their relation to HCI. Perhaps however, a
1
1. I NTRODUCTION
more evidently worrying picture for the future of HCI is painted in a Microsoft
Research Ltd. report (Harper, Rodden, Rogers, & Sellen., 2008). The report aims,
among other things, to identify the future challenges for HCI in the emerging
digital world for the 2010s. A statement is made for an outdated HCI field that
needs to move forward (Harper et al., 2008):
Technology is changing, people are changing, and the society is changing. All this is happening at a rapid and rather alarming rate. [...]
HCI needs to extend its methods and approaches so as to focus more
clearly on human values. This will require a more sensitive view about
the role, function and consequences of design, just as it will force HCI
to be more inventive. HCI will need to form new partnerships with
other disciplines, too, and for this to happen HCI practitioners will
need to be sympathetic to the tools and techniques of other trades. Finally, HCI will need to re-examine and reflect on its basic terms and
concepts. Outdated notions of the ‘user’, the ‘computer’ and ‘interaction’ are hardly sufficient to encompass all that HCI will need to
attend to. (p. 52)
Developing design perspectives that consider the implications of design on
user interactions and social experiences is particularly important for social media
development but also currently problematic. At first glance, issues and incidents
in social media demonstrate that some features implemented in the software are
not investigated thoroughly in terms of their impact on the users. Additionally,
the current HCI framework lacks the proper tools to deal with the challenges of
the social dimension in social media. Sociological and psychological theories are
sparsely implemented or considered in the work and research in HCI. In turn,
questions regarding the social aspects of social media software development lack
adequate answers. What is the effect of software on social interactions and behavior? Through what means can one develop software in order to proactively
engineer these interactions and behaviors? What social aspects should be considered relevant to social media software and how can someone identify them?
What approaches can be taken to design and develop software in relation to these
aspects of interest?
This dissertation attempts to make a contribution towards the resolution of
these issues. Theoretically ground theories are extracted and empirical research
is employed with a much needed interdisciplinary perspective in HCI. Social media should be designed and developed with goals for improving social interaction and experiences. I argue based on evidence that this is not only possible but
beneficial for social media. To achieve this, I present in this dissertation original
research demonstrating that changes in software design features currently being
used in social media, can affect social behavior and improve the social experiences of the users. I focus on two frequently studied sociological and psychological phenomena as a showcase: groupthink and aggression. I demonstrate how
designers and developers can make choices and aim to predict, manage or even
alter social behavior in social media. My dissertation research has found positive
evidence for several features affecting groupthink and aggression.
2
1. I NTRODUCTION
After a literature review on groupthink (chapter 3), I have conducted original research for software features that potentially had an effect on groupthink
(chapter 4). I have found compelling evidence that: (a) anonymity states empower users to contribute their alternative solutions to a discussion (Tsikerdekis,
in press-b), (b) pro/con lists produce more unique arguments as compared to
textual collaborative interfaces (Tsikerdekis, in press-c), (c) publicly identified
leaders do not affect decision-making voting processes, (d) item randomization
in voting processes has great potential for reducing groupthink as opposed to
standardization, and (e) dynamic voting status indicators do not affect voting
outcomes (Tsikerdekis, in press-a).
Similarly, after a literature review on aggression (chapter 5), I conducted original research for various software features that potentially affected aggression
(chapter 6). I found positive evidence that: (a) anonymity states and specifically
pseudonymity is a factor for aggression (Tsikerdekis, 2011, 2012), (b) message
submission errors could severely impact communication, be the cause of implicatures and potentially aggression, (c) quality of error messages in terms of helpfulness and comprehension can reduce frustration for users and in turn aggression.
Finally, I further used my findings to develop a framework through which
software designers can identify desirable as well as undesirable social interactions, behaviors, attitudes and phenomena of interest and established a methodology for identifying features which can promote or inhibit them (chapter 7). The
framework describes a new layer of Social Interaction Design (SIxD) based on
social experience goals which is applied to the activities currently found in Interaction Design (needfinding, prototyping, interactive versions, and evaluation)
(Rogers et al., 2011). I provide guidelines for each of these activities in relation to
social interaction design and provide examples of their application based on the
features investigated in this study for groupthink and aggression. The aim of this
framework is to provide a new definition and approach to social interaction design that distinguishes it from traditional interaction design perspectives that aim
to understand interactions and behaviors between a user and a system. Social interaction design aims to understand social interactions between users through a
system and find ways to engineer the system in order to affect these interactions.
In this chapter, I introduce and open a discussion for some of the definitions
used throughout this study. I proceed with presenting some examples of issues on
social media and consequences of their design mismanagement. I further present
the aims of this study and discuss the significance of this work on the current
domain of HCI and social media.
1.1 Definitions
1.1.1 Human-Computer Interaction
The world of computers prior to the first personal computer remained private for
scientists, information technologies professionals and dedicated hobbyists (Carroll, 2012). As software complexity and a new graphical interface came to be, the
need for an Software Engineering (SE) evolution became important. Usability is3
1. I NTRODUCTION
sues that arose with the appearance of the first computer users made it apparent
to scientists that there was a need for a new interdisciplinary project that could
produce software and hardware which aimed to be friendly to the end-user; a
user-friendly software.
According to Myers (1998), research for various aspects of usability was not
unlikely for researchers that wanted to establish evaluations for the first graphical
environments as well as hardware input objects such the mouse. Direct manipulation of objects within an interface with a pointing device was first demonstrated in a 1963 MIT PhD thesis (Sutherland, 1964). Another example that today
accompanies almost every computer in various forms is the mouse which was developed at Stanford Research Laboratory in 1965 (English, Engelbart, & Berman,
1967). In fact, most of the software and hardware usability tests were set as standards by these early studies.
A general definition of Human-Computer Interaction by the Curriculum Development Group of the Association of Computing Machinery Special Interest
Group on Human-Computer Interaction (ACM SIGCHI) describes it as ”a discipline concerned with the design, evaluation and implementation of interactive
computing systems for human use and with the study of major phenomena surrounding them.” (Hewett et al., 1992, p. 5). However, a unified yet simple way of
defining HCI is perhaps the description given by Carroll (1997) in an article.
Human-computer interaction (HCI) is the area of intersection between
psychology and the social sciences, on the one hand, and computer
science and technology, on the other. HCI researchers analyze and
design-specific user-interface technologies (e.g. three-dimensional pointing devices, interactive video). They study and improve the processes
of technology development (e.g. usability evaluation, design rationale). They develop and evaluate new applications of technology (e.g.
computer conferencing, software design environments). Through the
past two decades, HCI has progressively integrated its scientific concerns with the engineering goal of improving the usability of computer systems and applications, thus establishing a body of technical
knowledge and methodology. HCI continues to provide a challenging test domain for applying and developing psychology and social
science in the context of technology development and use. (p. 61)
This description encapsulates the basic goals of HCI and the epistemological
challenges that it faces although the article precedes major technological breakthroughs in the world of the Internet such as social media. For HCI to advance
forward, the ease of use and learning soon had to be replaced by more complex
measurements in order to account for qualities like fun, well-being, collective efficacy, aesthetic tension, enhanced creativity, support for human development,
and many others (Carroll, 2012). Today, HCI expanded to encompass visualization, information systems, collaborative systems, the system development process, and many areas of design. Additional fields include information technology, psychology, design, communication studies, cognitive science, information
science, science and technology studies, geographical sciences, management information systems, and industrial, manufacturing, and systems engineering. All
4
1. I NTRODUCTION
of these are used to achieve the fundamental goals of HCI “to develop or improve
the safety, utility, effectiveness, efficiency, and usability of systems that include
computers” (Diaper, 1989, p. 3). When designing HCI, the degree of activity that
involves a user with a computer should be analyzed extensively in all of the various different user activity levels such as physical (Chapanis, 1965), cognitive
(Norman, 1986, 2002), and affective (Picard, 2000).
Modern HCI seems to approach the vision of the “Man-Computer Symbiosis”
(p. 4) that was once described by Licklider (1960). HCI scientific work not only
was necessary for the development and launch of personal computers but helped
promote many Internet technologies which are in use today. As computers such
as IBM’s Watson (IBM, 2012) gain the ability to speak and process natural language, HCI’s unique role of establishing and studying interactions will ensure a
peaceful symbiosis with the new superstructure of networked people and computers.
1.1.2 Software Engineering
Much more challenging is trying to define Software Engineering. In an attempt to
try and dissolve the confusion that arose during the 1980s, Humphrey (1988) tried
to establish a basic definition. In the article he stated that “software engineering
refers to the disciplined application of engineering, scientific, and mathematical
principles and methods to the economical production of quality software” and
then proceeded by clarifying that “the term quality refers to the degree to which
a product meets its users’ needs, This may refer to functional content, error rates,
performance, extensibility, usability, or any other product characteristics which
are important to the users” (Humphrey, 1988, p. 82). Today, almost twenty years
since this definition was published, the Bureau of Labor Statistics from the United
States Department of Labor published their own similar definition for what is a
computer software engineer in their Occupational Outlook Handbook, 2010-2011
Edition (Bureau of Labor Statistics, 2011). The report states that “computer software engineers design and develop software. They apply the theories and principles of computer science and mathematical analysis to create, test, and evaluate
the software applications and systems that make computers work.”
These two definitions ostensibly state that software engineering is an interdisciplinary field which includes principles and goals in line with human-computer
interaction. However, even though in software design and development both approaches are expected to be implemented and work alongside, this is not always
the case. Bowles and Box (2011) describe the relationship as follows:
The fields of design and development are yin and yang, complementary yet opposed. They rely on each other, and thus either support or
damage the quality of each other’s work. Poor execution in development can ruin the experience you’ve worked so hard on. (p. 132)
In fact, there seems to be a fundamental difference in the perception of practitioners in the two fields. A review study concludes that human-computer interface specialists are user-centered and software engineers are system-centered
(J. Brown, 1997). This sets the stage a problematic alliance between two fields that
5
1. I NTRODUCTION
seem to be competing rather than cooperating. The same study made a further assertion that two factors seem to be the main constituents of the current state of
affairs between the two fields. First, the contributions of HCI specialists are never
included in software engineering textbooks and second, HCI methodologies do
not clarify the role of the SE process relative to the HCI process. In addition, both
disciplines seem to disagree in the use of formal and informal methods. Software
engineering relies more on formal methods while HCI values both formal and informal methods. Downton (1993) in his book tried to establish a common ground
for the two fields while trying to replace traditional software engineering intuitive methods for designing user interfaces with informal evaluation techniques
established by HCI. Attempts such as these are a step forward in a direction to
merge the two fields together in order to produce what was originally defined
by Humphrey (1988) as quality software. Based on examples from social media
software during the last decade it seems that HCI findings have yet to reach software engineers to a satisfactory degree. Additionally, HCI as a field seems to
still be retaining a main focus on hardware interaction. Today studies on HCI
show that “the new direction of research is to replace common regular methods
of interaction with intelligent, adaptive, multimodal, natural methods” (Karray,
Alemzadeh, Saleh, & Arab, 2008, p. 152). While these technologies will definitely
improve physical interaction between humans and computers, this direction still
ignores the social dimension which is a crucial component of social media.
1.1.3 Interaction Design, User-centered and Affective Design
An evolutionary step in the development of software that takes into account user
needs came with Interaction Design (IxD). This includes several other HCI fields
of which two are significantly important: User-Centered Design (UCD) and affective design.
A definition of user-centered design was given by Vredenburg, Mao, Smith,
and Carey (2002) in their survey of identifying the most widely used methods
and practices among UCD practitioners.
UCD is herein considered, in a broad sense, the practice of the following principles, the active involvement of users for a clear understanding of user and task requirements, iterative design and evaluation, and a multi-disciplinary approach. UCD methods are modular
or identifiable processes involved in UCD practice. You should NOT
think of UCD as merely usability testing or software engineering. (p.
472)
Active involvement of users is essential for the practice of user-centered design.
In a similar manner, the idea of affective design became popular as it emerged
from HCI (Norman, 1983). The field of affective computing was formed with its
main aim to deliver affective interfaces capable of eliciting certain emotional experiences for users (McCarthy & Wright, 2004; Reynolds & Picard, 2001). Based
on the ideas of affective computing, software products contain certain features
6
1. I NTRODUCTION
that can invoke physiological as well as psychological responses to users and can
enhance the emotional experience when using the product.
Both, user-centered design as well as affective design are incorporated in the
interaction design process. In fact, Rogers et al. (2011) in their book they demonstrate that interaction design is an extremely broad concept that combines many
fields that relate to the design of products. In an attempt to develop a working
and short description for interaction design they stated the following (Rogers et
al., 2011):
A central concern of interaction design is to develop interactive products that are usable. Generally meant to provide an easy to learn, effective to use and enjoyable user experience. (p. 2)
The most important question that interaction design addresses is: how does
one optimize the user interactions with a system so they match the user activities
that the system was designed for? To achieve this goal one needs to take “[...] into
account what people are good and bad at [...] considering what it is that might
help people with the way they currently do things [...] thinking through what
might provide quality user experiences [...] listening to what people want and
getting them involved in the design [...] using "tried and tested" user-based techniques during the design process” (Rogers et al., 2011, p. 8). Interaction design
utilizes knowledge from a variety of different scientific fields forming a new multidisciplinary field that investigates software design from multiple perspectives.
This is an extremely diverse field that tries to revise old conventions for software
development which generally avoid the identification of important human factors.
Interaction design has certain concretely identified goals. Rogers et al. (2011)
classify these as usability goals and user experience goals. Usability goals describe a set of goals derived from the field of usability engineering such as, effectiveness, efficiency, safety, utility, learnability, memorability. These goals are
located at center of interaction design. On the other hand, equally important are
the user experience goals which state that the system should be, fun, enjoyable,
satisfying, entertaining, helpful, motivating, aesthetically pleasing, supporting,
creative, rewarding, and emotionally fulfilling. In a figure describing the goals of
interaction design, Rogers et al. (2011) assigned this goal in the peripheral spectrum of interaction design.
It becomes reasonably obvious why considering both of the above goals is so
important for software development. By defining them, the software engineering team has a clearly defined vision for their software. This becomes a baseline
during the development of software and should always be evaluated in regular
intervals in order to establish when a project deviates from these early defined
goals. The downside for such a strategy is that it requires scientific expertise. Engineering based on intuition cannot work with such a model. All aspects of the
software design need to be based on solid ground theory and empirical studies.
As a side-effect of having the need for expertise, this approach is more costly for
projects however, the positive outcome can potentially outweigh the cost. The
software that is produced based on interaction design is significantly more ef7
1. I NTRODUCTION
ficient and serves user needs better. Additionally, it provides an improved user
experience which is good for the general marketing of the product.
1.1.4 Social Interaction Design
Interaction design is definitely a step forward towards HCI’s future described by
Harper et al. (2008), however its adoption is still limited and lacks an important
aspect which should be added in the initial model; the social interaction. When
it comes to developing social media, social interactions and behaviors become an
unavoidable aspect of the system itself which designers should also consider.
Social interaction design is considered by many to be a step forward in the
future of HCI (Chan, 2006; Huang & Deng, 2008; Jakobsson, 2006). However,
given the young nature of the field, definitions vary. Chan (2006) states the following about the nature of a social interaction designer:
The social interaction designer’s job is to anticipate the kinds of social
interaction a system will require if it is to sustain itself once launched.
Interaction designers think in terms of user behaviors. Social interaction design adds social and interpersonal attributes to the existing
technical and device interactions. Users must have social competencies with the community’s theme (what it’s about, e.g. dating, jobs,
etc.) and have a grasp of web navigation, links, use of form pages,
search, and so on. (p. 7)
A different approach was taken by Jakobsson (2006) which not only tried to
detach social interaction design from the traditional Computer-Mediated Communication
(CMC) research approaches but also suggested and demonstrated that the practice could in fact include ethnographic research. The following text makes this
clear (Jakobsson, 2006):
As the term computer mediated communication implies, much of the
research in this field has been based on [...] the context of the phenomenon [...] taken to be the physical work context to which the communication mediating technology was introduced. In the early nineties
the computer mediated communication research spawned a new branch
as the arena perspective started [...] to acknowledge both that the participants perceive that the technology allows them to share a common non-physical space and that this form of social interaction offers
a value in itself that makes participants spend time and effort in virtual places beyond their specific goal or purpose. (p. 62)
In other words, the environment that has been created by software engineering should be treated as a separate and unique entity with its own physical rules
since it is perceived as such by users. Additionally, it also pointed out the fact
that many individuals made use of these environments in ways that were not
originally intended by designers.
Furthermore, the approach taken by Jakobsson (2006) demonstrates several
implications that come from the ontological and epistemological approach of a
8
1. I NTRODUCTION
scientist. Most software development today and research in computer science and
informatics takes a positivist approach that accepts that reality is a given and
one can observe it objectively. However, when the social dimension is taken into
account, informal methods such as qualitative research methods and especially
ethnography, asserts that reality is perceived, can never be objectively stated and
has to be determined from within the social experiences of the researcher. This is
a good demonstration that as HCI tries to borrow from other scientific fields, it
also borrows many aspects that are parts of long debates that exist within these
fields. By adopting, psychological and sociological perspectives in the software
development process one has to side with epistemological questions that need
not to be answered with an absolute computer science perspective. While this
paper tries to maintain a positivist approach in its methods and interpretation of
results, it also accepts that such practice may be impossible at times. Therefore,
it does not advocate that one practice is better than the other, especially since
HCI is by definition an interdisciplinary field. Moreover, this way of thinking
can be restrictive since for example, Bryman (2006) has shown that quantitative
and qualitative methods can be combined in order to provide triangulation or
completeness for scientific results; a tactic that is used in this dissertation as well.
1.1.5 Social media
Social media aim to advance not only social communication but to go beyond
that and evolve social interactions. The most recent and perhaps most successful definition to describe social media was given by Kaplan and Haenlein (2010).
They defined social media as “a group of Internet-based applications that build
on the ideological and technological foundations of Web 2.0, and that allow the
creation and exchange of user-generated content” (Kaplan & Haenlein, 2010, p.
61). With a revolutionary way the “development of social media has substantially changed the way organizations, communities, and individuals communicate” (Leach, Komo, & Ngugi, 2012, p. 6).
Today, social media include blogs, microblogging, social networking sites,
video sharing sites, massive multiplayer games, virtual worlds, and so on (Kaplan & Haenlein, 2010, 2011). In fact, this is just a limited description of all classifications of social media services that made an appearance in the past decade and
still, new pioneering types appear regularly in an attempt to encapsulate another
piece of social interaction.
This potential for expanding the domain of social interactions to the digital
world is a driving force for companies that also see the potential for better marketing and in turn profits (Libert, 2010; Poynter, 2010). By analyzing identity,
conversations, sharing, presence, relationships, reputation, and groups, firms can
develop a congruent social media strategy based on the appropriate balance of
building blocks for their community (Kietzmann, Hermkens, McCarthy, & Silvestre, 2011). Whether companies developing social media or using them in order to communicate better with their clients, the success is evident in the reports
of some of the most popular social media services. Contrary to the conventional
marketing wisdom were one dissatisfied customer tells ten people, in the age
of social media, people have the tools to tell to 10 million customers virtually
9
1. I NTRODUCTION
overnight (Gillin, 2007).
At present time, Facebook which is one of the biggest social networking sites,
reports 800 million active users with more than 900 million objects that people
interact with such as groups, fan pages, events and community pages (Facebook,
2012). Another social media giant, YouTube, currently reports 48 hours of video
being uploaded every minute with 3 billion videos being viewed by its users
every day (Youtube, 2012b). These numbers paint a clear picture that it is not just
the companies favoring social media but users also seem to have made them an
integral part of their everyday lives. Chapter 2 provides a more detailed look on
social media.
Demand for new social media services that could expand to all kinds of social
interaction initiated numerous of companies providing a variety of services that
keep diversifying. The rapid growth of the industry has made it popular and
profitable. However, issues related with HCI have always overshadowed some
of these advancements.
1.2 Examples of Social Media Issues
Evidence of inadequate communication that exists between human-computer interaction and software engineering can be found in recent examples from those
involved in the social media industry. Additionally, I argue that most of these issues also demonstrate the lack of HCI to develop social media by taking social
interactions into account.
Perhaps one of the most prevalent examples of software mismanagement was
the case of the social networking website, Friendster. Friendster preceded most
major social networking services that are known today and in many ways set the
standards for many that followed. Boyd (2004) was one of the first to conduct
research on the highly popular website at the time. In one of her papers she suggested “how the HCI community should consider the co-evolution of the social
community and the underlying technology” (Boyd, 2004, p. 1279). Indeed this
suggestion could not have been any more prophetic about what followed.
The website started off as a dating website but encouraged users to join even
if they were not looking for a date (Boyd, 2003). It was built with the assumption that friends of friends are more likely to be good dates. The founders have
arbitrarily decided that the distance within a network in which one can find a
probable soul-mate is four degrees. As a result, any individuals beyond the four
degrees could not be reached by users. The choice was obviously restrictive for
the community if one takes into account the theory of six degrees of separation
and the findings from the small world experiment (Travers & Milgram, 1969).
There are still skeptics among the scientific community unwilling to sign on to
the idea of six degrees or what boundaries of a network are defined as small
(Schnettler, 2009). However, fifty years of scientific studies should have been a
cause for concern for Friendster’s designers and developers. At the time that was
not the case.
The website launched in the fall of 2002 its beta program and by mid-August
2003 the site had 1.5 million registered accounts and was still growing exponen10
1. I NTRODUCTION
tially (Boyd, 2003). The rapid growth created technical and social difficulties for
the software engineering team. The servers could not sustain the increasing traffic
of the website and the main page faltered regularly which in turn frustrated users.
Boyd (2003) also described some of the social issues that were created. Friendster
upset the cultural balance, contributed to the collapse of social contexts and created a phenomenon of users creating accounts based on their interests. These
users were called fakesters and what they were trying to do is connect beyond
the restrictive four degrees of separation that the software engineers set as a rule
for the community. Many fakesters had two thousand friends in their accounts
which resulted in tremendous processing loads for servers that were not designed
to handle such traffic. Eventually, when software engineers were faced with a decision to resolve the problem, they decided to delete the fakesters’ accounts. This
resulted in a rejection of the website in the U.S. by early adopters which were
faced with a combination of technical difficulties, social collisions and a rapture
of trust between users and the site (Ellison, Steinfield, & Lampe, 2007). Coincidentally, many of these issues violated also the hierarchy of needs for online users
as they were adapted by Kim (2000) from Maslow’s hierarchy of needs (Maslow,
1943) .
This distrust between the people and social media services which signals a
probable dysfunctional relationship between HCI and software engineering can
be found in many social media services today. As suggested on an article, “Social
networking websites should inform potential users that risk taking and privacy
concerns are potentially relevant and important concerns before individuals signup” (Fogel & Nehmad, 2009, p. 153). The cause for concern is justified considering that studies have demonstrated the potential for harm while on social media.
Social media services can become tools for stalking, re-identification, building a
digital dossier and have a fragile privacy protection nature (Gross et al., 2005).
Risks have also been found for cyber-bullying which should be a serious cause
for concern. Unlike face-to-face bullying, cyber-bullying has not been the subject
of research thoroughly and there are indications that it might be even more severe
than face-to-face bullying and have more negative long-term effects (M. A. Campbell, 2005). Today, unfortunately there is evidence that these effects can not only
be long-term but they can also have a negative impact on individuals with verbal and psychological bullying leading to physical abuse or self-abuse. Such was
the case when a female teenager that fell victim to a mother pretending to be a
16-year-old boy committed suicide (Atkinson, 2008).
These are not random events or accidents but testimonies of an industry that
is lacking the proper scientific understanding of the effects of software on their
clients. The success and excitement for the immense potential of these new technologies should not overshadow the need for a grounded understanding and a
scientific-guided software development of social media services.
1.3 Aim of This Study
Creating a viable and sustainable software design framework for developing and
managing social media projects is challenging. If anything, the examples that
11
1. I NTRODUCTION
were described in section 1.2 show some of the problems that designers and engineers face. To add to the confusion, the current knowledge pool of the field
lacks proper understanding of the effects of social media software on human factors and the social dimension. Researchers lack the ability to make casual claims
because of the lack of experimental studies and because the long-term effects of
these social tools are still unknown (Ellison et al., 2007). This becomes increasingly complex if designers have to account additionally for the social domain.
Social interaction design could be a way forward but lacks a framework for implementation. In addition, there are no guidelines for identifying what social interactions, behaviors, attitudes and phenomena are relevant for the development
and sustainability of social media services.
There are several phenomena that occur in online communities on an individual level as well as on a group level. Wallace (2001) in her book described aspects
varying from impression formation, group dynamics, online conflicts to even occurrences of online altruism. Other topics of particular interest deal with online
disinhibition (Gackenbach, 2007; Suler, 2004a, 2004c), social media and youth
development (Palfrey & Gasser, 2010; Subrahmanyam & Smahel, 2011), computer mediated communication (Thurlow, Lengel, & Tomic, 2004) and privacy
(Camenisch, Bner, & Rannenberg, 2011; Nippert-Eng, 2010) which has been the
focus of many scientists in the 2000s. The amount of social behaviors, phenomena
and aspects that exist and should be up for consideration when designing social
media is immense.
In the light of the above, I revised my study aims in order to conclude the
research within the time-frame of a doctorate study. I aimed to study cases that
would be of particular interest for software design related to social media, whether
for creating a virtual game world, a social networking site or even a brainstorming wiki for a corporation. As a general guideline, a specific theme was set in order to demonstrate that designers can increase collaboration and decrease conflict
when developing social media. Based on this theme, two showcase phenomena
were selected - groupthink and aggression.
Janis (1972) originally defined groupthink as “A mode of thinking that people engage in when they are deeply involved in a cohesive ingroup, when the
members’ strivings for unanimity override their motivation to realistically appraise alternative courses of action” (p. 8). It has been attributed as the cause for
Watergate, the Challenger disaster (Brooke & Tyler, 2011), and the 2008 financial crisis (Shiller, 2008). There is an extensive amount of studies on preventing
groupthink in groups (Hart, 1998; Jessup, Connolly, & Galegher, 1990; Kroon,
Hart, & Van Kreveld, 1991; Miranda, 1994; W.-w. Park, 1990). However, these
results in their current form still cannot sufficiently be implemented in software
design and development processes.
Based on the above, one of the questions that this study has raised is the viability of altering an interface in order to ensure that a decreased chance for groupthink will be observed for an online group. Preventing groupthink is a challenging task. Even in the real world when preventive measures are taken, there is
no guarantee for preventing groupthink. As an example, while the homogeneity
of a group is believed to be a contributing factor for groupthink (Janis, 1982),
12
1. I NTRODUCTION
simply adding individuals with a diverse background to a group would not benefit an organization already plagued by groupthink (Brooke & Tyler, 2011). Due
to this acknowledgment, the decision was made to conduct a study with broad
goals for contributing factors to groupthink than isolating the study by conducting an in-depth analysis for just one of the contributing factors. This, will also be
of great help to software engineers that may be dealing with social media that do
not include some of the contributing factors. Developers of a research wiki that
has many scientists from all over the world coming from different scientific disciplines may not have issues with the homogeneity of a group but should still consider other contributing factors such as poor information search, not examining
all alternatives (Kamau & Harorimana, 2008), opinions of leaders affecting the
group (Janis, 1982) and others. There is an unprecedented opportunity to study
how these factors may be translated in terms of their implementation on social
media design. Detailed literature review and research results on groupthink are
presented on chapters 3 and 4.
The second phenomenon under investigation was aggression. Aggression in
general “is the delivery of an aversive stimulus from one person to another, with
intent to harm and with an expectation of causing such harm when the other
person is motivated to escape or avoid the stimulus” (Geen, 2001, p. 3). This
definition also applies to online aggression. Since the intent to do harm is not
under the control of software engineers, other avenues that could contribute to
the prevention of such behavior were studied. Aggression can be malicious but
is also attributed to situational and environmental factors (Berkowitz & LePage,
1967; Castle & Hensley, 2002; Dollard, 1989). As such, changes in environmental factors can potentially impact aggression. Software seems to be a significant
contributor to online aggression (Wallace, 2001). This study maintained a focus
on decreasing frustration and in turn aggression (Dollard, 1989), that could be
attributed to technical or bad design issues in social media. Detailed literature
review and research results on aggression are presented on chapters 5 and 6.
Based on the aforementioned considerations, I itemize the main aims of this
study below.
•
To investigate if software can affect social behaviors in social media by conducting novel research on two showcase online behavioral phenomena groupthink and aggression.
•
To provide a new framework for social interaction design built upon current
interaction design practices that aims to engineer social behavior by altering
software design.
•
To provide recommendations and a methodology for software designers
based on the empirical evidence for the phenomena studied.
•
To advocate the benefits of social interaction design in the social media development process.
•
To promote scientific research and interest in the effects of software on social behavior from HCI and advocate a better way for communicating the
results from these studies to software engineers.
13
1. I NTRODUCTION
1.4 Significance
Research in the above phenomena as they occur online, aimed to identify if indeed software designers have more control over their users in social media. This
knowledge could also help develop a new channel of communication between
HCI scientists and software engineers in order to attempt to produce better quality software. As such social interaction design becomes a crucial framework for
social media design and development.
Social media are taking over all aspects of Internet use. Even websites that
used to be purely based on broadcasting information are shifting towards producing user-generated content and allowing users to share information. Additional benefits are also found when social media are implemented in the workplace (Ehrlich & Shami, 2010; Zhao & Rosson, 2009). The future of user experience online seems to be heading into a world where software and social interaction are intertwined. For example, a collaborative tool should be considered great
when it provides efficiency for project teams but also should guarantee measures
for successful decision-making. The boom in new technologies has caused the
development of software to lack HCI’s insights. In addition, HCI seems to be underdeveloped to handle the social dimension that is a centerpiece of social media.
This study aims to make a contribution and attempts to close the gap between
HCI and social media design and development. The goal was to not only establish
how software affects social behavior online but establish a framework through
which this information can be used by software engineers in order to produce
quality software that will allow for more efficient social interactions. Contrary
to research consisting of findings that have an observatory character, I wanted to
show that not only proactive prediction of online social interaction is possible but
also that designers can affect or even alter it altogether. This framework for social
interaction design will assist designers and developers in delivering social media
services for specific social experience goals. This information establishes a more
scientifically grounded approach to the architectural design and development of
software. In turn, a more sound understanding of the effects of the software on
the users will result in the development of more effective software. Finally the
contribution of this study also helps improve collaboration in online communities and groups by providing empirical evidence for inhibiting groupthink and
aggression when adapting software features and as such demonstrating methods
for future investigations of similar features.
14
C HAPTER 2
W EB 2.0 AND S OCIAL M EDIA H ISTORY
2.1 Introduction
Before presenting the study’s research on how social media design can be adapted
in order to affect groupthink and aggression, important definitions, history and
arguments for the philosophy of social media development need to be presented.
There is a variety of terms that relate to social media and given the young nature
of the field, establishing definitions and understanding the context is essential. In
addition, knowing the past can be important for predicting the future, or as President Harry S. Truman put it “You can’t know where you’re going until you know
where you’ve been” (Bateson & Klopfer, 1989). The essence of this sentence can
be traced all over history with names such as Adlai E. Stevenson, George Santayana and others. At its core this idea demonstrates a simple fact about human
history. Times change but humanity does not. The basic needs that define us as
a species remain with us as we travel through space and time. These qualities
made possible for Hiltz and Turoff (1978) to make observations in their book The
Network Nation which as stated by Howard (2010) were prophetic and show the
advent of YouTube and Hulu. Howard (2010) points to the fact that “they based
their predictions on a sociological understanding of the history of communication technologies and on deep-seated understanding about fundamental social
needs rather than technological possibilities” (p. 207). These are the same social
qualities that underlie in all aspects of human interaction today whether online
or offline.
The reality of social processes driving technologies forward is not always obvious. In the early history of technology advances and breakthroughs happened
each day. Scientists became fixated on the idea that people will adapt to technology. Bush (1945), which today is considered the inventor of hypertext (Wang et
al., 2007) along with other technologies that he predicted (Millard & Ross, 2006),
wrote in “As we may think”, one of his most famous essays (Bush, 1945):
Our present languages are not especially adapted to this sort of mechanization, it is true. It is strange that the inventors of universal languages have not seized upon the idea of producing one which better fitted the technique for transmitting and recording speech. Mechanization may yet force the issue, especially in the scientific field;
15
2. W EB 2.0 AND S OCIAL M EDIA H ISTORY
whereupon scientific jargon would become still less intelligible to the
layman. (p. 105)
The statement is extremely fascinating and the idea of a universal language
enticing. However, almost seventy years after this statement we have yet to see
any universal languages that are commercially sustainable and would best fit
computer technologies, especially for speech recognition. Instead, today’s speech
recognition is trying to adapt to different languages and a wide range of variations such as speaker, gender, speaking rate, vocal effort, regional accent, speaking style, non-stationarity (Benzeghiba et al., 2007). On the other hand, languages
can be adopted by a large proportion of a population. The adoption of the English
language seems to have been successful in non-English speaking countries in the
European Union and it is probably still on the rise (Labrie & Quell, 1997). As
stated by Labrie and Quell (1997) this increases the potential for communication
among Europeans which clearly identifies one sociological need behind this process. In fact, taking into account this sociological need for communication, one
can be fairly certain that this trend may increase in the future.
One sociological need that Bush (1945) saw was the need for a better communication between research scientists and for a better way of organizing the
information overload that existed at the time. His brainchild that will satisfy this
human-centric need was called a “memex.” He wrote the following (Bush, 1945):
Consider a future device for individual use, which is a sort of mechanized private file and library. It needs a name, and to coin one at
random, ”memex” will do. A memex is a device in which an individual stores all his books, records, and communications, and which
is mechanized so that it may be consulted with exceeding speed and
flexibility. It is an enlarged intimate supplement to his memory. (p.
107)
Today, we take the ability of doing such tasks for granted but this should
serve as a good reminder that such options were not always available. Technological needs are a direct implication of sociological needs and as long as this
chain is followed one can visualize near future. Coincidentally, in Information
Technology (IT) project management, the idea of a “vision” is the one that gives
birth to the mission statement for a project which also defines the general direction for the future development of a project (Webber & Webber, 2007).
This cognitive ability to visualize future information based on past information becomes particularly useful in software design given the limits for empirical
studies that could account for every situation possible. In fact, one may argue
that the best tool for software designers is this past knowledge that provides
them with evidence based predictions. Using past information they can potentially make sure that mental models between designers and users match.
2.2 Web 1.0, 2.0 and 3.0
Much has been said about the precursor of today’s web and arbitrary lines have
been drawn in order to define the transitional stages. Of course, identifying ver16
sions for such a broad concept as the Web is not similar as establishing versions
for traditionally produced software. It is fairly easy to version software based on
the major and minor changes in the product code and features (Ramesh & Bhattiprolu, 2006). On the other hand, the Web as a whole does not describe a single
entity and therefore there was no single event where people decided to shift from
version 1.0 to 2.0. With the variety of services that existed, the most likely thing
to happen was a gradual adoption of new ideas and concepts. Adoption of new
technologies seems to be influenced on people’s beliefs prior to adoption but even
after adopting the new technology (Karahanna, Straub, & Chervany, 1999). In
addition, age seems to play a significant role with the younger population being
influenced just by attitude towards technology as opposed the older populations
which is affected by subjective norms and perceived behavioral control (Morris
& Venkatesh, 2000). The young Internet during the period of 1995-2000 saw an
under-representation for the older age groups (Katz, Rice, & Aspden, 2001) and
as such adoption of new technologies was more likely to happen; and it did happen!
We can today identify the transitional period when at the time new ideas and
technologies were adopted by software engineers. Perhaps, a detailed description
of the Web 1.0 is made by Bernal (2009) which contrasted Web 2.0 to Web 1.0 and
hence defined the transition. He states that Bernal (2009):
Whereas the focus of Web 1.0 was on delivering products, Web 2.0 had
created a paradigm shift to delivering services that can be used and
combined with other services in new ways. Another key aspect is the
growth of interactivity with end users in new ways, enabling users to
drive what is important or of the most value. (p. 2)
The Web 1.0 had a sharp distinction between a user and a contributor or webmaster. Webmasters and content contributors delivered the information on the
servers from which users could finally retrieve. However, users could not contribute content on the servers. This may seem primitive now but at the time people were particularly excited about the potential of such technologies. McEnery
(1995) wrote an article about the potential uses of the early Internet and the one
of the first browsers that were able to display images along with text. Even at the
early static version of the Web 1.0 one can still identify the sociological needs for
interactivity and sharing.
This need was identified correctly by DiNucci (1999) in her article Fragmented
Future. She wrote (DiNucci, 1999):
The Web we know now, which loads into a browser window in essentially static screenfuls, is only an embryo of the Web to come. The first
glimmerings of Web 2.0 are beginning to appear, and we are just starting to see how that embryo might develop.[...]The Web will be understood not as screenfuls of text and graphics but as a transport mechanism, the ether through which interactivity happens. It will [...] appear
on your computer screen, [...] on your TV set [...] your car dashboard
[...] your cell phone [...] hand-held game machines [...] maybe even
your microwave oven. (p. 32)
17
Since the time in which this article was published, many of the features came
to be. Even more importantly, even though her statement described technologies
the underlying needs that led to these technological needs were still sociological
needs. These are the needs that evolved Web 1.0 to 2.0.
Moving forward a few years and Web 2.0 became a fascination of researchers
from various scientific disciplines. Many have attempted to define it. Bernal (2009)
gave a description of what Web 2.0 encompasses:
[...]in the Web 2.0 model, users actively participate and contribute to
a website. This bidirectional approach enables users to interact with
the site and each other in ways that provide for and foster a collective
community. Users can create, edit, rate, and tag content at will, which
provides other users with new information and guides the relevance
of what is important to the overall community. (p. 3)
This bidirectional interaction was only a matter of time with new technologies that allowed for information to be exchanged in the background between
browsers and servers. Bernal (2009) identified two core technologies that allowed
for such interactive experience for the end user, Asynchronous JavaScript and
XML (Ajax) and Representational State Transfer (REST). Ajax is a group of techniques that are used to create asynchronous web applications while REST is a
style for software architecture aimed for web applications. Additional concepts
are everyday being improved in order to foster this experience. Programming
languages such as PHP, Ruby, Perl, and Python are constantly improving the capabilities of server software while browsers are modernized to handle Adobe
Flash, JavaScript and other client frameworks. Database and database clusters are
also improving the way they handle data and constantly developing load balancing capabilities. These provide engineers with numerous combinations of server
and client hardware and software in order to design and develop modern Web
2.0 applications.
The idea of a Web as a platform became extremely popular by O’Reilly (2005)
which saw potential for business models. The idea also of combining Web 2.0
with service oriented architecture became popular. Schroth and Janner (2007)
specified that although the two have similarities as well as differences, the importance of unifying the two philosophies will help drive the development for
enterprises. Shuen (2008) in his book Web 2.0: A Strategy Guide Business thinking and strategies behind successful Web 2.0 implementations demonstrates the
strategies necessary to help customers build the website and create a user-driven
business. However, Hoegg, Martignoni, Meckel, and Stanoevska (2006) argued
that it is still debatable if Web 2.0 could ensure commercial success for all web
services.
No matter its commercial aspects, many enterprises are enjoying the benefits of Web 2.0 technologies with the majority of top executives favoring such
strategies (Murugesan, 2007). This is not without good reason since Web 2.0 has
significant differences than Web 1.0. Murugesan (2007) argues that Web 2.0 (pp.
34-35):
* facilitates flexible Web design, creative reuse, and updates;
18
* provides a rich, responsive user interface;
* facilitates collaborative content creation and modification;
* enables the creation of new applications by reusing and combining different applications on the Web or by combining data and
information from different sources;
* establishes social networks of people with common interests; and
* supports collaboration and helps gather collective intelligence.
Having all of the above in mind, it becomes obvious that Web 2.0 has significant advantages for communities which were not possible before. Or is this the
case? As discussed previously as long as there are sociological needs these need
to be satisfied by whatever technology exists at the time until new technologies
will help develop better versions of interaction.
Howard (2010) states that the excitement about Web 2.0 should not take away
the fact that people collaborating online and forming communities is not something new. He also expands on the idea that this could happen also on software
that is not Web based such as video games. However, one could still identify a
relatively new trend of providing more social tools in online games. This is not
surprising if one considers that gamers value personal interaction with the software as well as social interaction with other people (Choi & Kim, 2004). Web
2.0 represents many new ideas that are not confined in just Web applications but
they become part of a novel software engineering process that focuses on Service
Oriented Architecture principles and is trying to provide services across multiple domains whether web services, desktop services or gaming services. At the
rate that many services change, there is also an advocacy for avoiding traditional
Waterfall lifecycle methods and apply Agile Software Development principles
(Patterson & Fox, 2012). Approaches for a better development of software that
can accommodate if not welcome change is highly appreciated by Web 2.0 services.
Having that in mind, it becomes clear that the sociological need for social
interaction was not reinvented but the tools did in order to make it more efficient.
The sufficient freedom gave rise to new empowering tools that allowed Internet
users to explore social interactions like never before. As a result of this, the idea
of social media came to life.
2.3 Social Media
Social media and Web 2.0 are sometimes used interchangeably but they are far
from identical terms. Kaplan and Haenlein (2010) defined them as “A group of
Internet-based applications that build on the ideological and technological foundations of Web 2.0, and that allow the creation and exchange of user-generated
content” (p. 61). Deep into this definition lies the difference between social media and Web 2.0. Web 2.0 is a concept or philosophy while social media describe
applications. To put that into perspective with a metaphor, the hippie movement
during the 1960s was a culture or a set of ideologies and we can see applications
19
of these ideologies in Woodstock Festival in 1969 or the Sunset Strip in L.A. that
became a hippie gathering area at the time. The case is similar with social media
that encompass an incredible amount of various new applications.
Kaplan and Haenlein (2010) also created a classification for social media based
on social presence/media richness and self-representation/self-disclosure which
include, Blogs, Collaborative project (e.g., Wikipedia), Social networking sites
(e.g., Facebook), Content communities (e.g., YouTube), Virtual social worlds (e.g.,
Second Life), Virtual game worlds (e.g., World of Warcraft). They defined these
classifications based on principles that came from media theory. In short, media
differ in terms of social presence which can be influenced by intimacy and immediacy (Short, Williams, & Christie, 1976), and media richness which controls the
amount of information that can be transmitted in a given time (Daft & Lengel,
1986). The second combination of factors are self-representation which states that
people want to control their impression formation on others in social interactions
(Goffman, 1959) and self-disclosure which identifies information disclosure to
others whether consciously or unconsciously (Kaplan & Haenlein, 2010). Kaplan
and Haenlein (2011) expanded this classification to include micro-blogging which
stands “halfway between traditional blogs and social networking sites” (p. 106).
Figure 2.1 depicts the classifications of social media (Kaplan & Haenlein, 2010,
2011).
Figure 2.1: Classifications for social media
This is comprehensive list for social media. There is a great amount of literature on social media for all of these classifications. A short description for each
is provided here for reference however, most of this paper addresses general features that apply to most social media classifications.
2.3.1 Blogs
A blog (Web-log) is a Web site that contains entries arranged in reverse chronological order and it functions as an online journal where many people can contribute
(Boulos, Maramba, & Wheeler, 2006). Moreover, “Entries contain commentary
and links to other Web sites, and images as well as a search facility may also be
included” (Boulos et al., 2006). This function of linking between Web sites is what
makes them critical in sharing information.
Blogs are extremely popular and used in a variety of different ways. However,
their efficiency is debatable. Lawson-Borders and Kirk (2005) state that “blogs are
not the ideal tool for either political reporting or historical analysis” (p. 557). Re20
gardless, messages on blogs seem to have a higher longevity which makes them
ideal as journalistic tools, as shown in the Elections of 2004 (Lawson-Borders &
Kirk, 2005). Gumbrecht (2004) described blogs as “protected spaces” (p. 1). They
become a safe haven for bloggers that can feel safer from retribution. Benefits
seem to rise also with the use of blogs in distance learning. But again, certain
drawbacks still exist such as feelings of isolation and alienation for the students
(Dickey, 2004). As Dickey (2004) suggests, there are methods and strategies that
can be employed to alter the design for these blogs in order to counter these effects. Figure 2.2 depicts an example of a blog.
Figure 2.2: Google’s official blog.
2.3.2 Collaborative Projects
Blogs and collaborative projects have many similarities. One of the most famous
forms of a collaborative project is a wiki. This is described as “[...] a set of linked
Web pages created through the incremental development by a group of collaborating users (Leuf & Cunningham, 2001) as well as the software used to manage
the set of Web pages” (Khosrowpour, 2008, p. 204). Students generally find wikis
as good tools for project collaboration (Chao, 2007). In fact, there are numerous ways that people can use wikis especially for educational purposes (Duffy &
Bruns, 2006) and also in corporate environments where people report that their
groups are sustainable and wikis help them enhance their reputation, make their
work easier and help the organization improve its processes (Majchrzak, Wagner,
& Yates, 2006). Wikis also possess an extreme adaptability towards team goals
producing a great variety of wikis that are modified to serve the needs of their
communities. Wikis can be used to create collaborative research papers, encyclopedias and even facilitate debates (West & West, 2009). They allow for non-linear,
evolving, complex and networked text, argument and interaction (Black, Delaney,
& Fitzgerald, 2007). Ebersbach, Glaser, Heigl, and Warta (2008) have a detailed
21
description for installing and using Wikis especially for MediaWiki which is famous for its use in one of the most popular wikis today, Wikipedia and its sister
projects. Figure 2.3 depicts Wikipedia, an encyclopedia wiki.
Figure 2.3: Wikipedia’s encyclopedia page on pseudonymity.
2.3.3 Social Networking Sites
Social networking sites arguably became the center of a media frenzy since their
appearance early in the previous decade. Kaplan and Haenlein (2010) provided a
definition:
Social networking sites are applications that enable users to connect
by creating personal information profiles, inviting friends and colleagues to have access to those profiles, and sending e-mails and instant messages between each other. These personal profiles can include any type of information, including photos, video, audio files,
and blogs. (p. 63)
They become an extension of an individual’s real life containing a detailed
documentation of a person’s social network along with aspects such as their experiences, thoughts, beliefs and preferences. Social networking sites seem to be
extremely helpful for people with low self-esteem and low life satisfaction and
provide a tremendous advantage for managing social capital (Ellison et al., 2007).
Social capital as defined by Coleman (1988), described anything that can facilitate
individual or collective action generated by a network of social relationships and
established three forms for social capital, obligations and expectations, information channels, and social norms. Put simply, it is the value of social relations that
helps provide benefits to individuals or groups. As such, the term became a measure of well being for groups and the society. In fact, one study finds that the more
frequent users were participating in friend networking sites, the more numbers
22
of relationships are expected to form and receive positive feedback from these relationships (Valkenburg, Peter, & Schouten, 2006). As a result of this, it was further concluded that positive feedback does not only enhance social self-esteem
but well being as well. Furthermore, benefits of using social networking sites are
not just limited to individuals but apply on corporations as well. Research is favoring social networking sites for supporting brand promotion and marketing
campaigns (J. Brown, Broderick, & Lee, 2007; Sicilia & Palazón, 2008). Social networks can also be profitable business models (Murchu, Breslin, & Decker, 2004).
Figure 2.4 shows a company’s fan page on Facebook, a social networking site.
Figure 2.4: IBM’s fan page on social networking site Facebook.
Many of the benefits of social networking sites are overshadowed by their disadvantages. One of these is the differential adoption due to the digital inequality (Hargittai, 2008). This digital divide has economic, sociological and political
drivers that affect not just the adoption of social networking sites but the adoption
of the Internet (Guillen & Suarez, 2005). Furthermore, while the digital divide excludes a group of people from using this technology, problems arise for the ones
that actually use it. Perhaps one of the most popular issues currently on social networking sites is privacy (Barnes, 2006; Lenhart & Madden, 2007; Rosenblum,
2007). Specifically there is a tendency by young people to disclose personal information online without considering who the recipients are (Barnes, 2006). In
addition, Gross et al. (2005) have argued that this tendency of disclosing personal
or even sensitive information can lead to issues such as stalking, re-identification
by using information from various social networking services, building a digital dossier for an individual, and general change in attitudes where people will
tend to reveal information more freely. Perhaps, the magnitude of the problem
becomes apparent if one considers a study that showed a fifteen percent overall in user profiles between two major social networking sites overlapped (Liu &
Maes, 2005). Considering that at the time the usage of social networking sites
was not as popular as today, one could perceive that these numbers are probably
higher today. As an effect of this, an assailant wanting to identify a specific person
that uses a pseudonym can do so by cross-referencing information across various
social networking sites until he or she gets a match.
23
Regardless of the above dangers, people are more than willing to join social
networking sites especially when their perception of enjoyment from the service
is high (Lin & Lu, 2011). As seen before, as long as there is a sociological need
these services are expected to keep on growing.
2.3.4 Content Communities
Content communities are mainly geared towards exchanging and sharing content between users (Kaplan & Haenlein, 2010). The content can be in any type
of a multimedia format with websites allowing one or many formats. Content
communities such as YouTube (depicted on figure 2.5) are used by the media industries, amateur producers and communities of various interests providing new
opportunities that challenge the barrier between production and consumption
(Burgess & Green, 2009). In addition, although the main focus may be different
than in social networking sites, many of the benefits and drawbacks still exist
here. In fact, the idea of social networks forming on content communities is not
so farfetched either (Lange, 2008). Specifically, Lange (2008) conducted an oneyear ethnographic research studying the participants behavior in terms of how
much information they disclosed about themselves as producers and how big
was their audience. The social networking effects are also discussed on another
study that sees small-world characteristics in the links generated by contributors’
choices which in turn could improve service quality for other content communities if novel techniques are developed and implemented (Cheng, Dale, & Liu,
2008).
Figure 2.5: Masaryk’s University promotional video on content community,
YouTube.
On the other hand, challenges seem to increase when dealing with content
sharing. Issues of copyright and privacy that first started appearing with P2P
applications are now expanding in content communities and leading companies
need to establish a new set of guidelines (Pike, 2007). Furthermore, along with
24
these guidelines, new tools need to be developed. In fact, YouTube developed a
technology that can process the vast amounts of data being uploaded everyday
on their servers and identify which content breaches copyright rules (Youtube,
2012a). In another case, Stewart (2010) presented how the software works in front
of an audience in order to ensure transparency and alleviate any worries that
copyright owners may have.
2.3.5 Virtual Social Worlds
Virtual worlds are online environments similar to games but they have no specific
goal or narrative for their players other than socializing. Some may have certain
basic rules and limitations (e.g., users need to eat everyday) while others are less
restrictive (e.g., users have the ability to fly). As Kaplan and Haenlein (2010) argue, one expects to see behavior that is similar to real life settings (Haenlein &
Kaplan, 2009; Kaplan & Haenlein, 2007, 2009). Schroeder (2002) goes a step further by also defining that only certain virtual environments which he calls IVEs
are giving the ability to individuals to immerse themselves into the virtual world
and perceive themselves as part of them.
Individuals not only have the opportunity to immerse themselves in the virtual world but they have the power of controlling how they would achieve that
and in what manner. One of the interesting phenomena that can be observed
online is the practice of gender-switching where an individual assumes an online identity that has a gender that is different from their biological sex (Roberts
& Parks, 1999). This control helps them form or reform their identities according to their desires and have control over their impression formation (Wallace,
2001). Players not only have a set of tools at their disposal but they can assume
new roles and obtain new skills. As an example, aside from learning how to fly
in Second life, users also acquire skills such as handling unstructured problems,
dealing with the game’s economy and politics, socialize and sustaining relationships, negotiating and trading resources and so on (Papargyris & Poulymenakou,
2005).
Additional benefits which may not come to mind at first are for businesses as
well. Reeves and Read (2009) stated that concepts of the trend known as gamification will inevitably take over the business. The idea of a workplace that encompasses virtual awards, avatars and three-dimensional environments may seem a
bit farfetched but, as long as there is a sociological need for these processes one
could expect to see technology evolving in order to accommodate this need.
2.3.6 Virtual Game Worlds
Virtual game worlds are a more restrictive version of a virtual environment. Users
have to follow a strict set of rules and behave in a certain manner. These provide
people with a tremendous potential for role-playing and entertainment guided
by a fictitious story narrative. In fact research shows that “story, graphics, length,
and control are highly related to enjoyment” (Wu & Boulevard, 2008, p. 219).
They additionally provide strong social interaction for people to form not only
relationships with each other but eventually groups and communities. Their pop25
ularity especially for the young population makes them a potential tool for not
only helping people attain and improve skills such as logical reasoning and teamwork but also design their own storyline for games and eventually implement
such process in classrooms (Robertson & Good, 2005). Designing these worlds
is of particular interest to educators, business leaders, and entertainment executives looking to benefit from these environments (Squire & A., 2006). Researchers
from all fields are also excited about the potential for investigating various social, economic, medical, political and other effects in these online worlds. Lofgren
and Fefferman (2007) investigated a pandemic in a virtual game world (World of
Warcraft) and drew parallels on what could have happened in real life in case a
similar pandemic were to emerge. These worlds are also breaking new ground
for computer science research (Bainbridge, 2007). Trying to accommodate sociological needs that require advanced features for players such as being able to be
creative and develop their own items as well as being able to share these items
with others, creates severe delays in the communication of these systems. In turn,
delays in online games affect the willingness of an individual to continue using
the service (K.-T. Chen, Huang, Huang, & Lei, 2006). These results create new
technological needs that will help push the current momentum of social media
software forward.
2.3.7 Microblogging
According to Kaplan and Haenlein (2011), microblogs are a set of Internet-based
applications “which allow users to exchange small elements of content such as
short sentences, individual images, or video links” (p. 106). An example of such a
medium is Twitter which is depicted on figure 2.6. Similar to other social media,
microblogs have a great potential for individuals as well as the industry. They
are seen as potential great tools for collaborative work which can increase social
and emotional welfare in the workplace as well as the overall flow of information
within an organization (Ehrlich & Shami, 2010; Zhao & Rosson, 2009). They
also provide many opportunities for learning (Ebner & Schiefner, 2008). Another
research on Weibo, a Chinese microblogging service, reveals that the medium is
likely to increase participation in online protests (H. Li, 2012).
However, as with other social media cases, microblogging is plagued by several issues. C. Lee and Warren (2010) discussed some of the major issues that rise
with the use of microblogging services in the workplace. These relate to the information trust, privacy, confidentiality, and protection against scams and frauds.
Privacy and security was also a major concern discussed by Zhao and Rosson
(2009). Sensitive information that can be leaked without a user’s knowledge as
well as preset privacy settings for user accounts, should be a top concern for
designers and developers. Similarly, by having microblogging in the workplace,
sensitive work information may leak with severe consequences for organizations.
26
Figure 2.6: Twitter page for the Faculty of Social Studies, Masaryk University.
2.4 Additional Considerations
These classifications demonstrate the diversity of social media services and the
necessity of properly defining the boundaries between each. However, even social media services under these classifications are not always consistent. Boundaries may overlap and types can be combined to include two or more classifications along with their features. Additionally, even within one classification differences can be found.
One such example can be found on an article written by Ellison et al. (2007).
They argued that the terms Social Networking Sites and Social Network Sites are
used interchangeably but the two vary. The term Networking is describing formations of new relationships usually by strangers. In contrast, they argue that
what makes Social Network Sites unique is that they “enable users to articulate
and make visible their social networks” (Ellison et al., 2007). As such, they identify three unique elements that make social network sites based on the ability to
(Ellison et al., 2007):
•
construct a public or semi-public profile within a bounded system.
•
articulate a list of other users with whom they share a connection.
•
view and traverse their list of connections and those made by
others within the system.
These elements can vary however, depending on the structure of the social
network/ing service. In addition many of these features can be found in several
other social media that in effect become a hybrid.
The above idea is reinforced by the recent trend to use the term social networks and online communities interchangeably. However, Howard (2010) makes
a clear distinction between the two and provides the fundamental differences.
27
One difference is the lack of secondary relations that can be found in online communities but not on social networking sites. But perhaps the most important difference is the fact that communities can exist on a variety of social media services
while social networking sites are just one service. Put simply, online communities
are not confined to just one social media service necessarily. Furthermore, social
networks are also considered to be limited for certain types of activities that were
described in what is coined as Shirky’s ladder (Howard, 2010; Shirky, 2009). In
general, the ladder contains activities that increase in complexity and as such the
relationships between individuals need to be stronger in order to achieve higher
activities in the ladder. Howard (2010) concludes that while social networks are
ideal for sharing activities, when it comes to activities that require cooperation
or collective action they are insufficient. On the other hand, online communities
containing strong secondary relationships are more likely to be able to accommodate for such more complex activities. This as an idea should not sound so
absurd if one considers that online communities have a broad set of rules, ethics
and social norms that form a more complex social structure than just a simple
social network. Of course, one may still argue that today many social network
sites, or social networking sites, provide additional services that help organize
people into online communities. One such example is the famous Facebook application Causes which not only became a sustainable for-profit model of a social
networking application but also collective actions taken by Facebook users raised
$40 million dollars for 27,000 nonprofit organizations (Causes, 2012).
Many also have been mesmerized by the idea of merging Web 2.0 technologies with the principles of a semantic web in order to form the next platform for
web, the Web 3.0 (Giustini, 2007; Hendler, 2008, 2009; Lassila & Hendler, 2007;
Wahlster et al., 2006). The excitement is not unjustified considering the tremendous benefits that semantic web can bring to the current social media landscape.
Some of these benefits and implementations of semantic web in current social
media are demonstrated by Breslin, Passant, and Decker (2009) and Mika (2007).
Looking into the future we can now see that the ideas originally written by
Bush (1945) and tackling the information overload still exist today. Web 2.0 and
3.0 are nothing but evolutionary manifestations of the Web 1.0 that were originally created to serve the exact same sociological needs. These are the same sociological needs used to establish the predictions of Hiltz and Turoff (1978) as
pointed out by Howard (2010). Needs that one day may even transform our
workplace by using virtual social and game worlds in our businesses (Reeves
& Read, 2009).
But, as we move forward towards a brand new future with technologies that
are evolving with a rapid pace, one still needs to remember that technology may
change but people do not. In fact, while people are still referring to social media as something new that does not appear to be the case. Gundry (2006) points
out in the year 1991 in a world prior to the World Wide Web, there was a generation of social media that was successfully used in organizations. In his paper
at the time, he identifies the tremendous advantages seen by the use of collaborative technologies for educational purposes in organizations (Gundry, 1992).
Coincidentally, earlier on Hiltz (1985, 1990); Hiltz and Turoff (1985) identified in
28
their work the tremendous potential for the networked software that one day will
become part of the new Web which among its various application one can find
social media.
2.5 Conclusion
This is the current state of the social media services that exist out there along with
some of the major issues they are facing. This chapter aimed to demonstrate the
variety of the services that exist and also the challenge of producing software that
is deeply intertwined with everyday social activity. When a decision is made by
an individual or a corporation to implement a social media service to accommodate for a set of interactions, the design of the service can have a great impact
on the efficiency of that interaction and the overall individual and social experience. As such, developing social media that will account for these experiences
requires a revolution in the way designers and developers think and progress
through software development. The first goal for such a development requires
one to identify what desirable and undesirable social experiences are for social
media communities. This will also depend on what type of a social media service
is developed. In the following chapters I demonstrate how research on software
features and social behaviors can assist in a development of better social media
guided towards achieving goals related to these desirable and undesirable phenomena. A deep understanding of sociology and psychology is important for
understanding the effects of software on social interactions.
29
C HAPTER 3
G ROUPTHINK T HEORY AND A PPLICATION
3.1 Introduction
This chapter provides an in-depth review on groupthink. The relevance of groupthink in this dissertation research was already mentioned in chapter 1. In this
chapter, I elaborate more on the relevance of groupthink in social media. Definitions for groupthink are presented along with a literature review for groupthink
in real world and on the Internet. Furthermore, the features under examination
by this study are presented here and will be analyzed in the next chapter.
3.2 Defining Groupthink
Janis (1982) defined groupthink as:
A mode of thinking that people engage in when they are deeply involved in a cohesive in-group, when the members’ strivings for unanimity override their motivation to realistically appraise alternative
courses of action. (p. 8)
This definition identifies some of the causes of groupthink. People within a
group try to minimize conflict and reach a consensus without proper and critical evaluation of all the information available to them, individually as well as a
collectively.
A more modern definition for groupthink was given by the Merriam-Webster
Dictionary (Groupthink, 2012).
a pattern of thought characterized by self-deception, forced manufacture of consent, and conformity to group values and ethics
This definition demonstrates the effects of groupthink. It describes what happens while a group is experiencing groupthink. This behavior goes past simply
conforming with the group. Individuals force themselves to accept information
in order to help a group reach an agreement. Arguably, this combination is what
makes groupthink severely destructive in group decision-making processes.
30
3. G ROUPTHINK T HEORY AND A PPLICATION
It is not difficult to imagine why groupthink is an unwanted behavior in
groups and communities in everyday life. Furthermore, it is an unwanted behavior in online groups and communities that exist in social media, especially for
services meant for collaboration. For example, it is a serious issue for the development of Group Decision Support Systems (GDSS) which include wikis (Tsikerdekis, in press-c).
3.3 Examples of Social Media Endangered by Groupthink
To understand the involvement of software designers for preventing groupthink
in social media, two examples are presented. A team was asked to develop a social media software, and more specifically a collaborative project for a classroom.
A decision was made to use wiki software which will lead students in participating in a collaborative effort to create and improve on the current learning
material, but also discuss issues that exist in class and try to provide solutions. A
defective decision-making process in this setting could have devastating and unforeseen consequences especially when dealing with a population of non-adults.
Considering that wikis are considered to have tremendous potential for classroom settings and education in general, this scenarios is not far-fetched (Ferris
& Wilder, 2006; D. Turner, 2004). Another example could be the assignment of
developing software that will accommodate decision-making for an IT company.
A group may be developing new projects or working towards problem resolution for issues in current projects or even their clients’ projects. It becomes clear
that a solid and efficient decision-making process is essential for business success.
Employees may need to hold multiple meetings and information may be stored
in wiki software in order to help weight their options better. The idea of wikis
for business collaboration, although requires changes in the current management
culture (Edwards, 2007), is still considered beneficial (Kussmaul & Jack, 2009;
Wood, 2005).
These two examples demonstrate the tremendous potential for social media
being used in various settings that will involve collaboration. In fact, groups are
really easy to form in any community in general. D. H. Smith (1967) gives a good
definition of a group based on a couple of criteria.
A group is defined here as (1) the largest set of two or more individuals who are jointly characterized by (2) a network of relevant communications, (3) a shared sense of collective identity, and (4) one or more
shared goal dispositions with associated normative strength. (p. 141)
The above statement demonstrates that groups not only are easy to form but
they can vary in size. Another definition was given by Levine and Moreland
(1994). They stated that a group is several people that “interact on a regular basis,
have affective ties with one another, share a common frame of reference and are
behaviorally independent” (Levine & Moreland, 1994, p. 306).
Based on the above, it becomes clear that social media are bound to have
groups within them whether a collaborative project for a company or an online guild of individuals in a virtual game world. Wherever one can find groups,
31
group phenomena are bound to exist and this rule applies also to social media.
However, people can alter the virtual environment produced by the software of
social media, unlike the real world situations where people have limited control
over the environment. In fact, the software is entirely manufactured. As such, a
question that comes to mind is, could we alter the software in such a way that we
can affect group interactions, and as such protect groups from dangerous group
phenomena such as groupthink?
3.4 The Beginnings of Groupthink and Research
To understand how groupthink prevention became a popular idea one has to
understand how the model of groupthink was formulated. Groupthink traces its
roots to March 1952. Whyte (1952) wrote about groupthink:
Groupthink being a coinage - and, admittedly, a loaded one - a working definition is in order. We are not talking about mere instinctive
conformity - it is, after all, a perennial failing of mankind. What we are
talking about is a rationalized conformity - an open, articulate philosophy which holds that group values are not only expedient but right
and good as well. (p. 114)
Later on, Janis (1971) developed the initial model on groupthink. He wrote
that groupthink leads “to a deterioration in mental efficiency, reality testing and
moral judgments as a result of group pressures” (p. 43). He followed up with his
research on several publications (Janis, 1972, 1982).
3.4.1 The Groupthink Model
The model of groupthink that Janis (1972) created provided explanations for groups
with defective decision-making. Janis and Mann (1977) further established the
groupthink model as a five-step casual model (Henningsen, Henningsen, Eden,
& Cruz, 2006).
Antecedents
The first step included a set of antecedents such as high levels of cohesiveness,
structural defects and provocative situational contexts (Janis & Mann, 1977).
Structural defects can be analyzed further as insulation of the group, lack of impartial leadership, lack of norms that require methodological procedures, and
homogeneity of the social background ideologies of members. Similarly, provocative situational contexts can appear in terms of highly stressful external threats
for the group, other recent failures, significant difficulties in the decision-making
task and moral dilemmas that may arise in the collaborative progress.
Cohesiveness in particular is perceived as the single most important hazard
for groupthink (Janis, 1982). This seems to be in agreement with a later metaanalysis study conducted by Mullen, Anthony, Salas, and Driskell (1994) which
associated high group cohesiveness with defective decision making. Also, Janis
32
(1982) noted that groupthink cannot occur unless all the antecedents are present.
However, Hart (1991) described that high cohesion is important but groupthink
cannot occur unless one or both of the other antecedents are present; asserting
that groupthink could occur without the need for all antecedents to appear in a
group.
Concurrence seeking
In the second step, concurrence seeking is leading members to openly agree with
a group’s perceived position although they privately disagree (Janis, 1972). Coincidentally, this behavior is similar to conformity described by Asch (1958, 1992)
where the distortion at the action level affects individual judgment in group
decision-making. It is understandable that it takes a significant force to lead someone to consciously avoid expressing their opinion to a group. This may be why
this behavior is so dangerous. If an individual withholds a key piece of information, this information will never be put up for consideration by his or her group.
Symptoms of Groupthink
As a result of the previous two steps, symptoms of groupthink start occurring.
These include illusion of invulnerability, illusion of morality, collective rationalization, excessive stereotyping, self-censorship, illusion of unanimity, pressure
of conformity and self-appointed mindguards (Janis & Mann, 1977). These were
later revised into three categories: overestimation of the group, close-mindedness,
pressures toward uniformity (Janis, 1982). They establish pressure on group members to not only conform with the group but also perceive that the decisions made
by the group are productive and successful (Henningsen et al., 2006).
Decision-making defects
Consequentially, a group that starts experiencing the symptoms of groupthink
eventually may experience deficiencies in the decision-making process as well.
Janis and Mann (1977) identified the following defects: incomplete survey of alternatives, incomplete survey of objectives, failure to examine risks associated
with the preferred choice, poor information search, selective bias in processing
information, failure to reappraise alternatives, and failure to provide contingency
plans. Essentially, the group is “[...] processing information ineffectively” (Henningsen et al., 2006, p. 38).
Poor decision outcomes
The final outcome of all the previous steps lead to a faulty decision by the group
decision-making process (Janis, 1972, 1982, 1989). The faulty decision at this point
is usually not the optimum one. Additionally, while Janis (1989) pointed out that
groups may arrive in a good decision even under the influence of groupthink,
Henningsen et al. (2006) reported that most accounts under which groupthink
had occurred point to a negative outcome and poor decisions being taken.
33
3.4.2 Historical Examples of Groupthink
The model of groupthink might have become popular because the above model
is fairly easy to understand and follow, but also due to popular historical cases
that Janis analyzed (Esser, 1998). Some of the fiascoes in human history that are
attributed to groupthink are: the Bay of Pigs Invasion, the attack on Pearl Harbor,
and the space shuttle Challenger disaster. These cases are going to be described
below but there are additional cases such as the North Korea escalation or the
Vietnam escalation (Rose, 2011) that were originally studied by Janis (1972) for
groupthink as well.
Pearl Harbor
On December 7, 1941 the Imperial Japanese Navy attacked Pearl Harbor in a surprise military strike. Janis (1982) argued that the naval officers that were stationed
in Hawaii had shared illusions and rationalizations which contributed to lack of
being on alert. The United States intercepted messages from the Japanese side that
was preparing for an offensive attack and warned the officers in Pearl Harbor, but
their warnings were not taken seriously. The army officers had succumbed to social pressures and additionally were rationalizing their beliefs for how unlikely
an attack on Pearl Harbor was by the Japanese. Some of the statements are illuminating (Janis, 1982, pp. 83-85):
•
•
•
•
“The Japanese would never dare attempt a full-scale surprise assault against Hawaii because they would realize that it would
precipitate an all-out war, which the United States would surely
win.”
“The Pacific Fleet concentrated at Pearl Harbor was a major deterrent against air or naval attack.”
“Even if the Japanese were foolhardy to send their carriers to attack us, we could certainly detect and destroy them in plenty of
time.”
“No warships anchored in the shallow water of Pearl Harbor
could ever be sunk by torpedo bombs launched from enemy aircraft.”
All of these statements testify to a sense of invulnerability that the army felt.
This was part of symptoms that should have raised red flags. At the time nobody wanted to question or oppose mainstream beliefs. Eventually this lead to
decision-making defects by dismissing information and being inactive and finally
the outcome of this decision that lead to the disaster of Pearl Harbor. Arguably,
had they known about groupthink and applied preventive measures, this disaster
might have been limited, if not averted.
Bay of Pigs Invasion
Twenty years after Pearl Harbor, the Bay of Pigs invasion incident stands as
the primary example used by Janis (1982) to describe groupthink. The invasion
34
plan that Kennedy used targeted to overthrow Castro’s Cuban government. Originally, the plan was initiated by the Eisenhower administration and later the
Kennedy administration adopted CIA’s plan. There is enough evidence that portrays that this information was not critically evaluated (Janis, 1972) but also those
that tried to raise concerns or object were questioned about their loyalty. The
group believed in the morality of their plan or even self-censored their thoughts.
The sense of invulnerability was also high as well as stereotyping Castro’s administration. At the end, within few hours after the invasion, all the invaders
were killed or captured. Later on, in the case of Cuban Missile Crisis, president
Kennedy employed preventive measures for meetings in order to avoid a similar fiasco. On the other hand, concluding that the Bay of Pigs fiasco is an event
caused just by groupthink may also be presumptuous. As noted by Kramer (1998)
aside from group dynamics there may have been political reasons that could also
explain the Bay of Pigs event. They called for a “multiple lens” perspective when
investigating such organizational fiascoes.
Space Shuttle Challenger
Another popular disaster that given the evidence is attributed to groupthink, is
the case of the space shuttle Challenger disaster. The official report released by
the NASA History Office (Scobee et al., 1986) after the accident is illuminating:
The decision to launch the Challenger was flawed. Those who made
that decision were unaware of the recent history of problems concerning the O-rings and the joint and were unaware of the initial written
recommendation of the contractor advising against the launch at temperatures below 53 degrees Fahrenheit [...] They did not have a clear
understanding of Rockwell’s concern that it was not safe to launch
because of ice on the pad. If the decision makers had known all of
the facts, it is highly unlikely that they would have decided to launch
51-L on January 28, 1986 [...] based on incomplete and sometimes misleading information, a conflict between engineering data and management judgments, and a NASA management structure that permitted
internal flight safety problems to bypass key Shuttle managers. (p. 82)
Research into this specific accident in relation to groupthink has been positive.
Esser and Lindoerfer (1989) conducted quantitative research by using archive
records and positively identified an increase in the observable antecedents and
consequences defined by Janis (1982) in the last 24 hours prior to the accident.
Another study into the accident by Moorhead, Ference, and Neck (1991) also
found substantial evidence of the same antecedent conditions, the symptoms, and
the decision-making defects and concluded that the Challenger disaster should
be classified as a “groupthink situation.” Additionally, they revised the original
model in order to include two more variables, time and leadership style.
35
3.4.3 A Model of Controversy
Literature review articles today seem to agree that research and experimental
studies on groupthink have been limited (Esser, 1998; Rose, 2011). In general
most of the offline research investigated popular cases such as the 1999 baseball
umpire strike (Koerber & Neck, 2003), Kent State University gymnasium controversy (Hensley & Griffin, 1986), France’s 1940 WWII defeat (Ahlstrom & Wang,
2009), Iran Hostage Crisis (S. Smith & Jan, 1985) and other case studies which provide support for the theory. According to a review on groupthink by Rose (2011)
there have been also fifteen experimental studies that investigated groupthink.
Erdem (2003) showed that a high degree of trust can increase groupthink, Ahlfinger and Esser (2001) showed that promotional leadership partially supported that
it can affect some of the antecedents and consequences, and M. E. Turner, Pratkanis, Probasco, and Leve (1992) found support for groupthink by testing many of
its antecedents and consequences.
While it may seem that researchers have reached an agreement after decades
of studies since the creation of initial model, there is still a lot of controversy over
the topic. As Rose (2011) pointed out, while Mitchell and Eckstein (2009) state that
groupthink has been widely accepted, W.-W. Park (2000) states that “very little
consensus among researchers on the validity of the groupthink model” (p.873).
This is far from an unfounded statement given that studies have also failed to
make a direct connection between groupthink and the processes studied (Kramer,
1998; Maier, 2002). Some of the well tested areas such as insulation and impartial
leadership show support of the model, while cohesion produced mixed results,
mainly because of the diversity of perspectives in cohesion (Rose, 2011). Some
attribute those mixed results in the lack of standardization for measuring the antecedents and combining the results (Esser, 1998), while others in the general
lack of multiple studies that explore all the antecedents and consequences (Rose,
2011).
As Rose (2011) indicated this started a debate. The result of this is some stating
that “the model represents a brilliant construction founded in part on the existing group dynamics literature” (Paulus, 1998, p. 371) while others state that “in
our view, groupthink is a compelling myth” (Fuller & Aldag, 1998, p. 177) that
was accepted without substantial evidence. Perhaps the middle ground could be
considering positive as well as negative evidence for groupthink, and provide
certain revisions to the initial model. Some modifications to the initial model are
provided by Chapman (2006), Flippen (1999), and Neck (1995) with a main focus
on adding more variables to the initial model. A full survey of all the proposals
is covered by Rose (2011).
For now, the current model is highly popular and being applied to jury decisionmaking (Mitchell & Eckstein, 2009), hockey teams to improve performance (Rovio,
Eskola, Kozub, Duda, & Lintunen, 2009), academia and other cases (Klein &
Stern, 2009; Rose, 2011). Paying close attention for preventing groupthink in
groups is not unjustified. Having looked in the popular fiascoes with case studies
as well as experimental studies, if groupthink is not a myth the implications can
be severe. One can survive a false positive but can never survive a false negative.
Whether the model is a figment of the imagination or reality, it does not exclude
36
the fact that it is a model that tries to explain what happened in all of these popular cases and probably is the best model that does this to this day. The debate
goes on with research trying to argue for and against the model, regardless of
the initial acceptance of the model. This paper agrees with such a strategy. In the
light of past and current evidence which can be found in literature reviews (Esser,
1998; Rose, 2011), some antecedents and consequences of the original groupthink
model have been positively observed. It is also evident that some variables have
not been tested at all, supporting the argument made by Rose (2011) for a limited
research by experimental studies.
This paper argues that from an HCI perspective, this debate can also become
trivial. Whether the model provides a substantial predictability of the decision
outcomes of groups, or it needs revisions, the implications of such realizations are
not limiting HCI research that investigates groupthink. In HCI, the focus should
mainly be the application of such processes into the software development based
on empirical evidence. As an example, consistent studies show that insulation
(Moorhead et al., 1991) and impartial leadership (Flowers, 1977; Fodor & Smith,
1982; Leana, 1985) play a significant role to defective decision-making processes
and poor decision outcomes. Hence undoubtedly, even if the groupthink model
needs to be revised, the implications and goals for HCI researchers and software
engineers that need to apply this knowledge in software design and development
practices do not change.
This approach seems to be in accordance with the initial goal of Irving L. Janis
that as indicated by M. E. Turner and Pratkanis (1998), he was always interested in
the practical significance of the research. That is, the goal of the groupthink model
was not just to explain why groups make bad decisions but also to establish ways
that this can be prevented. This is what makes the groupthink model a valuable
tool for software designers wanting to apply this knowledge in various social
media services.
3.4.4 Preventing Groupthink
Given the variables presented from the groupthink model we can make certain
speculations when attempting to prevent groupthink. In effect, by inhibiting the
antecedents and consequences one can expect that the probabilities for groupthink occurring within a group will decrease.
In general, one can expect that groups with more information available to
them will produce more alternatives. In addition, Janis (1982) showed that a
group’s high cohesion will lead to a group making compromises and impair the
decision-making ability of the group. Consensus may be reached faster but critical thinking may be impaired for the members of a highly cohesive group. These
processes can be found in larger groups as well (Koerber & Neck, 2003; McCauley, 1998). Hart (1991) considered extremely important to actively combat
the effects of groupthink.
Seven measures were provided by Janis that could prevent groupthink (Janis,
1972, pp. 209-215):
•
One member of the group should always be assigned the role of Devil’s
37
advocate. This person needs to be different for each meeting.
•
Each member should be assigned the role of a critical evaluator by the leaders. Objections and doubts should be presented freely.
•
Leaders should not present their preference initially.
•
All effective alternatives should be examined.
•
Independent groups should be set up and work on the same task.
•
Each member should discuss ideas with outsiders.
•
Outsiders should be brought in the meetings.
When applying the following guidelines, groupthink can be averted. In fact,
they were applied during the Cuban Missile Crisis by President John F. Kennedy
(Hart, 1991). Of course, these are not but just rough guidelines. As described before, any given method that can inhibit antecedents or consequences can be used
as a preventive measure. Each individual condition that inhibits those variables
that are considered to be a cause for groupthink can help avert potential disastrous decision outcomes. Identifying conditions that have such power is essential
especially when developing software where software engineers have a lot of control over the environment that their users will collaborate in.
3.5 Groupthink and Online Groups
Developing an online environment in order to reduce groupthink behavior has
seen even less interest from the scientific community. However, the model is arguably, tremendously useful if one considers the benefits of implementing preventive measures that can reduce groupthink.
Back in the two original examples that were given early on in this chapter,
one can consider the necessity of assigning an individual the role of the devil’s
advocate or the iconoclast. An iconoclast always tries to redefine whatever is considered already as a well-accepted consensus. In effect, this role exists with a sole
purpose to disagree and point weaknesses to the group’s general plans. In both
examples this role can automatically be assigned based on the profile characteristics of individuals. The assignment of the role would not only be less biased
since the software would do most of the work, but it could also ensure that the
individual selected as the devil’s advocate fits the psychological profile of person
comfortable with that role. This can also be reinforced with the use of anonymity
which for certain people can be a liberating factor to voice out their opinions.
However, such rules cannot be easily implemented in groups collaborating on
the physical world.
On the other hand, software for online collaboration is extremely useful. In
fact, one of the arguments supported by this dissertation is that developers do
not consider the tremendous potential and opportunities that online software has
for controlling behavior. Most of the aspects described in the previous paragraph
can easily be implemented in social media software.
38
3.5.1 Researching Online Groupthink
Most of the research that investigates online effects for groupthink, is focused on
GDSS. Right from the emergence of Group Decision Support Systems researchers
became excited about the potential of technologies to constrain the environment
and produce designs that could actually prevent certain group phenomena. DeSanctis and Gallupe (1984) wrote about the potential benefits of GDSS in a list
that stated several benefits along with the following statement:
The GDSS contains built-in mechanisms which discourage development of negative group behaviors, such as destructive conflict, miscommunication, or “groupthink". (p. 24)
It becomes evident that since the early beginnings of collaborative technologies people thought highly of their potential for preventing negative group phenomena. This idea was at the core of some early studies that followed on group
collaboration and GDSS (DeSanctis & Gallupe, 1985; Driscoll & King, 1988).
Such interest was not unjustified as later on group support systems were suspected of having an effect on groupthink (Miranda, 1994). Miranda (1994) reviewed antecedents and procedural conditions in order to establish how groupthink can be prevented in groups and what the implications are for these early
group support systems.
A similar research, only a few years back, investigated the effects of anonymity
in GDSS. Jessup et al. (1990) conducted a study that investigated the effects of
anonymity as a separate condition but it was also partially connected to groupthink. This is a versatile approach of tackling groupthink by investigating individual aspects that may contribute to groupthink. As mentioned before, groupthink may be caused by a number of variables that may or not be present. Hence,
it becomes particularly difficult to evaluate the actual occurrence of groupthink.
However, one can study the individual conditions that are known or suspected to
be causes for groupthink according to the case studies and afterward infer about
their implications.
Indeed group support systems show promising results by giving groups advantages of accessing external information (Miranda & Saunders, 1995); an important preventive measure for groupthink. On the other hand, the same study
shows that groups tend to discuss alternatives less which may lead to defective
decision-making. Alternatives play a key role for the chances of observing groupthink behavior in a group. It is also evident that it is not sufficient to achieve
enough contributions for alternatives if the group will not spend time considering them. In fact, one may argue that even the process through which alternatives
are evaluated is important; that is, to ensure an unbiased way for evaluating those
alternatives.
Troyer (2003) mentioned the importance of taking into account group phenomena when designing GDSS. Specifically one of the main arguments investigated the effects of positive and negative evaluations and their effects to groupthink. Specifically, the case of negative evaluations that is particularly important
for critically assessing all alternatives and the fact that shortcomings should be
39
thoroughly explored. It is no wonder that Cordell (1996) described that GDSS
software has the ability to prevent groupthink in a large extend, mainly because
of the ability to provide anonymity. The idea and potential benefits for anonymity
in GDSS is described by many (Kraemer & King, 1988; Nunamaker, Vogel, &
Konsynski, 1989; Nunamaker Jr., Applegate, & Konsynski, 1987) and it becomes
a core concept for this study as well.
Other research in GDSS in the past decade seems to investigate several group
phenomena in relation to GDSS with a general consensus that GDSS are effectively preventing Groupthink (Lawrence, 2002; Marreiros, Ramos, & Neves,
2004; Wadhwa & Bhattacharya, 2000). However, Shirani and Lee (2008) made
the following suggestion for research into computer-mediated communication
technologies:
Groupthink: CMC environments eliminate or reduce verbal and social
context cues. Which CMC environments, under what conditions, are
better at avoiding groupthink? (p. 5)
Arguably the question stated here is valid. While most researchers and in turn
software designers and developers seem to agree that groupthink is a condition
that can be avoided with group collaborating technologies, it is naive to state
that this applies to all GDSS software. This is particularly true if one considers
the lack of experimental studies for the conditions known to cause groupthink
for GDSS. In fact, even if most of the statements are grounded in theory, it is
essential communicating these results to software engineers so that software can
be developed properly. Evidently, it seems that the current consensus is that any
group decision support system and other broader collaborative projects that are
part of social media will guarantee up to a point avoidance of groupthink.
As it will become obvious during the next sections, this study showed that
this is not the case. Certain software features affect antecedents and consequences
that contribute to groupthink. If while designing software these are not considered properly, they can have devastating consequences and lead to groups experiencing groupthink. In addition, the current consensus about GDSS preventing
groupthink can make things significantly worse by having usability engineers or
software engineers acknowledging a groupthink-proof software while this may
not be the case always.
3.6 Software Features Studied for Groupthink
Based on the literature presented here, several software features were studied
to help assert the main goals of this dissertation. These were selected based on
the cost of studying them and being examples that showed potential based on
the literature. Since this dissertation work wanted also to encourage similar research, demonstrating an “easy to investigate” approach was of particular importance. In addition, elements had to be diverse as well. This guaranteed that
the study would investigate multiple perspectives of the problem of groupthink
but, it would also demonstrate a variety of ways for researching such aspects in
40
HCI. This study aimed to focus on features of social media software that had preventive potential based on the factors described in this chapter that are known to
contribute to groupthink. Features that contribute to information production and
reduce conformity pressures were at the core of this study. Several diverse features currently in use in social media were identified and investigated in terms
on their effect on individual and social behavior. Coincidentally, many of these
features were found to be easily adjusted on social media. A summary of the
main aspects that were studied is presented below.
•
Pro/con lists and their potential for reducing groupthink as compared to
textual collaborative interfaces. How effective are they in terms of argument
production, uniqueness and comprehension?
•
Anonymity states and their effects on non-conformity and groupthink. Can
they be implemented to improve collaboration?
•
Dynamic voting interface indicators and their effects online voting processes. Is the final voting in decision-making process biased and could contribute to groupthink?
•
Item randomization in voting process for improving collaboration and shieldproofing against groupthink.
•
Publicly identified administrators and leaders in online groups as potential
candidates to influence open voting results and contribute towards groupthink.
41
C HAPTER 4
R ESEARCHING THE I NDIVIDUAL F EATURES FOR
G ROUPTHINK IN S OCIAL M EDIA
4.1 Introduction
In this chapter, research analysis and findings will be presented for the social media features described in the previous chapter in relation to groupthink. Each feature being investigated is presented in a separate section. Individual conclusions
are presented in each section. Additionally, all results are combined later on in the
chapter where the social interaction design framework is presented. Furthermore,
some of the features that were investigated for the purposes of this dissertation
have been published in journals and conferences. In such cases, summaries and
highlights from the published papers are presented. One exception to this is the
article regarding dynamic voting interfaces and their effects on affecting decision
outcomes (Tsikerdekis, in press-a), which was not included in this chapter due
to size restrictions on the dissertation. However, results that came from that particular study are mentioned in the chapter 7 where the social interaction design
framework is presented.
4.2 Pro/con Lists
One of the features that were explored to assert a relation between design and
social behavior in terms of groupthink was the case of implementing pro/con
lists as a feature that can potentially reduce groupthink. Information that can
be organized in a more concise and probably more comprehensive format may
be more effective than traditional textual collaborative interfaces. I conducted a
comparative analysis that investigated argument production and comprehension
between pro/con lists and textual collaborative interfaces. This was a previously
unexplored software feature in HCI, especially in terms of its relation to groupthink. The study’s findings were published on a scientific journal (Tsikerdekis, in
press-c). A short version of the process is presented here along with additional
statistical analyses of the results which were not included in the published version. Finally, I present the conclusions and remarks based on the work conducted
in the original paper as well as in this dissertation.
42
4. R ESEARCHING THE I NDIVIDUAL F EATURES FOR G ROUPTHINK ...
4.2.1 Method and Initial Results
In my study, I proceeded with the analysis using real world data from two popular websites that used the two interfaces. The first website was Wikipedia that
uses a textual collaborative interface while the second was Debatepedia, a website
that facilitates debates with the use of pro/con lists. I obtained a random sample
of articles from both websites that collaborated on the same topic. Content analysis was used to code the text into quantitative data that was later used in the statistical analysis. From the qualitative analysis several additional conclusions were
extracted in relation to argument production. Finally, I used readability formulas
to develop quantitative data to assert the potential for comprehension between
the two interfaces.
The results showed that pro/con lists may not be more advanced than textual collaborative interfaces in terms of argument production however, statistically significant differences were found for the production of unique arguments
that favored pro/con lists. Additional comparisons that examined the readability
of the text between the two interfaces produced no statistically significant differences. Several other results coming from the qualitative portion of the study
showed that the original segmentation and categorization of an article had a significant effect on the development of articles for both interfaces.
The use of a variety of qualitative and quantitative methods helped triangulate results and validated some of the conclusions. This demonstrates the variety
of approaches that can be employed by researchers to assess certain hypotheses. Human-computer interaction as an interdisciplinary field can use the diverse variety of methods that exist without being constrained to a strict research
framework. In addition, the current approach demonstrates the wealth of information that is currently on the web and can be used for research. Finally, comparative content analysis is arguably a lot less costly as opposed to experimental
design methods, and examines real people in real situations as opposed to having “guinea pigs” in laboratory or simulated conditions. On the other hand, the
downside for the above methods is that real examples that match the conditions
that need to be investigated are difficult to find.
4.2.2 Analysis for Argument Production
A detailed account of how the data were obtained and processed can be found
in the original article Tsikerdekis (in press-c). In this dissertation, I present a
Bayesian analysis for the original hypotheses on the article that further helps assess the overall results of this study. The analyses that follow investigate two
aspects of pro/con lists over textual collaborative interfaces: production of total
arguments and production of unique arguments.
Descriptive statistics are shown on figure 4.1. Total arguments favored Debatepedia (M = 32.27, SD = 10.90) over Wikipedia (M = 25.47, SD = 16.88).
The result was similar with the unique arguments where Debatepedia (M =
59.84, SD = 16.76) showed higher measurements for unique arguments over
Wikipedia (M = 41.72, SD = 24.99). In the original article however, a statistical analysis using independent t-tests achieved significance only for the sec43
ond comparison. Therefore, a Bayesian analysis may more efficiently help assess
these results and establish the statistical likelihood of the two major hypotheses
being valid. Furthermore, while in null-hypothesis statistical testing, “p-values
are incapable of providing support for the null hypothesis” (Wagenmakers, Lee,
Lodewyckx, & Iverson, 2008, p. 200), Bayesian hypothesis testing does quantify
the probability that the null hypothesis is true (Kaptein & Robertson, 2012; Wagenmakers et al., 2008).
Figure 4.1: Descriptive results for pro/con lists and textual collaborative interfaces in terms of total argument production and unique arguments.
For the Bayesian statistical analysis, I used a model developed by Lee Michael
D and Eric-Jan Wagenmakers (2010) which I briefly present here. The model es44
tablishes that X and Y represent the observed data that follow a normal distribution with a shared variance σ 2 .
X ∼ N ormal(µX , σ 2 )
(4.1)
Y ∼ N ormal(µY , σ 2 )
(4.2)
The means for X and Y are defined as:
µX = µ + α/2
(4.3)
µY = µ − α/2
(4.4)
Therefore α becomes their difference in their means. The effect size δ is given by
the equation:
δ = α/σ
(4.5)
The priors for δ, µ and σ are established as:
δ ∼ Cauchy(0, 1)
(4.6)
µ ∼ Cauchy(0, 1)
(4.7)
σ ∼ Cauchy(0, 1)+
(4.8)
Based on the model, the hypotheses are the following:
H0 : δ = 0
(4.9)
H1 : δ ∼ Cauchy(0, 1)
(4.10)
H2 (δ < 0) : δ ∼ Cauchy(0, 1)−
(4.11)
H3 (δ > 0) : δ ∼ Cauchy(0, 1)+
(4.12)
The null hypothesis (H0 ) asserts that there is no effect while the alternative (H1 )
asserts that there is a difference between X and Y . Further, H2 supports the claim
that X has larger values than Y while H3 supports the opposite.
Using MCMC sampling I drew 100,000 samples from the posterior for the
effect size δ with a burnin for the first 10,000 samples and with three chains.
Next, I obtained the Bayes factor using the height of prior and posterior distributions for δ at δ = 0. The height of the prior was computed directly from
the Cauchy(0,1) distribution (using function dcauchy(0) in R) and the posterior
was estimated from the MCMC samples by applying a non-parametric estimator
(Stone, Hansen, Kooperberg, & Troung, 1997) which can be found in the polspline
package in R (using function dlogspline).
45
The Savage-Dickey ratio at δ = 0 yielded a Bayes factor of BF01 = 1.85, indicating that the null hypothesis (H0 ) is 1.85 more likely given the data than the
alternative hypothesis H1 . In terms of strength of evidence this Bayes factor is
classified as weak, meaning that while the null hypothesis still has to be accepted,
caution is advised (Jeffreys, 1998; Kass & Raftery, 1995; Wagenmakers et al.,
2008). The dots indicate the height of the two distributions at the point δ = 0, the
point of interest. A similar result was obtained for the order-restricted hypothesis H3 with a Bayes factor of BF03 = 1.05. This indicates that the null hypothesis
is 1.05 times more likely to occur than the hypothesis H3 . The result becomes extremely hard to interpret. In other words, given the data we can reject H3 but only
by an extremely small fraction. A more concrete result was obtained for H2 with
a Bayes factor of BF02 = 8.08. This is classified as substantial positive evidence
that the null hypothesis (H0 ) is 8.08 times more likely to happen than the H2 . Put
simply, having strong evidence against H2 means that Wikipedia did not produce
more arguments than Debatepedia. Figure 4.2 shows the graphs illustrating the
prior and posterior distributions obtained for all three hypotheses.
12
H0 : δ = 0
10
H0 : δ = 0
10
8
H2 : δ < 0
8
H3 : δ > 0
6
BF02 = 8.08
6
BF03 = 1.05
H1 : δ ≠ 0
3
BF01 = 1.85
Density
Density
12
2
●
4
Density
4
H0 : δ = 0
4
1
2
●
2
●
0
−3
−2
−1
0
δ
●
●
0
1
2
3
0
−3
−2
δ
−1
0
0
1
δ
2
3
Figure 4.2: The prior and posterior distributions of effect size δ, when comparing
total arguments between Debatepedia and Wikipedia. The top graph illustrates
the unrestricted t-test for H1 , the middle graph illustrates the order-restricted test
stating that Wikipedia produced more arguments than Debatepedia (H2 ), and the
bottom graph illustrates the t-test for the opposite order-restriction (H3 ). The dots
mark the height of the prior and posterior distributions at δ = 0.
Using the previous procedure, I evaluated the likelihood of the same unrestricted and order-restricted models but for the uniqueness of the arguments between Debatepedia and Wikipedia. The Bayes factor for H1 was BF10 = 2.28
indicating that given the data H1 (which corresponds to hypothesis H2 ) is 2.28
times more likely than H0 . In terms of strength of evidence this is far from strong
and it is different from the frequentist t-test which showed a medium effect size,
r = .404. Nevertheless, both indicate that there is a difference in the uniqueness
of the arguments between the two websites. The order-restricted testing yielded
Bayes factors BF02 = 11.73 and BF30 = 4.48. These seem to provide substantial
evidence that Debatepedia is more likely to produce more unique arguments than
Wikipedia. Figure 4.3 shows the plots for prior and posterior distributions for all
three hypotheses.
46
12
H0 : δ = 0
H0 : δ = 0
10
8
H2 : δ < 0
8
H3 : δ > 0
6
BF02 = 11.73
6
BF30 = 4.48
H1 : δ ≠ 0
3
BF10 = 2.28
Density
Density
12
10
2
H0 : δ = 0
●
Density
4
4
4
1
2
2
●
●
●
0
0
−3
−2
−1
0
δ
1
2
3
0
−3
−2
δ
−1
0
●
●
0
1
δ
2
3
uniqueness of arguments between Debatepedia and Wikipedia. The top graph
illustrates the unrestricted t-test (H1 ), the middle graph illustrates the orderrestricted test stating that Wikipedia produced more unique arguments than Debatepedia (H2 ), and the bottom graph illustrates the t-test for the opposite orderrestriction (H3 ). The dots mark the height of the prior and posterior distributions
at δ = 0.
4.2.3 Bayesian Analysis for Readability
A similar assessment was performed for the readability data that were presented
in the original paper. In the study, three readability measures were used in order
to increase internal validity: FRE, Flesch-Kincaid and SMOG. Descriptive results
for all three readability measures are depicted on figure 4.4. These show that Debatepedia had slightly superior results over Wikipedia. However, none of the
readability comparisons between the two interfaces achieved significance which
may be an indicator of readability being similar between the two interfaces. An
additional Bayesian analysis can further help assert this claim.
Following a similar approach as described in the previous section, I conducted
a Bayesian statistical analysis on the data. The Bayes factors for the unrestricted
and order-restricted hypotheses for the FRE formula were, BF01 = 1.69, BF02 =
8.29 and BF30 = 1.05. In general results agree with the t-test on the original paper but substantial evidence was found only for rejecting the H2 hypothesis. In
contrast, more substantial results were obtained for rejecting all three hypotheses
for the data based on the Flesch-Kincaid formula. Bayes factors were returned all
in favor of the H0 hypothesis, BF01 = 3.13, BF02 = 2.18, and BF03 = 5.98. Similarly, the Bayes factors for the data based on the SMOG formula also provided
substantial support for the H0 , BF01 = 3.17, BF02 = 2.26, and BF03 = 5.87. Figure
4.5 shows the plots for prior and posterior distributions of all these results.
4.2.4 Discussion
The results from the Bayesian statistical analysis are in accordance with the results
from the Null Hypothesis Statistical Testing (NHST) that was applied in the original paper. Results suggest that the two interfaces produce no different amount
of arguments but do vary in terms of the unique arguments. The Bayesian analysis however allows for order-restricted hypothesis that can further help assert
47
Figure 4.4: Descriptive results for pro/con lists and textual collaborative interfaces in terms of readability measures FRE, Flesch-Kincaid and SMOG.
more specific questions. Strong evidence was produced against the case of the
order-restricted hypothesis that stated that Wikipedia produced more arguments
than Debatepedia. Similarly, according to the order-restricted hypothesis for the
amount of unique arguments between the two interfaces, Debatepedia seems to
be favored by the analysis. Put simply, pro/con lists, while they do not seem to
produce more arguments than textual collaborative interfaces, they are definitely
producing more unique arguments.
The readability analyses also produced similar results with the original article.
The Bayesian analysis is supporting the null hypothesis that states that both interfaces produced similar levels of readability for all three readability measures.
However, that does not mean that articles produced similar ratings. As can be
seen in the results provided in the original article, there is a great variance within
the articles of Wikipedia and Debatepedia in terms of readability.
Aside from these results coming from the NHST and Bayesian analysis, there
were additional features that were discovered during the qualitative analysis.
One of the most prominent observations made during this analysis was the power
that the original segmentation of the article had on the further development of the
articles on both interfaces. In Debatepedia, pro/con lists were divided in terms
of categories and claims per categories. Similarly on Wikipedia an article is segmented in terms of categories as well. For example, both interfaces occasionally
skipped categories that were however included in one of them in their corre48
12
H0 : δ = 0
H0 : δ = 0
10
8
H2 : δ < 0
8
H3 : δ > 0
6
BF02 = 8.29
6
BF30 = 1.05
H1 : δ ≠ 0
3
BF01 = 1.69
Density
Density
12
10
2
●
Density
4
4
H0 : δ = 0
4
1
2
●
2
●
●
0
●
0
−3
−2
−1
0
δ
1
2
3
0
−3
−2
δ
−1
0
0
1
δ
2
3
(a) FRE formula
12
H0 : δ = 0
H0 : δ = 0
10
8
H2 : δ < 0
8
H3 : δ > 0
6
BF02 = 2.18
6
BF03 = 5.98
H1 : δ ≠ 0
3
BF01 = 3.13
Density
Density
12
10
2
Density
4
4
1
4
H0 : δ = 0
●
●
2
2
●
●
●
0
●
0
−3
−2
−1
0
δ
1
2
3
0
−3
−2
δ
−1
0
0
1
δ
2
3
(b) Flesh-Kincaid formula
12
H0 : δ = 0
H0 : δ = 0
10
8
H2 : δ < 0
8
H3 : δ > 0
6
BF02 = 2.26
6
BF03 = 5.87
H1 : δ ≠ 0
3
BF01 = 3.17
Density
Density
12
10
2
Density
4
4
4
H0 : δ = 0
●
●
1
2
2
●
●
●
0
0
−3
−2
−1
0
δ
1
2
3
●
0
−3
−2
δ
−1
0
0
1
δ
2
3
(c) SMOG formula
readability between Debatepedia and Wikipedia. Each sub-graph illustrates the
unrestricted t-tests (H1 ), the order-restricted test stating that Wikipedia scored
higher in readability than Debatepedia (H2 ), and the Bayesian t-test for the opposite order-restriction (H3 ). The dots mark the height of the prior and posterior
distributions at δ = 0.
sponding articles. This happened regardless that many articles were at least a
year old. Therefore, while many articles between the two interfaces have presented similar arguments, there were occasions where articles took different directions and produced their own unique arguments.
Pro/con lists had an advantage on producing unique arguments as it was
found in the statistical analysis, however they are also limited. According to the
original analysis, pro/con lists seem to have an upper limit for arguments per
each claim (Tsikerdekis, in press-c). In the study the limit was found to be between two and three arguments per claim and rarely arguments surpassed this
limit.
49
Recommendations
The recommendations based on the results produced by the Bayesian analysis,
are still in accordance with the recommendations made in the original article.
Given the evidence, pro/con lists are not superior for producing arguments that
could increase information flow within a group and help prevent groupthink.
However, there is strong evidence that they produce more unique arguments
which is a desired feature for successful decision-making. Additionally, both interfaces are limited in the original categorization of the article that could have
severe consequences in the production of arguments and is detrimental for preventing groupthink. Finally while pro/con lists seem to be favored by the results, they still have limitations when it comes to argument production. As such,
pro/con lists are recommended to be used in software that supports decisionmaking for preventing groupthink however, a combination of the two interfaces
investigated in this study may prove to be just as effective or potentially more
effective.
4.3 Alternatives Randomization in Voting Interfaces
While having sufficient alternatives is essential for a decision-making process, it
may not be enough to guarantee a positive outcome. After all, as indicated by
Henningsen et al. (2006), groupthink as a five-step model has implications all
throughout the decision-making process. One of the potential issues becomes apparent when considering that a list of alternatives is like any other list for the
individuals reviewing it. In other words, the design of a list is subject to limitations which may or may not contribute to groupthink. In this study, the order in
which lists represent items in collaborative social media was a feature examined
to determine if it could affect the choices of individuals.
In this part, I focus specifically in the formation of voting procedures and the
way alternatives are presented to individuals. This usually describes the final part
of the decision-making process where all the alternatives have been introduced,
analyzed and the final stage requires members to vote and reach a consensus. In
web design, this would usually be displayed as a multiple choice list of items.
It is tempting for designers to deliver the alternatives in terms of chronological
order or to allow for leaders to design the item order in the multiple choice list.
I introduce an alternative hypothesis which states that the randomization of the
list of alternatives for each individual is a better choice for preventing groupthink
behavior compared to a fixed-order list of items.
The choice of item presentation is far from trivial. Further, it is a realistic choice
for any voting system. Whether, one designs a corporate brainstorming group or
a collaborative project for educational purposes, this decision is an intricate part
of the software especially if there are implications on the final outcome and could
be the cause for groupthink.
For this hypothesis, I decided to follow a theoretical approach to establish
whether the randomization feature for lists may reduce groupthink. There are
two reasons for this choice. First, I can demonstrate the diverse approaches that
50
can be taken in order to evaluate software interfaces. Second, that parts of software evaluation for the new social interaction design framework can be based on
existing research findings. The result demonstrates a tremendous advantage for
HCI researchers that can produce cost-efficient results based on previous research
and therefore avoiding having to “reinvent the wheel.”
4.3.1 Literature on Response Order and Behavior
Multiple choice is an extremely intuitive design and extremely popular for educational testing, research, elections, and other potential social media uses. In its
simplest form, it requires from an individual to select one or more items from a
list. It is highly popular in the United States (Phelps, 2005) but in the rest of the
world as well. In fact, many social media today utilize this feature. Each option
in a multiple choice design is usually called an item, and one theory governing
the creation and evaluation of such items is the item response theory. This is a
paradigm for the design, analysis and scoring of such representations that are
measuring variables.
Van Der Linden and Hambleton (1997) pointed that in experimental designs
the issue of controlling for experimental errors in responses is essential when
designing tests. This is a similar issue in designing any kind of item response design such as multiple choice. They identified based on previous literature (Cox,
1958), that there are choices for coping with experimental error such as standardization or randomization. Standardization and randomization belong under the
category where the experimenter (or in our case designer) intervenes and his or
her decisions are expected to affect the outcome. Standardization is ideal for experimental designs (Field, 2009; Lewin, 2005) because it allows groups to be
compared based on the same categories. If variance exists due to the way the
items were presented, the error should be equal because the items were the same
for all groups. With randomization on the other hand, this assumption would not
hold due to the fact that different groups received questions in different orders.
Evidence for this can be traced in literature. Schwarz and Hippler (1991) identify the impact of response order in surveys. Specifically they state that the “presentation order may strongly influence the obtained results.” (Schwarz & Hippler, 1991, p. 51). The expected effects based on theoretical evidence are further
identified as primacy and recency effects. Primacy effects create higher endorsements for items early in the list while recency effects have higher endorsements
for items late in the list. Schwarz and Hippler (1991) also indicate that this effects are more likely to manifest when each response alternative represents different opinions for a topic rather when using ordered scales (Mingay & Greenwell,
1989) such as a Likert scale.
Primacy and recency effects in verbally-presented response categories are due
to memory bias (Krosnick & Alwin, 1987). In verbal speech it becomes obvious
that limited amount of time for reasoning and issues with retaining information
from a list, may affect individual choices on items. However, today the same idea
has been endorsed by literature for paper or digital surveys as well (Bishop &
Smith, 1997; Jobe & Mingay, 1991; Jobe, Tourangeau, & Smith, 1993; Krosnick & Presser, 2010). In addition, there seems to be a general agreement that
51
such effects in surveys are more likely to have primacy than recency effects, since
memory does not become an issue for the respondent in most survey designs
(Schwarz & Hippler, 1991). However, Schwarz and Hippler (1991) noted that
while memory may not be a factor affecting the decision-making process of an
individual, contrast effects may be a cause for primacy or recency effects.
Contrast effects can manifest in any list including multiple choice as long as
there is variance in the preferences of individuals for certain items. For example, when having an extremely likeable alternative for most of the group, the
order that items will be positioned in relation to the extremely likeable item may
affect the outcome. A moderately likeable item that will be presented after the
extremely likeable solution may seem less appealing as an alternative while if
presented before the popular item, it may lead people to viewing it more favorably.
4.3.2 Assessing the Appropriate Response Order
The overall likelihood of observing primacy effects and/or contrast effects becomes a serious issue in surveys as well as decision-making processes. The measurement error may lead to false decisions and conclusions for the groups involved. It is quite surprising therefore that these conditions have never been
linked with the groupthink model and in turn with the design of software for
social media, especially for collaborative projects.
Given my hypothesis and the information provided by literature, I compared
the two competing options for designing how lists are represented in a voting
interface; that is, standardization or randomization.
Standardization of Response Order
Standardization is ideal for controlling for measurement errors and seemingly obtaining a result that can be compared between individuals. In fact, one can argue
that the differences that one observes between individuals are only due to their
reasoning about the responses and not any other unknown factors. However, literature tends to disagree with this argument. Both empirical and theoretical evidence show that in survey designs choices may be made due to subconscious processes or as elsewhere defined, unsystematic variance (Field, 2009). Additionally,
primacy and contrast effects may sufficiently contribute to such differentiations
that in turn may affect the voting outcome.
One way for controlling such errors while using standardization for the response order, is to develop the order based on previous information and display
items appropriately. Prior information may come from results of a pilot study or
the insights of an individual. This prior information can then be used to assign
the order of the items in such way that will be unfavorable for the items that were
favorable before. The easiest way to achieve this is by inverting the original response order. After the final results are obtained, one can determine if there was
an effect due to the response order.
However, such method is not without issues. One of the problems is that the
overall selection of the response order is left to an individual which introduces
52
subjectivity. Even if that individual is an expert on the topic, has discussed and
understands the general preferences of the group, he or she is still required to
make a decision based also on intuition. Moreover, even if one detects presence
of primacy or contrast effects when comparing the outcome with the prior information, there is no ideal course of action. As an example, one can revise the
response order once again and repeat the voting process or merge the final results
with prior information in order to control for the error. Either can be costly and
does not guarantee a genuine outcome. As Schwarz and Hippler (1991) pointed
out, because “the relative power of the factors involved is difficult to evaluate a
priori, the emergence of response order effects is likely to remain a surprise in
many specific cases” (p. 56).
Randomization of Response Order
Randomization on the other hand, is likely to produce more variability in the
results but I argue that errors are more likely to cancel each other out between
individuals. Especially the case of primacy effects can be diminished on a group
level since even if participants have a tendency of selecting the early responses,
the effect will be diffused in the overall group.
However, contrast effects will still be present in a randomized response set. In
fact, one may argue that a standardized response set may at least provide some
control over contrast effects. But, the variance that will be created by the randomization is likely to end up diffusing the contrast effects on a group level. Moreover, while a standardized response set ideally can reduce the contrast effects, it
is extremely difficult to determine if that is the case a priori.
Randomization for multiple choice designs has been praised by McLeod, Zhang,
and Yu (2003) for educational purposes. As they suggest, randomization provides
advantages for preventing cheating and has no adverse effect on student performance. Coincidentally, this may provide support that contrast effects may be diluted within a sample of randomized sets. Multiple choice however, were found
to also have certain negative effects. One of them is that students may actually
try to guess that the more deceptive option may be the correct one (Roediger &
Marsh, 2005). While this may not be an issue for decision-making processes as
people have no need to guess for correct answers, it is still an issue worth considering. As advocated here, understanding the effects of the interface can only
improve the design of the software.
Finally, randomization can be applied along with standardization across randomized sets. This provides a way for evaluating any effects caused by randomization and evaluating issues with voting procedures. For example, one approach
may be to deliver randomized sets in a stratified manner just as it is applied in
stratified random sampling in quantitative research. However, as mentioned by
Hensher, Rose, and Greene (2005), research is limited in determining what would
be the ideal number of randomized sets for a sample. Perhaps, such procedures
can be applied in a similar way to conducting survey research. Lewin (2005) indicated minimum estimates for the subgroups within a sample to be between
twenty to fifty depending on the design. Using these estimated one can determine the number of randomized sets to be delivered to a group. For example, a
53
group of a hundred members may receive five different versions of response sets
based on randomization.
4.3.3 Conclusions
Based on this literature review, I attempted to make a connection between the
design of the order for alternatives and the chances of occurrence for groupthink.
The choice between a standardized or a randomized design may not be clear for
controlling contrast effects but randomized response sets have a clear advantage
over standardized response sets for controlling primacy effects.
When designing software that may affect voting processes it is important to
control for errors caused from these effects. Given the substantial advantage for
randomized sets, algorithms can be developed that could create such sets for the
list of alternatives. In addition, as I mention before, a designer can also have control over how many randomized sets will be created and even manufacture sets to
be delivered to the end users in a stratified manner. These sets allow for post-hoc
procedures to be applied so that software can indicate whether one of the randomized sets was plague by certain issues due to the response order. Algorithms
can be developed to not only test that but also raise red flags to the group after
the votes are in. Tests of statistical significance can be automatically applied for
such tests and give an estimate of the differentiation between randomized sets.
The end result is that designers can claim that the voting process provides
measures for preventing effects that could introduce bias in the voting results and
guarantee that they have taken steps to prevent groupthink. Even in cases where
members knowingly are willing to attempt to conform with a group’s suspected
favorable position are likely to face different response sets, which in turn will
produce different contrast effects. The result is that the software ensures that all
alternatives are given equal consideration in terms of the response order on a
group level.
4.4 Leaders in Social Media Affecting Opinions
One of the early hypotheses for groupthink described in my dissertation proposal
involved the issue of leaders being publicly identified in social media and their
power to affect voting procedures. Janis (1982) in his groupthink model assigned
significance on the role of the leader and the decision-making process. One of
the recommendations involved not allowing the leader to express their opinion
at least when initially are discussing a problem. However, I argued that if left
untested one can sufficiently claim that an identified leader by other members
cannot affect the voting outcome even at a later stage. It was therefore important to determine how much of an effect physical design features would have for
identifying leaders in a voting processes.
54
4.4.1 On Leaders and Groupthink
Studies on leadership show that leaders can affect behavior in several ways.
George (1995) has found that a leader’s positive mood can improve group performance. Similarly, Sy, Côté, and Saavedra (2005) investigated the effects of leaders’
mood on groups. They found that a leader’s mood can affect the group members’
mood, the affective tone between group members and the coordination within
a group. In turn, mood can also affect a leader’s effectiveness and attraction to
a leader for group members (Bono & Ilies, 2006). Further, leaders seem to have
even more power than just affecting mood and improving efficiency within corporate environments. Jones and Olken (2005) found positive evidence that leaders are detrimental to the economic development of nations and can affect policy
outcomes. As such, their role in group decision-making processes is crucial and
hold a great potential for affecting a group’s decision in terms of groupthink.
According to early literature, leadership is associated as a condition that could
lead to groupthink. Leana (1985), in a study utilized students which were divided
into groups that needed to resolve a hypothetical business problem. Some teams
had leaders assigned to them and were instructed to be directive while others had
leaders who encouraged member participation. The result was that groups with
directive leaders discussed fewer alternative solutions. Similar studies favored
these results (Z. Chen, Lawson, Gordon, & McIntosh, 1996; McCauley, 1989).
Ahlfinger and Esser (2001) in their study, tested the hypothesis that groups with
leaders that promoted their own preferred solutions will likely develop groupthink. The findings showed that groups with promotional leaders produced more
symptoms of groupthink, discussed fewer facts and reached to a consensus much
faster than groups with non-promotional leaders.
Promotional leadership is not only found in laboratory experiments but in
real life scenarios as well. In one of the case studies for groupthink involving the
space shuttle Challenger, Moorhead et al. (1991) indicated that top level managers
actively promoted their pro-launch opinions in the face of opposition. In addition
several managers ignored the warning about low temperatures and tried to push
for launch.
In social media geared towards collaboration, leaders exist in a variety of
ways. The terms administrator or moderator are highly popular in forums and
discussion boards. Other types of social media such as virtual game worlds have
been more creative with terms such as guild leaders or dungeon masters. No matter the title, these individuals are in key positions within an organization and as
leaders can affect decisions. When social media software is designed one has to
evaluate if these leaders will affect group decision-making in a negative manner
and be a cause for groupthink.
It is challenging however, to determine aspects through which one can balance the negative effects of leaders and promote the positive effects. Rosenthal
and Okie (2005) suggested that anonymous forums with no administrators overseeing the messages may be more effective and increase participation for students
suffering from depression. Disabling leaders may be possible in certain cases,
however, this may not be possible in other cases where the presence of a leader
is necessary for increasing positive attitude and performance in a group (Bono &
55
Ilies, 2006; George, 1995). A better approach however, may be to avoid making
leaders appear so prominent especially under crucial processes such as the final
stages of a decision-making process (e.g., voting).
There is a great variety of voting mechanisms which definitely have an effect
on the outcome of a voting process (Brams, 1991; Dummett, 1984; Gavish &
Gerdes, 1997; Z. Li, 2003). Dynamic open voting is an attractive idea for many
groups that wish to able to see individuals and their choices as the voting process progresses. In many cases, leaders are distinguished from users using titles
in front of their names or even color variations; a choice extensively used by software designers in forums and similar discussion boards. Given the power that
leaders hold over a group and the potential for groupthink, this public identification of leaders and their distinction from the rest of the group can potentially
affect voting outcomes. This helped formulate the main hypothesis for this part
of the study:
H1 : Visually identified leaders can affect a voting process where individuals
vote publicly.
4.4.2 Method
In order to evaluate this hypothesis, I developed a set of questions and included
them in a larger survey that I had administered to evaluate other hypotheses that
were part of the overall dissertation work. Hereby, the large survey is denoted as
GR1 in other sections of the dissertation. The type of survey was a randomized
controlled trial “in which participants are allocated truly randomly to an experimental group and a control group” (Lewin, 2005, p. 221) which strengthens the
claims for internal validity.
The survey involved a scenario about global warming where people had to
vote for one of four potential solutions. The simulated voting procedure involved
sixteen individuals which had already voted for their solutions and the result
was a draw. The respondent was called upon to vote and essentially break the tie
with his or her decision. Individuals were identified publicly by name, while the
administrator was distinguished from the rest of the individuals with a different
color as well as a title in front of the name. Two versions were created where in
the first the leader voted for the first solution while in the second the vote was
placed on the third solution. Figures 4.6 and 4.7 depict these two versions.
Additionally, a further open-ended question was included in the survey in order to help understand the reasons behind the respondents’ choices and in order
to help triangulate and complete the results (Bryman, 2006) (“Why did you select
this solution? Please elaborate.”).
The survey was administered through Facebook. Invitations for the survey
were circulated in several academic and non-academic groups and fan pages on
the social networking website. Additionally, a snowball sampling method was
introduced by allowing users to invite their friends in order to reach farther
users within the network. The survey was originally developed in English but
also translated to Greek and Spanish as well. The language was automatically
adapted based on the respondents’ system settings and Facebook profile settings.
The survey took place during the period of February 15 to March 19, 2012.
56
Figure 4.6: The first version of the voting process with the leader supporting the
first solution.
4.4.3 Results
In total, 123 people visited the application during the survey and 89 ( 72.4%) went
through the whole survey. Most of the non-respondents quit the survey while on
the first page. The sample of 89 respondents was used for the analyses. Their demographic information reveals that the sample was well balanced in terms of sex
but in terms of age it was relatively young. The self-reported geographical information reveals that most participants came from Europe. Figure 4.8 summarizes
the demographic data.
Looking at the results between treatments the differences in the votes did not
seem to be large. Figure 4.9 shows the counts for each category per treatment. In
both cases the voting outcome was the same. However, there was a small difference in the preference of the third solution. It seems, that by having the leader
being identified and voting for the third solution did affect people’s choice up to
a degree. However, was this a significant effect?
A chi-square test for the contingency table 4.1 showed no statistical significance, χ2 (3) = 4.336, p > 0.05, V = .221. However, two of the assumptions for the
chi-square test were violated. Seventy five percent of cells had an expected count
less than five and the minimum expect count was less than one. This severely impacts the accuracy of the result and therefore a Fisher’s exact test was performed
in order to obtain the exact significance value. The result was also not significant,
p > 0.05.
I used Bayesian statistics to determine the probability of the null hypothesis.
I used a procedure developed by Albert (2009) which can be accessed from R
using the LearnBayes package. The ctable function returned a Bayes Factor in favor of the null hypothesis, BF01 = 11.33. This indicates that the null hypothesis
57
Figure 4.7: The second version of the voting process with the leader supporting
the third solution.
Treatment / Response
Leader supporting
the first position
Leader supporting
the second position
Reduce... A man... No need... Wait...
Total
42
4
1
1
48
33
3
5
0
41
Table 4.1: Contingency table for each treatment and votes given by the respondents
of independence was 11.33 times more likely over the alternative hypothesis of
dependence. This is considered strong evidence (Jeffreys, 1998; Kass & Raftery,
1995; Wagenmakers et al., 2008). Additionally, using the bfindep function, I evaluated the case of having a hypothesis “close to independence” allowing for some
variability due to random factors. This hypothesis was tested in terms of its likelihood over the hypothesis of independence for K precision parameters between
2 and 7. Having 100,000 simulations, the highest Bayes factor that was obtained
was BF10 = 1.16. The result indicated weak evidence in favor of the hypothesis
“close to independence” against the null hypothesis of independence.
Most of the respondents (86,5%) have also elaborated on their choice on the
open-ended question while a small percentage (13,5%) left the question unanswered. Qualitative results seem to agree with the statistical analyses. Many respondents showed that their decision was based on their emotional reasoning.
For example one respondent under the first treatment stated that he or she chose
58
Figure 4.8: Demographic information for the leadership survey.
the first solution “Because i [sic] believe is the most applicable.” Similarly, a respondent elaborated further by saying that “It’s the most active solution. Something we can do right now. It will also efficiently reduce costs.” Another individual preferred the first choice while however under the second treatment. He or
she said that the choice was made “because it is what i [sic] already believe is
an imperative for people.” Furthermore, an individual with a rationalized choice
under the first treatment, voted for the second solution while stating that it is a
better option “cause [sic] are day to day habits would not change.”
In general, almost all the answers for the qualitative question followed a similar pattern and at no place there was an indication that the position of the leader
affected their decision. One answer however was particularly different from the
rest. An individual under the second treatment that selected the third solution
(the leader’s choice) stated that he or she was “unsure.” While a single word
59
Figure 4.9: Simple graph for the votes for each solution per treatment.
barely classifies as informative it is still worth mentioning given that this was the
only individual of those that had answered that they could not elaborate on his
or her choice. Under the second treatment, there were additionally two respondents that preferred the third solution however did not answer the qualitative
question. It may be the case that these additional individuals may have felt in
a similar manner and were unsure about committing to one of the choices provided. This may indicate that in general leaders may not affect the voting process
but under circumstances where individuals hold a degree of uncertainty over a
topic an effect may be found.
4.4.4 Discussion
The results have been promising for social media software and especially online collaborative communities. All statistical analyses pointed to the same direction. Leaders while identified in the voting process do not have enough power to
change the voting outcome. This result is promising given the popularity of having administrators and other members holding leader positions being publicly
identified in many online communities. In addition, it also shows a great difference with the general consensus over promotional leadership and its relation
to groupthink. According to the results, by having administrators publicly show
their preference while being publicly identified would not necessarily contribute
to groupthink.
However, some minor differences were observed especially for the third so60
lution. The difference between treatments indicated that some users did follow
the leader’s position. This is a small difference and did not change the outcome;
which could have otherwise been a cause for concern in terms of groupthink.
However, it is still worth considering. This is especially important given the qualitative findings. While one can design a voting interface with leaders being identified and expect that this would not affect the voting outcome, it has the potential
to distort the results and affect individual votes.
Recommendations
Given the evidence, it is safe for software designers and developers to utilize in
social media open voting processes where leaders are publicly identified and distinguished by the rest of the users. Voting outcomes are likely to not be affected
and therefore the potential for groupthink is small at the final voting stage of
a decision-making process. However, this design should be implemented with
caution. While the overall outcome is not affected by the leaders’ positions, individual votes may be affected under conditions where individuals are uncertain
or confused about a topic. Additional information that can be provided about a
topic and each solution may help individuals better evaluate their decisions.
Limitations
This study has similar drawbacks with any laboratory experiments. Occasionally
laboratory settings cannot sufficiently reproduce real world situations. For example, in groups usually people have affections and preferences for other people
within the group. If the administrator is popular, he or she may have more power
over the voting process. In addition, members may not wish to make their votes
public in favor of one solution that goes against the crowd and it is an issue that
should also be addressed.
4.5 Anonymity States and their Effect on Non-Conformity and
Groupthink
One of the features investigated by my dissertation work was the effect of the
different anonymity states that can be found on social media today in relation to
groupthink. These anonymity states were: pseudonymity and individuals being
completely anonymous. According to literature, anonymity has long being found
to be a condition that may reduce compliance pressures and is an antecedent for
groupthink (Ahlfinger & Esser, 2001; McCauley, 1989; Rovio et al., 2009). However, anonymity states were never studied in regard on their effect on groupthink.
My aim was to establish if there is an effect in terms of ensuring alternative solutions for situations that require problem-solving. An adequate amount of alternatives that should be explored is an important measure for preventing groupthink
as stated by Janis (1972, 1982). The outcome of this investigation led to a published article on a scientific journal (Tsikerdekis, in press-b). In this section, I
provide the method and major findings of this research.
61
4.5.1 Method
For the purposes of this study, I decided to use a survey which had the form of
a randomized controlled trial (Lewin, 2005). I created three different vignettes
that featured scenarios of a group making a decision about a particular problem. The majority of the group had already decided on a particular position and
the respondent was required to make a choice on whether he or she would propose an alternative solution for discussion or follow the group’s opinion. Prior to
that, their opinion was obtained for each problem presented on the vignettes and
the software always simulated the group’s opinion to be different from the one
selected by a respondent. Several additional questions were retrieved from individuals regarding how important they thought the resolution of each problem
was, along with their perception of anonymity under each vignette.
For each of the three different vignettes, three versions were produced that
were different only on the presentation of the anonymity state that the group
was collaborating under. These were pseudonymity, complete anonymity and an
additional state under which participants used their real names. Each respondent
received the three vignettes in a random order. Additionally, each was randomly
assigned one of the versions representing a different anonymity state.
The research was conducted in collaboration with the Wikimedia Foundation
Research Committee which reviewed and approved it. A random sample of respondents was invited through the English version of Wikipedia. A detailed account of the survey design, process and overall survey information can be found
in the original article (Tsikerdekis, in press-b).
4.5.2 Results
Several hypothesis were tested in the study (Tsikerdekis, in press-b)
•
H1 : With higher levels of anonymity, the likelihood of not conforming increases.
•
H2 : With higher levels of anonymity, the perception of anonymity increases.
•
H3 : As the perception of anonymity increases, the likelihood of not conforming increases.
•
H4 : There will be a relationship between a given scenario and the likelihood
of not conforming.
•
H5 : There will be a relationship between a given scenario and the perception
of anonymity.
•
H6 : As the level of importance assigned to a problem increases, the likelihood of not conforming increases.
•
H7 : As the level of importance assigned to a problem increases, the perception of anonymity decreases.
A research model was also formed based on these hypotheses which is depicted in figure 4.10 along with the results based on the statistical analyses.
62
Figure 4.10: Research model for research on anonymity states and groupthink
with results. Note: * p < .05, ** p < .01, *** p < .001
Hypotheses H1 , H2 , H3 , H4 , H7 were supported while H6 was conditionally
supported and H5 was not supported. Anonymity states had an effect on conformity and as such on groupthink. The more anonymous an individual was,
the more likely he or she would not conform. Figure 4.11 depicts counts for the
likelihood of not conforming for each one of the anonymity states. A statistically
significant effect was found, however, it was small, rs = .102, p < .05 (1-tailed).
Similarly, using the scale of perception of anonymity, a small effect was found
with the likelihood of not conforming, rs = .101, p < .05 (1-tailed). Figure 4.12
depicts a line chart between the two variables. In general, the medians indicate
that up to the point where the perception of anonymity is reported higher, the
likelihood of not conforming is one value lower.
In order to establish whether the perception of anonymity was indeed dependent on the anonymity states the two variables were analyzed using Spearman’s
correlation. The result indicated that there was a statistically significant medium
effect size, rs = .466, p < .001 (1-tailed). Figure 4.13 clearly displays this association between the two variables.
Additionally, scenarios and level of importance had an effect either directly or
indirectly through other variables. Scenarios also seemed to have had inhibiting
effects for correlations. For example, the scenarios involving a noisy neighbor and
a misbehaving employee showed conformity being dependent on anonymity,
rs = .178, p < .05 (1-tailed) and rs = .160, p < .05 (1-tailed) respectively. However,
the scenario involving an employee’s unpaid overtime did not show a statistically significant association between these variables. Instead, it showed that the
respondent’s likelihood of not conforming depended greatly on the self-reported
level of importance assigned to the problem, rs = .222, p < .01 (1-tailed). Figure
4.14 depicts the likelihood of not conforming for all three scenarios in respect to
the self-reported level of importance.
Qualitative evidence while agreeing with the quantitative data, in many cases
showed individuals that stated anonymity was a factor for their decisions. Many
respondents made clear statements that they would not propose their alternative
63
Figure 4.11: Likelihood of not conforming for each anonymity state.
solutions under circumstances where real names were used and have done so.
Many were afraid of the consequences especially in some of the scenarios where
the penalties might have been more severe for going against the rest of the group.
Additionally, among the findings discovered from the qualitative data was the
issue of trust between respondents and social media services. There were statements made by the participants that indicated that they were skeptical about the
anonymity provided by a service. As such, this was identified as a probable factor
that may affect perceptions of anonymity and in turn the likelihood of conformity.
Recommendations
Recommendations can be found in detail in the original article (Tsikerdekis, in
press-b). It is recommended based on the evidence brought forth by this study
that software designers and developers should promote the use of anonymity
especially in social media aimed for collaborative activities. While the effect is
small, it is still significant so that all potential solutions should be examined before dismissed. Between pseudonymity and complete anonymity, the latter holds
the greatest potential for reducing groupthink. An alternative course of action
could also be the use of real names or pseudonyms with an additional option for
anonymous information exchange. Additionally, designers should find ways to
communicate with their users and alleviate any issues of trust that may exist for
anonymity.
64
Figure 4.12: Likelihood of not conforming and perception of anonymity.
Additionally, several other factors should be given enough consideration. For
example, the level of importance that participants assign for a resolution of a
problem, has been found to affect the likelihood of conforming. When designing collaborative software, it may be beneficial to assess a self-reported level
of importance from the collaborators in order to establish prior to a conference
what would the likelihood for conformity and groupthink be. Finally, developers should try to better communicate how anonymity is being implemented and
generally to achieve higher levels of trust for the anonymity provided to the final
users.
65
Figure 4.13: Line chart displaying the gradually ascend of the perception of
anonymity as progressing to higher anonymity states.
66
Figure 4.14: Likelihood of not conforming for all three scenarios based on the selfreported level of importance. The second graph is significantly being impacted by
this dependency.
67
C HAPTER 5
O NLINE A GGRESSION T HEORY AND A PPLICATION
5.1 Introduction
In this chapter, I provide a literature review on aggression and present features
that can potentially inhibit aggression. The purpose is to provide a clear view for
the motivation and the importance into researching online aggression as well as
what should be considered as important when designing social media. Specifically, the aim of this chapter is to demonstrate that one can engineer aggression
through software features. As with the case of groupthink, having an understanding of sociological and psychological literature is important for developing procedures that establish how changes in the software may alter social behaviors.
5.2 On Aggression in General
Aggression is a phenomenon that can be seen in all aspects life. On the other
hand, online aggression as a phenomenon is relatively new compared to the span
of human civilization. However, knowledge of the two is necessary in order to
understand the differences between them as well as for identifying design opportunities. Many studies in aggression take a comparative approach between online
and offline aggression in order to examine such differences (Dooley, Pyżalski, &
Cross, 2009; Hinduja & Patchin, 2010; Schoffstall & Cohen, 2011; Yen, Yen, Wu,
Huang, & Ko, 2011).
5.2.1 Definition
Defining aggression can be difficult especially when attempting to identify various classifications of aggression. As Baron and Richardson (2004) pointed out,
Berkowitz (1993a) described the linguistic issues for defining aggression.
When people describe someone as being aggressive, they might be
saying that he frequently attempts to hurt others, or that he is often
unfriendly, or in a quite different sense, that he is typically very forceful and tries to get his own way in his dealings with others, or maybe
that he assertively stands up for his beliefs, or perhaps that he usually
attempts to solve problems facing him. (p. 4)
68
5. O NLINE A GGRESSION T HEORY AND A PPLICATION
The main issue is whether aggression is defined by the intention to do harm to
others or when harm is actually inflicted on others. Baron and Richardson (2004)
provided a working definition that includes both intention to harm and harming
someone:
Aggression is any form of behavior directed toward the goal of harming or injuring another living being who is motivated to avoid such
treatment. (p. 7)
This definition identifies aggression as a behavior, not as an emotion or motive. In addition, it also requires a clear intention by the assailant while the victim
should always try to avoid harm. Additionally, within this definition one can also
identify violence, which is aggression that has extreme harm as its goal (Anderson & Bushman, 2002). While this definition is endorsed for the purposes of this
study, it should also be acknowledged that in an online setting damage can be
inflicted unintentionally due to software design. However, classifying accidental
damage inflicted on someone as aggression can be problematic.
5.2.2 Types and Categories of Aggression
Aggression can be identified as physical or verbal (Baron & Richardson, 2004)
but one can also observe indirect aggression (Bjtirkqvist, 1994). Verbal aggression
is often described as psychological aggression which is usually perceived to precede physical aggression (Capaldi & Crosby, 1997). This is a particularly interesting type of aggression since it is essentially an online type of aggression as well.
Follingstad, Rutledge, Berg, Hause, and Polek (1990) identified several types of
psychological abuse such as threats of abuse, ridicule, jealousy, threats to change
relationship, restriction and damage to property. Additionally, Kessler (1990) included insults, swearing, spiteful behavior, and abrupt withdrawal from interaction. Arguably many of these cases of psychological aggression can be found in
social media today. While one may argue there are not as damaging as physical
aggression, real world studies show otherwise (Follingstad et al., 1990). Hence, it
is important to explore such issues and find ways that aggression can be reduced
in social media.
Furthermore, aggression can be divided into two major categories; hostile aggression and instrumental aggression (Anderson & Bushman, 2002). Hostile aggression is also called affective, impulsive, reactive, or retaliatory. It can be distinguished by instrumental aggression where there is a need for premeditative
means of obtaining some goal aside from harming someone (Berkowitz, 1993b;
Geen, 2001). Hostile aggression is also found to be associated with people that
have a lower IQ score (Vitiello, Behar, Hunt, Stoff, & Ricciuti, 1990). Ramírez
and Andreu (2006) suggested based on their research that personality measures
can be used to clearly differentiate aggressive subjects. Arguably, the same personality measures can be used for a population in social media to identify which
category of aggression is more likely to appear given the types of users that exist
in the online community.
69
On the other hand, while this categorical distinction between hostile and instrumental aggression is supported by some researchers (Mcellistrem, 2004; Weinshenker, 2002), others insist that such a dichotomy fails to consider aggressive
acts with multiple motives and advocate against it (B. J. Bushman & Anderson,
2001). While in this dissertation, I will not elaborate more on this debate, I do consider both perspectives as likely to be better explanatory and descriptive models
depending on the situational context. As mentioned previously in chapter 3, I
further advocate that HCI research involved with psychological and sociological
behaviors should always maintain a broad view with the concepts involved and
should not try to act as the agent for resolving long-term debates from other disciplines. Instead, it should include using the knowledge coming from these fields
in order to improve over the current design of software and maintain a practical
perspective for the issues studied.
5.2.3 The Nature of Aggression
Aggression based on the definition presented in this chapter is not restricted in
the human species. For example, as indicated by Lorenz (2002), aggression is a
life-preserving mechanism for insects. It is not surprising that such mechanism
can be found in other species including ours. However, the origin of aggression
is debated. Researchers seem to disagree on whether the origin for human aggression has a biological or evolutionary basis (Somit, 1990).
Proponents of the evolutionary basis have indicated how some of these behaviors may have evolved (Buss & Duntley, 2006; McCall & Shields, 2008) however, there is difficulty in distinguishing which were truly selected and which
have been a by-product of the process, as it is in the case with collective violence
(Durrant, 2011). In addition, selection seems to have been part of the differences
between sexes for aggression. The differences between male and female are consistent with sexually-selected behavioral differences (Archer, 2009) with a support that women may have evolved to be less physically dangerous and more
covert (A. Campbell, 1999, 2005).
On the opposite side of the seesaw, advocates of the physiological basis identify several root causes. Studies in animals have found that electrical stimulation
of the hypothalamus causes aggressive behavior (Kruk et al., 1983). Others psychological responses can be found in chimpanzees which as continuous breeders
show significantly raised testosterone levels and aggression between males when
females are present (Muller & Wrangham, 2004).
No matter the case, it becomes evident that aggression is an intricate part of
our existence and reality.
5.3 Research on the Causes for Human Aggression
Research on aggression was conducted in a variety of contexts such as in culture
(Fujihara, Kohyama, Andreu, & Ramirez, 1999; Harman, Klopf, & Ishii, 1990;
Lomas, 2009), aggression in media (Aronson, Wilson, Akert, & Akert, 2009;
Freedman, 2002), in children (Bongers, Koot, Van Der Ende, & Verhulst, 2008;
70
Network, 2004; Tremblay, 2000) and even between sexes (Björkqvist, Österman,
& Lagerspetz, 1994; Card, Stucky, Sawalani, & Little, 2008; Hines & Saudino,
2003). There is an increasing interest over the recent years to understand why
humans occasionally act aggressively (Anderson & Bushman, 2002). Anderson
and Bushman (2002) in their review identified a number of causes for the increased aggression such as accessibility to guns (O’Donnell, 1995), global warming (Anderson, Bushman, & Groom, 1997), violence against children in schools
and homes (Hyman, 1995), and the widespread exposure to media (B. Bushman
& Huesmann, 2001). Explanatory theories were at the core of all of these studies
that aimed to understand the nature of human aggression.
5.3.1 Theories on Aggression
Some of the most prevalent theories were presented by Anderson and Bushman
(2002). These are described here as they are an important part for identifying
design features that may have an effect on social interaction related to aggression.
Neoassociation Theory
The theory built upon the ideas developed by Berkowitz (1989, 1990, 1993a) where
aversive events such as frustrations, provocations, loud noises and similar events
produce a negative effect that automatically stimulates various thoughts and
memories, expressive motor reactions and physiological responses associated with
fight or flight tendencies (Anderson & Bushman, 2002). The theory was built
upon the frustration-aggression theory developed by Dollard, Doob, Miller, Mowrer,
and Sears (1939) which at its simplest form states that the more frustrated an individual is, the more aggressive he or she is likely to become.
Social Learning Theory
A different approach of explaining the cause of aggression is taken by the social
learning theory. In this case, aggression becomes a social behavior that is obtained
through observing others and in effect learning how to act in situations (Bandura,
2001a; Mischel, 1973, 1999; Mischel & Shoda, 1995). This theory is ideal for
describing differences in aggressive behaviors between cultures. In addition as
indicated by Anderson and Bushman (2002) the theory is ideal for explaining
acquisition of aggressive behaviors as well as instrumental aggression.
Script Theory
Script theory resembles some similarities with the social learning theory. It states
that as individuals observe violence in mass media they learn aggressive scripts.
This was first proposed by Huesmann (1986, 1998) to explain behaviors of children. Once a script is learned it can be used at any time to guide a behavior. Further research into script theory by Abelson (1981); Anderson (1983); Anderson
and Godfrey (1987); Marsh, Hicks, and Bink (1998); Schank and Abelson (1977)
71
indicated the power that these scripts have in dictating social behaviors. As Anderson and Bushman (2002) puts it, “a child who has witnessed several thousand
instances of using a gun to settle a dispute on television is likely to have a very
accessible script that has generalized across many situations. In other words, the
script becomes chronically accessible” (p. 32).
Excitation Transfer Theory
Excitation transfer theory as developed by Zillmann (1983), describes how physiological arousal can dissipate slowly and be transferred between several arousal
events. If the event following a previous event brings forth anger then the additional residual arousal can make the person angrier. Anderson and Bushman
(2002) also has indicated that anger can stay for long periods of time with the
person even after the arousal has dissipated.
Social Interaction Theory
Social interaction theory seeks to interpret coercive actions as part of social influence (Tedeschi & Felson, 1994). The choice to act in an aggressive manner falls
under an individual’s judgment based on his or her cost-benefit analysis of a situation. This theory helps explain aggressive acts that are motivated by higher level
goals (Anderson & Bushman, 2002).
5.3.2 General Aggression Model
As described by Anderson and Bushman (2002); DeWall, Anderson, and Bushman (2011); Parrott (2008) the General Aggression Model (GAM) becomes a unifying model for all the theories on aggression that were presented previously.
As an effect of this integration of all theories into one model, the general aggression model as a social-cognitive model incorporates biological, personality development, social processes, basic cognitive processes, short-term and long-term
processes, and decision processes into understanding aggression (DeWall et al.,
2011). Several articles describe in detail the inner working of the model (Anderson & Bushman, 2002; DeWall & Anderson, 2011). The most important parts
and structure of the model are presented here.
The model contains three critical stages in understanding a single episodic
cycle of aggression: (1) person and situation inputs, (2) present internal states
(i.e., cognition, arousal, affect, including brain activity), and (3) outcomes of appraisal and decision-making processes (DeWall et al., 2011). Additionally, Anderson, Buckley, and Carnagey (2008); DeWall and Anderson (2011) indicated
that a feedback loop can influence future cycles of aggression and start a violence
escalation cycle.
In the input stage, variables based on person and situational factors play
a critical role in the overall process. Beliefs, traits, attitudes, sex, values, longterm goals, scripts are all factors related to an individual (Anderson & Bushman,
2002). Similarly, situational factors include aggressive cues, provocation, frustration, pain and discomfort, drugs, and incentives.
72
In turn, these input variables develop the present internal state and through
this process the final outcome behavior can be affected. These internal states are:
(1) cognition which includes hostile thoughts(i.e., chronically accessible prime aggressive thoughts) and scripts, (2) affect which includes mood and emotion and
expressive motor responses (i.e., unexpected pain which produces angry expressions), and (3) arousal which can influence in several ways such as, (a) if coming
from an irrelevant source strengthens the action tendency (especially aggressive
tendencies), (b) if elicited by an irrelevant source and is mislabeled as anger in
situations involving provocation, (c) if high and low levels of arousal can be aversive states and as such stimulate aggression (Anderson & Bushman, 2002). All of
the above states are interconnected; effectively one causing the other, and vice
versa.
The final stage describes appraisal and the decision-making process. Appraisal
can be further analyzed into immediate appraisal, which is considered to involve
relatively automatic information, processing and reappraisal which is considered
to involve more heavily controlled information processing (Anderson & Bushman, 2002). In the cases of an immediate appraisal if the person has sufficient
resources (i.e., time, cognitive capacity) and the immediate appraisal outcome is
both important and unsatisfying, then an individual will seek a more cognitively
demanding set of reappraisals. Otherwise, an individual exhibits impulsive actions which can be aggressive or non-aggressive depending on the content of the
immediate appraisal.
Due to its unifying nature, the model has been successful in explaining several aggressive behaviors. DeWall et al. (2011) indicated how the model can be
applied to understand intimate partner violence, intergroup violence, global climate change effects on violence, and suicide. Similarly, Parrott (2008) has used
this model to develop a theoretical framework for understanding antigay aggression.
However, the model is not without its critics. Ferguson and Dyck (2012) reported that the model may not be satisfactory due to inconclusive evidence. They
further based their claims on the fact that social science is based mainly on NHST
which provides no road to falsification and that most researchers have a tendency
to attribute a non significant result in the small effect that is attempted to be detected and not considering that there may be such a small effect that could become
trivial. Hence, misinterpretations of the effect sizes are considered an additional
cause for supporting the general aggression model according to Ferguson (2009);
Ferguson and Dyck (2012).
5.3.3 Additional Related Phenomena
There are several phenomena in literature that could act also as explanatory models for certain cases of aggression. Some are listed below as they were described
by Anderson and Bushman (2002).
73
Opportunity
Opportunity or the social situation may restrict or allow for aggression (A. P. Goldstein, 1994). Schools have such restrictive rules where aggression is discouraged
through strong social norms and the presence of witnesses. On the other hand,
bars in which there is a presence of alcohol, aggressive cues, relative anonymity
and competitiveness between males for females are more likely to develop more
aggression for individuals; a model that fits well with the general aggression
model as well (Anderson & Bushman, 2002).
Overriding Inhibitions
Research on overriding inhibitions identified certain patterns in which individuals that would not normally act aggressively, act in such a manner (Bandura,
2001a; Bandura, Barbaranelli, Caprara, & Pastorelli, 1996; Keltner & Robinson,
1996; Staub, 1989). The two important mechanisms for this are moral justification in which individuals rationalize a decision (i.e., when an individual sees
that the end for a common good justify the means) and victim dehumanization
in which moral standards are not applicable (i.e., war propaganda) (Anderson
& Bushman, 2002). Within this phenomenon there has also been research about
online disinhibition which becomes relevant for this study as well (Suler, 2004a).
Shared Motivations
Social groups have several common needs that may feel inclined to protect when
a threat to those needs appears (Anderson & Bushman, 2002). Aggression is often the dominant response to such threats which as suggested by Anderson and
Bushman (2002) positions of power are abused as well as the probable evolutionary tendency to emit aggressive behaviors when faced with pain, whether it be
physical or psychological.
Role of Anger
Anger also seems to act as a factor for aggression. This view has been challenged
by Berkowitz (1993a); Geen (2001) however Anderson and Bushman (2002) identified five causal roles that anger plays for aggression. These are: (1) anger reduces
inhibitions against aggressing by either providing a justification for aggressive
retaliation or by having anger interfering with higher-level cognitive processes,
(2) anger allows for maintaining an aggressive intention over time, (3) anger is
used as an information cue, (4) anger primes against thoughts, scripts etc., and
(5) anger energizes behavior by increasing arousal levels.
5.4 Research on Online Aggression
It is of no surprise that people may become aggressive online. While most of the
online aggression is through psychological means where people are not physi74
cally harmed, residual aggressive tendencies may be expressed in physical reality
as well.
One recent example was the case where an adult male shot an adult female on
a college campus and then committed suicide in 2009 (Taylor-Bonds, 2009). Both
individuals had accounts on YouTube while the male had publicly expressed aggressive intentions in his videos for people opposing his religious beliefs (YouTube,
2010). In one of them, one can clearly see an overriding of inhibitions by victim
dehumanization. He stated “[they] are not humans, they are monkeys, they are
animals [...] worms and dogs” (YouTube, 2010). While statements such as these
are not a guarantee for the violent acts that followed, they should be a cause for
concern and raise red flags.
Of course, not all online incidents lead to physical manifestations of violence.
Some remain online however the psychological damage on the victims can be life
damaging as well. Cyberbullying, a type of bullying “in which the aggression
occurs through modern technological devices, and specifically mobile phones or
the internet” (Slonje & Smith, 2008, p. 147), is a popular method for psychological
aggression online. One of the most popular cases, is the one involving a 15-yearold Quebec boy that had one of his videos leaked on the internet by his classmates
and has been edited and submitted in various forms by individuals from all over
the world ever since (Brady & Conn, 2006; M. A. Campbell, 2005). According
to media reports, the student’s family filed a lawsuit against the families of the
students responsible for releasing the video on the Internet. In the lawsuit they
stated that the student “had to endure, and still endures today, harassment and
derision from his high-school mates and the public at large” (Popkin, 2007).
The above incidents while arguably rare, demonstrate that just as in real life,
online life in social media has at times many unforeseeable dangers. This probably makes them more dangerous and their development and management becomes more problematic. The blowback from these incidents in online communities can be devastating with lawsuits waging against the companies as well as
the individual perpetrators. These situations can become a PR’s nightmare. It is
therefore justified that many researchers try to understand such online incidents
and why they occur.
5.4.1 Media Violence Research
Research in media violence can expand in various types of media such as novels,
comic books, television but it also applies in social media found today, whether
online collaborative communities, social network sites or social game worlds.
While some believe that there is a link between media and violence (Anderson
et al., 2003; APA, 2004) others have expressed that such effects may have been
overestimated (Ferguson, 2009; Freedman, 2002).
Media research has been influenced by the social learning theory (Bandura,
1978) and the social cognitive theory (Bandura, 2001b) which has many similarities with script theory and interconnectedness of socio-cognitive factors which
can also be found in the GAM. Additional theories however were developed
with their main focus on the media such as, the catalyst model and the moral
panic theory.
75
The catalyst model has been developed by Ferguson et al. (2008) in order to
become an explanatory model for media violence. However, the model itself advocates that media violence plays a small role as a casual influence to aggression. The model suggests that stressful environment circumstances and a genetic
and early social influence creates a combination in which predisposed individuals may become aggressive. The theory is not well tested but it does represent
modern ideas that media violence is not produced because of the medium alone
but because of additional factors as well. In particular, this theory presents an
interesting aspect which has not yet been thoroughly explored in research. The
environmental stressful parameters have a role to play in aggression. These parameters are part of the social media software design.
On the other hand, moral panic theory explores the media violence from a
historical and cyclical perspective. The theory was developed by Gauntlett (2005)
not as an explanatory model for whether there is an increasing violence by media
but whether the violence is perceived to exist in any novel media by individuals.
Put simply, there is bias from researchers that have a tendency to try and confirm
the preexisting belief that media causes aggression. In addition this pattern is not
just temporary but repeats itself throughout history with every new invention
classified as “new” media.
No matter which theory one may prefer over the other, I attempt to maintain
a neutral position in respect to social media violence. My aim was not to infer
that social media increase or decrease aggression but to show that there is an opportunity for software developers to intervene and play a more proactive role
in preventing aggression through changes in the features provided on these services. By understanding the effects of media violence leading to aggression, one
can sufficiently determine the ways that the software may play a key role in this
process and alter it. The modifications can vary from providing parental control
constraints to the software to reducing the blood and gore violence for young
individuals. I argue that these are all valid considerations which should be part
of project planning and development process of social media and that currently
such constraints are only imposed mostly by the law (e.g., in the development of
video games: “Game Experience May Change During Online Play”).
5.4.2 Online Disinhibition Effect
Another centerpiece for online research into aggression is the online disinhibition
effect. The model was developed by Suler (2004c) as an explanatory theory for the
reasons behind individuals acting in a disinhibited manner when online. Such behavior has been observed in many previous studies. One in particular, conducted
by Siegel, Dubrovsky, Kiesler, and McGuire (1986) before the advent of the Internet, clearly demonstrated that computer-mediated groups produce more uninhibited behavior. In another paper, Kiesler, Siegel, and McGuire (1984) discusses
the potential for sociological and psychological research that can lead to a better
understanding for computer-mediated technologies. Many have attributed phenomena of online aggression such as internet violence, flaming, verbal attacks,
self-disclosure and other to online disinhibition (A. Joinson & Gackenbach, 1998;
A. N. Joinson, 2001; Kiesler et al., 1984). Those negative behaviors were coined
76
by Suler (2004c) as toxic disinhibition while behaviors with no negative consequences or with even positive effects are classified as benign.
The concept behind the online disinhibition effect starts with a comparison
between the offline and online world. In everyday life, social interactions are
governed by unwritten rules for social conduct. These are not enforced by authorities but there are social consequences for those that do not follow them. On
the Internet and therefore in social media these social consequences are reduced
compared to face to face communication and therefore behaviors may differ. As
an effect, people act with less restraint. Suler (2004b, 2004c) identified the following psychological aspects of online communication as causative: dissociative
anonymity, invisibility, asynchronicity, solipsistic introjection, dissociative imagination, minimizing authority.
Dissociative Anonymity
Dissociative anonymity describes the condition where individuals behave online
under the assumption that nobody knows who they are. This was coined by Suler
(2012) as “You Don’t Know Me”. Dissociative anonymity creates a universe where
identities are concealed and as such social consequences could be perceived by
individuals as negligible. This factor is not necessarily negative. Under the same
token one can see a positive result of being able to speak more freely and disclose information that would otherwise be too embarrassing for an individual.
However, similarly one can act in an aggressive manner if they wish to do so.
Disinhibited behaviors are more acceptable by individuals; they are not vulnerable to others since they cannot be identified.
Invisibility
Invisibility as a factor is coined by Suler (2012) as “You Can’t See Me”. It is described as a state where individuals are able to conceal their physical presence.
This state provides individuals with the ability to conceal their physical characteristics should they wish to do so. An example for this could be an individual
choosing to upload a picture which would not reveal his or her psychical characteristics, or even concealing their surname to avoid being ethnically identified.
Additionally, this state also provides individuals with the ability to avoid having
to be conscious about their standard social cues. In other words, in social media
many of the standard social cues such as small changes in facial expression, tone
of voice, aversion of eyes, nostril flaring and others may be absent in social interactions between users. In fact, many of the social media classifications (Kaplan
& Haenlein, 2010, 2011) are missing many if not all of these social cues. Furthermore, while this factor is similar to dissociative anonymity, it is distinctive because even if people are not anonymous there is still the factor of invisibility that
is ever-present in online communication. In fact, a recent study on online disinhibition by Lapidot-Lefler and Barak (2012) found that the most important factor
between anonymity, invisibility and eye-contact which created flaming behaviors
(a product of online disinhibition) was the lack of eye-contact.
77
Asynchronicity
Anynchronicity accounts for the fact that online communication is rarely as synchronous as in real life. There is void between the exchange of messages which
leads individuals to never expect immediate feedback and therefore being less
afraid about their actions. Wallace (2001) has sarcastically described the web as
the “World Wide Wait” (p. 113) in order to demonstrate the seriousness of the issue. While real life exchange of messages vary between 1.35 to 3.21 seconds (Zimmerman & West, 1975), online communications even in synchronous chat rooms
could reach significantly higher delays. Today, most social media still prefer asynchronous means of communications for their users with text communication being arguably the most preferable. Bandwidth constrains still prevent technologies from providing synchronous communication to their users and many virtual
game worlds (or virtual social worlds) still avoid it as an option. In fact, the expectations of users for online communications have been transformed to such degree that not only immediate feedback is not expected but it has become optional.
Many individuals do not expect a response from others. A “See You Later” (Suler,
2012) attitude seems to have been embedded into individuals while online.
Solipsistic Introjection
Solipsistic introjection can be seen as the result of all previous factors presented.
It involves a projection of one’s identity onto media. Simply put, lacking any kind
of input (e.g., standard social cues), the human mind assigns its own traits and
characteristics to another individual when interacting online. This factor acts as
a form of stereotyping where by having someone looking at lines of text one can
reach to certain conclusions about the author’s character and traits without necessarily those conclusions being true. As Suler (2012) further points, individuals
may end up playing stories in their heads and writing as if they were authors
writing their own novel without being fully aware of it. It becomes apparent,
how meanings can be distorted without individuals being aware of such things.
Dissociative Imagination
“It’s All a Game” (Suler, 2012) is the feeling that individuals may have due to the
effects of dissociative imagination. Cyberspace creates a sense of escapism from
a mundane reality to a world where different rules may apply. In such a place,
characters become alive in a make-believe dimension which is separate from the
responsibilities and limitations of one’s reality. This is the key difference between
dissociative imagination and dissociative anonymity. While in the latter identities
are simplified or reduced, in dissociative imagination there is a disconnect from
one’s personality and instead creativity and role-playing commence to form a
new online identity in a “new” world.
78
Minimizing Authority
The Internet and social media tend to be looked upon by individuals as egalitarian communities. Individuals in social media have a “We’re equals” (Suler,
2012) state of mind. This in turn, leads to them expressing opinions and ideas to
people with otherwise intimidating authority. Everyone regardless of their status,
wealth, gender or race is on a level playing field in social media. As Suler (2012)
points out, since the Internet was engineered with no centralized control, ideas
about online communities of equal individuals became the core of the Internet
experience.
5.4.3 Benign & Toxic Online Disinhibition Effects
As mentioned previously, the online disinhibition effect does not necessarily have
online negative effects for online individuals. This is a pattern that can be observed for many social and psychological phenomena where their existence seems
to be agnostic towards a group; in a non-deterministic setting they cannot be
identified as neither good nor bad. This point is important for understanding
how HCI research should behave when examining research phenomena but also
for software developers that need to develop software in order to affect social and
psychological behaviors.
Suler (2004c) provided a dichotomy for the online disinhibition effect, with
effects being described as either benign or toxic. Benign effects of disinhibition
do not have negative consequences or can even prove to have a positive effect in
social interactions. On the other hand, toxic instances are those perceived to be
harmful to participants involved.
Examples of benign effects can be seen in online collaborative projects (e.g.,
wikis, GDSS, online support systems) where invisibility along with anonymity
provides a “complete” anonymity. Participants in online alcoholic anonymous
groups are enjoying these effects. Another popular case is virtual game worlds
or virtual social worlds in which individuals can select characters that have any
qualities they desire. Such effects can help individuals that could be described as
introverts to express their true selves online (Sheeks & Birchmeier, 2007).
On the other hand, toxic effects are a serious cause for concern. Online collaborative projects can also present cases where one can find toxic disinhibition
such as environments for educational purposes or the workplace. An individual
acting in a disinhibited manner may be facing detention or lose his or her job. In
fact, toxic disinhibition effects can be found in all the spectrum of the social media
world today. Content communities are suffering from the effects of cyber-bulling
and flaming just as much as virtual game or social worlds do. The online nature
of the environment for social interaction that is provided by social media seems
to amplify disinhibition which can potentially lead to aggression.
5.5 Preventing Online Aggression
Most of the explanatory theories that were developed to explain offline and online aggression are powerful at displaying the causal influences that lead to ag79
gression but are less effective in demonstrating ways for preventing it, specifically
in online settings. Their usefulness is as such isolated to fields of sociology and
psychology. However, these theories hold tremendous applicative potential for
other scientific fields such as HCI.
I argue that there is a need for a new perspective. Most strategies put the individual at the center of interest for investigation. While this is definitely efficient
for gaining understanding, it is of little use for practical applications without taking into account that this is a human-computer symbiosis as described by Licklider (1960). Therefore, it is unwise to assert that factors of individuals are the
cause for behaviors when computers are an intricate part of the system. Theories
on aggression have successfully identified situational and environmental factors
playing a key role in affecting aggressive behavior but did not establish preventive measures. One probable explanation for this is mainly the observational nature of sociology and psychology, as well as the general lack of interest by HCI to
design social interaction. However, this lack of interest leads to a faulty thinking
pattern.
For example, one may expect that familiarity with technologies and online
communication will reduce the online disinhibition effect. This strategy is deeply
engraved into computer science philosophy. However, it is problematic. Just as
Bush (1945) expected that people may develop and attain a language in order
to fit the technology’s needs, similarly this expectation develops the idea that
people will become familiar with technologies in order to work in accordance
with a designer’s expected online behavior in social media; that is, avoiding being aggressive. Hence, behavior is expected to be forced on individuals through
familiarization or even by law. Such a way of thinking, takes the responsibility
away from software designers and puts the blame on the users. Evidence today,
shows that this is truly the case. A fallacy has been created where, while software
designers are not concerned with user behaviors, users seem to have tremendous
issues with the software that in turn affects their behavior and the way they interact with others online.
In fact, the issue is such that Wallace (2001) has named one of her sections
in the chapter describing flaming and fighting as “THE #$@@!!!& SOFTWARE”
(p. 127) in order to demonstrate that software is a factor in the aggression that
can be seen today. If one looks at aggression through a lens where a humancomputer interaction becomes the centerpiece, then part of the environmental
factors are transformed into the software itself. Rather than expecting individuals
to familiarize themselves with the software and to behave in a desirable manner,
the software can form guidelines that enforce behavior directly or indirectly. The
outcome from this revelation, is that aggression becomes a side-effect of social
interaction that can be inhibited and additionally social media can benefit from
having communities that will display less toxic disinhibition effects and lesser aggression in general. Such approach is also endorsed by Suler (2004c) which stated
that “The self does not exist separate from the environment in which that self is
expressed” (p. 325). He also argued for the potential benefits of understanding
environmental effects in different online settings and for designing software by
having such knowledge (Suler, 2000).
80
5.5.1 Software Features Studies for Aggression
I demonstrate using HCI research methods, how software can affect aggressive
behavior and how one can identify certain elements of the software and alter
them to rid social media of such a negative behavior. The basis for identifying
social media features with potential for affecting aggression is based on current
psychological and sociological literature while merging it with the current practices for social media software development. My main aim was that the features
investigated be broad enough so that results can be generalized and be applied
in most if not all social media classifications. Below I itemize a list of features that
were explored for which findings are presented on the next chapter.
•
Different anonymity states produce different effects of online disinhibition
and in turn aggression.
•
When online messages are lost due to submission errors, there is a loss in
the quality of messages which can indirectly lead to aggression through
frustration or textual misunderstandings.
•
The quality of error messages and suggestion of alternatives in them can
reduce the frustration caused by the incident and therefore reduce aggression.
81
C HAPTER 6
R ESEARCHING THE I NDIVIDUAL F EATURES FOR
A GGRESSION IN S OCIAL M EDIA
6.1 Introduction
In this chapter, I present the research findings for the features that were explored
in relation to aggression. Each feature examination was given a separate section
since most literature involved for each feature as well as its method may not apply to another. Additionally, some of the work has been published in journals and
conferences and as such, summaries of the background, processes and findings
are presented here as well. Finally, while individual conclusions for each section
are presented here, they are all combined later on the chapter describing the social
interaction design framework.
6.2 Anonymity States and Aggression
For this part of the study, I wanted to investigate the effects of anonymity states
and aggression. Since the online disinhibition effect theory directly described the
effects of anonymity as a factor for disinhibited behavior, it became an essential
part of my investigation. Keeping in mind that the research results are required to
always have had direct applications in the software design and development process, I decided to investigate how anonymity states may contribute to aggressive
behavior. Establishing that there is an effect will directly give control to designers
over how much disinhibited behavior may be expected in their social media services. While the choice of providing an anonymity state is an important decision
when designing social media, it has never been investigated in terms of the effect
that it could have on aggressive behavior for the users.
This study has been part of an investigation that resulted in a conference publication (Tsikerdekis, 2011) as well as a published paper on a scientific journal
(Tsikerdekis, 2012). Both papers describe in detail my research study involving
anonymity states and aggression. While specific details can be found on these
papers, a summary is also presented here. Additional statistical analyses for the
original hypotheses are applied in order to provide a better perspective for the
82
6. R ESEARCHING THE I NDIVIDUAL F EATURES FOR A GGRESSION ...
findings. Examining results from multiple perspectives could give a better insight into the overall processes involved. Furthermore, since the publication of
the original paper, there was a similar study involving anonymity and online disinhibition (Lapidot-Lefler & Barak, 2012) that is worth discussing along with the
results obtained from the work involved in this dissertation work. A brief theoretical introduction follows.
6.2.1 Background Theory
The theoretical background of the study revolves around two anonymity states
identified by computer science literature (Pfitzmann & Hansen, 2010). While
more anonymity states exist in literature, I argued that these cannot be easily
identified by individuals online (Tsikerdekis, 2011, 2012). The states that can be
identified by users are pseudonymity and complete anonymity. Additionally, I
was interested in the state where participants used their real names which acted
as my control state. Studies on anonymity have confirmed a link with aggression
(A. N. Joinson, 2006; Suler, 2004a, 2012; Wallace, 2001). However, none at the
time produced research that investigated if there is a difference between these
anonymity states that would provide designers with control over aggression.
Anonymity is associated with positive (Ainsworth et al., 2011; Lea, Spears, &
de Groot, 2001; G. B. Lee, 1996; Tanis & Postmes, 2007) as well as negative effects (Davenport, 2002; Farkas, Ziegler, Meretei, & Lörincz, 2002; Keesom, 2004;
Rains, 2007; Wallace, 2001) for users in social media. It is identified as an important part of the online disinhibition effect in which dissociative anonymity, invisibility, solipsistic introjection and dissociative imagination affect the social interactions of individuals through anonymity (Suler, 2012). In my study, my main
focus revolved around dissociative anonymity and dissociative pseudonymity
(Tsikerdekis, 2011, 2012). I argued, that while dissociative anonymity can be a
factor for the case of complete anonymity as well as pseudonymity, the same cannot be said about dissociative imagination. Dissociative imagination is a factor
that involves more creativity and as such it is more likely to play a significant
role in the case of pseudonymity rather than complete anonymity. Furthermore,
if that is true, then there may be a difference in the disinhibition that individuals
would exhibit between the states of pseudonymity and complete anonymity.
6.2.2 Method
In order to assess this hypothesis, I conducted a survey on Facebook, a social
networking website in which respondents faced different situations in controversial topics. These contained problems that in turn respondents were required
to respond to with one of the predetermined set of written messages. The research was an experimental design survey with three treatments, real names,
pseudonymity, and complete anonymity. The messages consisted of four-point
ordinal scale ranging from messages that were very polite, to very rude messages. This way standardization was achieved between all the responses which
increased internal validity. Additionally, under the suspicion that the opinion on
a topic may affect the outcome, participants were asked prior to facing the scenar83
ios, their opinion about each topic on a three-point ordinal scale. This additional
variable would help determine if the strength of one’s opinion could play a significant role. The survey was administered in a form of a Facebook application
and a volunteer/snowball sample was obtained through advertisement in academic and non-academic groups and fan pages on the social networking service.
A more detailed description of the survey design and process can be found on
the research article (Tsikerdekis, 2012).
6.2.3 Findings and Results
According to the analysis of data in the study, anonymity states were found
to have an effect on the way individuals reacted but only conditionally to the
self-reported level of opinion on a topic. This can be seen on figure 6.1. Under
conditions where individuals stated that their opinion about the topic was “extremely strong” their behavior was significantly different and more disinhibited.
However, that effect was found only for the condition of pseudonymity while
analysis for complete anonymity did not produce statistically significant differences compared to the control state as well as pseudonymity. The finding that
pseudonymity produced more aggressive responses was in accordance with the
literature on online disinhibition.
The effect of an anonymity state in the way a participant would respond was
also found to be greatly dependent on the scenarios that were used. The scenario
involving animal rights showed a great difference compared to the other two
scenarios used. Figure 6.2 depicts these differences.
The conclusions in the original paper rendered complete anonymity as a safer
choice for software designers and developers that want to provide a form of
anonymity for their users while wanting to avoid negative behaviors that may
come with it. On the other hand, pseudonymity was not recommended for social
media services under which discussion of highly controversial topics is expected
to take place since this can potentially increase disinhibited behavior and in turn
the chances for aggression.
6.2.4 New Hypotheses for the Additional Analysis
Several statistical analyses from the original study are worthy of examining based
on a Bayesian framework. This can potentially help in reassessing some of the results as well as help determine if null hypotheses were more likely for results that
did not achieve statistical significance. The following hypotheses were formed for
this re-evaluation:
•
H1 : Responses are dependent on the anonymity state through which individuals respond.
•
H2 : There is a dependency between importance of a topic and the responses.
•
H3 : There is a dependency between scenarios and the responses.
When examining the data as a repeated-measures design across all anonymity
states for all scenarios and self-reported importance of a topic, a Friedman’s test
84
Figure 6.1: 3D clustered graph depicting the responses that individuals gave under each anonymity state and their reported strength of opinion on a topic. Responses under the state of pseudonymity (here depicted with a light brown color)
show a substantial effect
failed to show statistical significance, χ2 (2) = 4.671, N = 160, p > .05. This failed
to provide support for H1 . However, the same data can be reexamined under a
Bayesian network.
A contingency table was created using the three anonymity states and the
four responses of individuals. This can be seen on table 6.1. Variables lose their
ordinal nature and become categorical, however, they can still provide evidence
on whether a dependency exists. H1 becomes the hypothesis of dependence and
H0 becomes the hypothesis of independence between the two variables. Using
the ctable function from the LearnBayes package for R, the Bayes factor that was
produced provided support for the hypothesis of independence, BF01 = 138.06.
This is considered decisive evidence that anonymity states have no effect on the
answers of the respondents.
However, in this result there is unwanted variance due to other variables
as it was found in the original paper. A small effect was found between the
answers and the importance of a topic, rs = .142, p < .01. A second dependency was found between the scenarios and the importance of a topic, χ2 (4) =
30.371, p < .001, V = .177. Looking for the standardized residuals within each
cell, the scenarios about abortion and animal rights had some cells with residuals higher than 1.96. Most of the respondents rated the scenario about the animal
85
(a) Abortion scenario
(b) Animal rights scenario
(c) Death penalty scenario
Figure 6.2: Responses based on anonymity states for each one of the three scenarios. Each left column represents complete anonymity, each middle column
represents using real names and each right column represents pseudonymity.
rights higher in the scale of importance while the opposite occurred in their responses. Additionally, I conducted an analysis to look for a dependency between
responses and scenarios which showed a statistically significant result as well,
χ2 (6) = 56.737, p < .001, V = .242. In general, the NHST analysis provided evidence for both H2 and H3 .
Using Bayesian analysis the results became harder to interpret. After forming
a contingency table based on the variables, Bayes factors were obtained through
an analysis using the ctable. The test between the answers and the importance of a
topic produced a Bayes Factor in favor of the null hypothesis, BF01 = 1.32. Given
that the result is so close to one, it becomes harder to interpret. However, when
testing for a model “close to independence” (here denoted as Hc for clarity) using
100,000 simulations and a Dirichlet parameter K between 2 and 7, the bfindep
function produced the highest Bayes factor in favor of the model “close to independence” BFc0 = 12.76. This result shows that if one is willing to accept that
86
Response
Anonymity State
Very polite
Polite
Rude
Very Rude
/
Complete Anonymity Real names Pseudonyms
82
47
15
20
62
53
23
24
67
49
29
17
Table 6.1: Contingency table for anonymity states and responses for all scenarios
there are uncertainties in data, then a dependency between these two variables is
unlikely. This may seem different than the result obtained by the NHST analysis
but this is not the case if one takes into account the small effect that was produced
by Spearman’s correlation analysis. On the other hand, if we consider that the
importance of a topic was found to be dependent on the scenarios then a model
close to independence is unlikely. In fact the Bayesian analysis using the ctable
function produced a Bayes Factor in favor of the dependency between the importance of the topic and scenario, BF10 = 11, 245. Strong evidence was also found
for a dependency between the responses and scenarios, BF10 = 391, 556, 237. The
Bayesian analysis while providing support for H3 had showed little evidence for
H2 .
While a dependency between responses and strength of opinion was established, there was a need to evaluate this in relation to anonymity states. In my
original paper I proceeded by conducting correlation tests for the importance of
a topic and responses within each anonymity state. I was able to trace the degree
through which the importance of a topic affected each anonymity state. As the importance of a topic would rise so would the response become ruder. I conducted
1-tailed tests that produced non-significant results for the state where participants
used their real names, rs = .068, p > .05, and for the completely anonymous state,
rs = .083, p > .05. However, the state of pseudonymity produced a statistically
significant result, rs = .288, p < .001. This analysis was similar to the one conducted on the original paper (Tsikerdekis, 2012).
I wanted however to reevaluate these results using the Bayesian framework
since I wanted power to assert conclusively that the states where participants
used real names and complete anonymity, were independent of the self-reported
opinion about a topic. The Bayes factors obtained using the ctable function for the
states of real names, pseudonymity and complete anonymity were BF01 = 11.20,
BF10 = 8.04, BF01 = 48.83, respectively. The results seem to agree with NHST
analysis. Both, the state where respondents used their real names as well as the
state that they were completely anonymous support the hypotheses of independence from the importance of a topic. On the other hand, under the state of
pseudonymity, responses were dependent on the self-reported importance of a
topic.
87
6.2.5 Discussion
The reexamination of the original results using Bayesian analysis seemed to provide evidence to support that indeed pseudonymity as a state is different than the
other two states in the study. Additionally, the level of importance did not affect
how participants acted under complete anonymity or when using real names.
Additional remarks were also illuminating. One of them was the powerful
dependence that was found between the importance of a topic and the scenario,
as well as for the effect that scenarios had on the responses of the participants. In
both cases the NHST analysis showed relatively small effects while the Bayesian
analysis was definitely more favorable of dependencies.
Another major difference was found when examining a dependency between
importance of a topic and responses across all anonymity states. While the NHST
analysis indicated that there was a dependency, the Bayesian analysis showed
that a model “close to independence” is also highly likely. However, further examination of this dependency within each anonymity state revealed that this
was not the case. The dependency was not due to unsystematic variance but
was caused because of the differences between pseudonymity and the other two
states. Pseudonymity seems to be enhancing the dependence between the importance of a topic and the way individuals respond; a probable cause of dissociative imagination having a larger effect as speculated at the beginning of the study
(Tsikerdekis, 2011).
Furthermore, the results from this study may seem to contradict the study
conducted recently by Lapidot-Lefler and Barak (2012). In their study strong evidence was not found between participants that used their real names along with
other personally identifiable information and their pseudonyms. However, there
are two important differences between the two studies. In their study, LapidotLefler and Barak (2012) automatically assigned pseudonyms to its participants,
while this study requested participants to assign their own pseudonyms a priori.
This may indicate that self-assigned pseudonyms have more power to produce
toxic disinhibition than automatically assigned pseudonyms. Furthermore, there
was no measurement of a self-reported strength of opinion on a topic which was
an essential link that helped discover the effect that it had on pseudonymity.
Considering all the above and given the new results using the Bayesian analysis, a revised causal model can be created for the variables measured. Figure 6.3
depicts the model.
The following causal model is still conditioned only for the state of pseudonymity
where an effect was found. According to the results, the states where respondents
used their real names and completely anonymity produced similar results and
did not affect responses.
Recommendations
As I described in the original paper, given the evidence, the decision for providing users with the option of pseudonymity in social media is bound to have
negative consequences. In general, it is expected that it will raise the levels of
aggression in the community. Unfortunately, for certain classifications of social
88
Figure 6.3: Model for level of aggression in responses under pseudonymity based
on situational context and strength of opinion about a topic.
media, this is the only option. However, for other classifications such as online
collaborative projects, software designers are not limited to just pseudonymity.
A collaborative wiki can have the choice of real names, complete anonymity or
even both with the latter provided as an option at any time a user wishes send an
anonymous message. Since complete anonymity was not found to have the same
effects as pseudonymity, it can be considered as a safer alternative and therefore
can provide designers with control over aggressive behavior on social media.
6.3 Quality of Error Messages and Suggestion of Alternatives
One of the features suspected of contributing to aggression is the design of error
messages. In social media software today, software errors can happen frequently
and users are confronted with all sorts of unknown situations which they are required to resolve. To add to the confusion, a limited amount of time for finding
the information that a user truly seeks could make matters worse. The final outcome of such interaction results in an increased frustration for individuals that
could lead aggression. The reasons for this outcome can be found in neoassociation theory (Berkowitz, 1989, 1990, 1993a) as well as the frustration-aggression
theory originally developed by Dollard et al. (1939).
This part of the study investigated if the quality of error messages as well
as the suggestion of alternatives can potentially improve user perceptions about
such unfortunate incidents and achieve higher levels for comprehension, helpfulness and appeal. These features can potentially reduce the negative effects of an
error message. In this part of the study, I administered a survey which aimed to
investigate the potential for error messages with better quality of text, graphics
and suggestion of alternatives to reduce frustration on users.
6.3.1 Error Messages and Frustration
Nielsen (1994a, 1994b) developed several heuristic guidelines to be followed by
software designers and developers. One of the most important aside from the
avoidance of errors was the use of error messages that can be helpful and be
89
understood by the users. In particular, a message should help users recognize
the problem, diagnose and recover from errors. Similarly, in the extended version
of Nielsen’s heuristics, Ryu (2007) suggested that each and every error message
should offer at least one solution or a link to a solution in the error page.
However, evidently these heuristics are not followed in many cases. In their
observation study on the causes for user frustration during their work with computers, Ceaparu, Lazar, Bessiere, Robinson, and Shneiderman (2004) found that
many of the causes were attributed to a clear violation of the guidelines discussed previously. In addition, such frustrated events also reduced the time spent
when performing actual work on a computer. Error messages are among the chief
causes for end-user frustration.
While in the isolated universe of yesterday, frustration and even aggression
was contained to local distances, today technical issues that do not prompt for a
clear path for resolution may be more costly. Hara and Kling (1999) conducted
a qualitative case study of a Web-based distance education course. They discovered that some students had severe issues with error messages and they lost a lot
of time trying to resolve them. One of the participants clearly linked their frustration with the anger that they felt. Their conclusion was that issues that can cause
frustration inhibit opportunities for education.
In today’s complex system such issues do not just haunt the users but also
system administrators. Maglio and Kandogan (2004) described two cases under which system administrators were frustrated with resolving error messages.
Many errors were completely disconnected from the original causes and the error
messages were not helpful. Their recommendation was to try and display messages that can be interpreted within the context of which they occur.
In social media design, all of the cases described above should be of great
cause for concern. They are not just technologically complex systems but also
include a social ecosystem which is fragile to events that can cause frustration
and aggression. The degree where a message can be understood by a user, the
aesthetics of the unfortunate incident and the proposition of alternative paths are
all important factors to be considered. Are such heuristic guidelines going to have
a significant impact on these factors and in turn help reduce frustration levels for
the users? This research question helped formulate the three main hypotheses for
this part of the study:
•
H1 : Messages that explain the problem and provide more relevant information will result in an increased comprehension.
•
H2 : Messages that provide alternative paths and solutions will be rated as
more helpful.
•
H3 : Messages that have better graphics will be more appealing to users.
6.3.2 Method
In order to assert these hypotheses I developed a set of questions which were
included in a larger survey that was administered for the purposes of this dissertation work. This survey will hereby be referred to as AG1 since parts will be
90
used for asserting other features studied in this dissertation later on. The segment
consisted of two different versions of error messages that users encountered in a
hypothetical scenario. Users were randomly assigned to groups. For the purposes
of realism, I decided to use already existing error messages which coincidentally
were developed by the same company for two different products. These are presented side by side on figure 6.4.
(a) First version
(b) Second version
Figure 6.4: The two versions of 404 errors that respondents encountered.
The first error message uses code to state the problem and is not specific in
its first statement with the type of error. Additionally, the final statement may
be considered irrelevant with the problem. However, the image may add to the
aesthetics and appeal and therefore the message may be seen as less dramatic to
the user. On the other hand, the second version does not use code to state the
error and makes a clear statement about the type of error. Additionally, there are
clear suggestions and alternative paths that a user can take so that the problem
can be resolved.
Users were asked a set of questions that aimed to measure comprehension,
helpfulness and appeal of the messages. These questions can be seen on figure
6.5. Additionally, a qualitative question was added to the study in an attempt to
uncover the reasons behind user choices.
The survey was launched on Facebook as an application, and advertised in
several academic and non-academic pages and groups. Additionally a snowball
sampling method was used throughout the survey allowing participants to invite
their friends to take part. The survey was originally designed in English but later
91
(a) First version
Figure 6.5: Questions included in the survey investigating the effectiveness of
error messages.
was translated in Greek as well as Spanish. The survey took place between May
1, 2012 and August 21, 2012.
6.3.3 Results
A total of 206 individuals visited the application that hosted the survey of which
157 successfully completed it. The demographics of the sample are presented on
figure 6.6. The sample consisted of a young and middle-age population with a
limited however population for the extremities. There was balance between sexes
and the majority of people came from Europe (72%), although other continents
also made significant contributions to the sample. There were no respondents that
came from Africa in the sample.
Figure 6.7 depicts the results for the three variables. Comprehension was higher
for second version of the error message (M = 7.64, SD = 2.573) than the first
version (M = 6.35, SD = 3.386). Helpfulness also had a higher rating for the second version of the error message (M = 6.49, SD = 2.873) than the first version
(M = 2.87, SD = 2.569). Smaller differences were found for the aesthetics variable between the second (M = 5.09, SD = 3.003) and the first version of error
messages (M = 4.96, SD = 2.766).
NHST analysis was used in order to assess the hypotheses for this study.
92
Figure 6.6: Graphs depicting demographic information for the sample that took
part in the AG1 survey
While the ten-point scale variables could have been used as interval variables,
violations of normality assumptions made it impossible to be used in parametric tests. Instead, Mann-Whitney U tests were used to determine if the groups
reported significant differences for the variables of comprehension, helpfulness
and appeal. The analysis showed significant differences for comprehension (U =
2424.5, Z = −2.344, p < 0.05, r = .187) as well as for helpfulness (U = 1088, Z =
−7.099, p < 0.001, r = .566). These results provide support for H1 and H2 . On the
other hand, results for the appeal did show any statistically significant differences
(U = 3024, Z = −.202, p > 0.05, r = .016).
Qualitative analysis
Findings from the qualitative analysis are in accordance with the quantitative
analysis. The differences between the error messages are apparent in the comments of participants especially for the first error message. One respondent that
received the first version of the error reported that “the message should include
93
Figure 6.7: Results for the three aspects of the two versions of error messages
evaluated by respondents.
a solution so I can fix the error.” Another respondent specifically reaffirmed hypotheses H1 and H2 : “Not at all useful. Didn’t explain what the problem actually
is and didn’t tell about the solution as well.” Evidently the need for solutions is
extremely high among these respondents. The language used in the text also went
under scrutiny as this respondent demonstrates: “It’s the ’That’s all we know’ that
sounds uncaring and makes me give up hope.” While users giving up hope because of these error messages could decrease productivity, communication and
interaction on social media, frustration is a more severe consequence that could
be linked to aggression in a community. A respondent clearly demonstrated this
link between a faulty design of error messages and frustration by stating that
it is “frustrating. if an [sic] URL is not available, the link for that URL must be
taken down.” Finally, one can find even evidence for the non-significant result
regarding the aesthetics of the first error message: “The picture and message is
not user understandable. they [sic] can change into the website you are looking
is not available not a message 404.”
On the other hand, while quantitative results seem to have favored the second error message, respondents did not seem to do so in their comments. For
example, one respondent requested additional solutions for resolving a problem:
“The message could indicate other ways to resolve the issue. One can use the Way
Back machine to view a broken link in hopes of discovering its non-existence any
longer [sic].” Another respondent stated that “The suggestions are not helpful. I
generally know to try a different approach in the search engine. The only helpful
information is that the link is broken.” Evidently, computer and software efficacy
is an important factor that determines the usefulness of a message. Users with
94
more advanced skills will probably require more solutions. The same statement
was also made by another respondent: “I think there’s something wrong with
the website. Normally, if the website isn’t a first priority, I always quit and try to
search other websites though there’s [sic] suggestions below.”
Finally, qualitative results contained some mixed messages about aesthetics.
As an example, a respondent viewing the first error message said “cute but not
helpful,” indicating that aesthetics actually mattered in this case. Similarly, another respondent when viewing the second error message dismissed the solutions as not helpful and made a request for better aesthetics: “give a funny face
will make me feel better.” While quantitative results did not achieve significance
for the aesthetics of the message, it may be the case that an effect may exist but
is probably small. However, these cases demonstrated that for some individuals,
aesthetics matter even for error messages.
6.3.4 Discussion
In general, results from this study seem to reaffirm heuristics that are widely used
as guidelines in human-computer interaction. More importantly however, they
demonstrate that such differences in the comprehension of the text as well as the
helpfulness of error messages statistically matter. In turn these features have the
power to decrease negative effects such as frustration which could be the cause
for aggressive behavior in social media. Aggression in social groups can be contiguous (Cairns, Cairns, Neckerman, Gest, & Gariepy, 1988; N. E. Goldstein,
Arnold, Rosenberg, Stowe, & Ortiz, 2001) which could contribute towards developing flame wars online (Wallace, 2001). As such these features are not helpful
only for improving user experience on social media software but also social experience.
An important find for this study is that users seem to have higher expectations
than what heuristics provide. Even in the case of the second error message where
alternative solutions were provided, respondents still seem to be craving for more
and more radical solutions. While providing the obvious solutions will lead to an
increase in helpfulness, an effort to develop solutions that go beyond the current
expectations of users may be appreciated by individuals.
Additionally, while aesthetics may still play a role in error messages, strong
evidence was not found for the H3 hypothesis. However, some individuals seem
to have an artistic sensitivity for aesthetic details according to the qualitative results.
Recommendation
It is recommended that designers and developers will avoid the use of codes,
and cryptic language for error messages. Additionally, text that may not be considered necessary or related to the error messages should be omitted from the
message. Error messages should clearly communicate in non-technical terms the
problem and provide solutions. Solutions need not be limited to the bare obvious
but attempt to provide more innovative ways for users to work around a problem. Finally, while evidence for aesthetics was not found in this study, designers
95
and developers should make an effort to design messages that would achieve
near-universal appeal. Changes such as the ones described here can significantly
decrease the potential for frustration by users and in turn aggressive behavior in
social media communities.
Limitations
This study has used a mixed-methods experimental design survey to assert that
the design of error messages can help decrease frustration and the potential aggression. However, this is still an indirect measurement for establishing that error
message could contribute to aggression. The study relied on self-reported measures of helpfulness and comprehension to determine their effect on frustration.
Additional future studies may wish to explore scenarios where aggression can
actually be observed shortly after individuals face an error message.
6.4 Message Submission Errors and Recovery
Another important aspect to investigate in relation to aggression was incidents
of submission errors. Packet loss is nothing new to computer science and even
with the sophisticated degree of hardware and software occasionally information is lost. In many occurrences this information is part of messages being exchanged between users in social media. While the event by itself is unfortunate
and definitely contributes to user frustration, I wanted to investigate the recovery events followed by users after the unfortunate incident. My main concern revolved around what was the action taken by the user since the original message
was lost. Was there a rewritten message being submitted or did users abandon
their effort to send the message completely? If the user decided to rewrite the
message then a loss in the quality of the message could result in a less effective
communication. The latter is highly likely if the maxims of the cooperative principle (Grice, 1989) are not satisfied with the rewritten message.
6.4.1 Messages, Communication and Aggression
Messages and the exchange of messages online is the subject of study of computermediated communication. Computer-mediated communication is defined as any
“communicative transaction that takes place by way of a computer, whether online or offline” (McQuail, 2010, p. 552). The field of study investigates the effects
of such communication on individuals, groups of individuals and the society at
large.
Gricean maxims (Grice, 1989) or as else known the cooperative principle,
have been of interest for studies in computer-mediated communication (Herring,
1999; Nastri, Peña, & Hancock, 2006; C. B. Smith, McLaughlin, & Osborne,
1997). Grice (1989) described maxims that rather than guidelines are seen as conditions for successful communication. There are four maxims: maxim of quality,
maxim of quantity, maxim of relation and maxim of manner. The maxim of quality requires from individuals to be truthful, while the maxim of quantity requires
96
from individuals to be specific with the amount of information that they share
(no more and no less than what is required of them). The maxim of relation requires for individuals to maintain relevance to the topic. Finally, the maxim of
manner requires for individuals to be clear such as avoid ambiguity, obscurity of
expression and be brief. These are mutually assumed from the speaker and the
listener but whenever they are not followed implicatures could occur. These described meanings which are not explicitly conveyed in the actual conversation.
In turn, these implicatures have a great potential for misunderstandings occurring in communication which not only affects the efficiency of communication
but also can be the cause for conflict between individuals in online communities.
The effects and attempts of individuals to be in accordance with these maxims
in order to successfully communicate to other their position can be seen widely
on the Internet. For example, Nastri et al. (2006) studied the abbreviations used in
away messages by users and argued that these were made by users in an attempt
to be in accordance with the maxims of quantity and relevance. Such text abbreviations can be seen in all sorts of social media from collaborative projects for the
workplace, classrooms but also video games (Peña & Hancock, 2006). Additionally, non-adherence to Gricean maxims, especially the one relating to the maxim
of quality could violate social norms within a group which could become the start
of flame wars (C. B. Smith et al., 1997). Herring (1999) found that in many cases of
online exchange of messages, the maxim of relevance was violated which led the
invitation of jokes and humorous play. While in this case Herring (1999) argued
that this was a positive effect that explained the popularity of computer-mediated
communication in many cases the development of implicatures could influence a
conversation in a negative manner.
Several hypotheses were developed to establish if maxims have been violated
and under which conditions. Specifically the maxims of interest were related to
quantity and quality which was likely to be affected by a potential rewrite of
the original message. Messages with shorter amounts of information or of less
quality, could lead to implicatures. It was expected that two important factors,
the time and the importance of the message, will affect the action that will be
taken by the user and will also affect the quality of the message. Additionally,
the action taken by the user will probably have an effect on the quality of the
message. Based on the above, I itemize below the main hypotheses:
•
H1 : The condition of somebody being pressed for time will affect the action
that he or she will take.
•
H2 : The importance of the message will affect the action that he or she will
take.
•
H3 : The importance of the message to be sent will be dependent on the
whether someone is pressed for time.
•
H4 : The quality of message that will be sent again will be dependent on the
importance of the message.
•
H5 : The quality of message that will be sent again will be dependent on
whether someone is pressed for time.
97
•
H6 : The quality of message that will be sent again will be dependent on the
action taken.
6.4.2 Method
A survey was used to assess the following hypotheses and was administered as
part of the AG1 survey (see section 6.3.2). The method used for assessing these
hypotheses relied on the respondent’s memory to report events where a message
submission error had occurred in their past experience online. It was anticipated
that this method will sufficiently be able to demonstrate the relation of external
factors such as time and importance of a message which would have been difficult to be replicated in laboratory conditions.
The survey consisted of a set of five questions which are depicted on figure
6.8. The primary question asked respondents on whether they have experienced
such an incident in their past. Those that answered negative did not answer the
rest of the question. The second and third question aimed to develop variables
for time pressure and importance of the message. The fourth and fifth questions
aimed to see what action the respondents took and what the quality of the second
message that they sent again was (in case there was a second message).
Figure 6.8: Survey questions for the message submission error and recovery survey.
98
6.4.3 Results
Demographic statistics about the consistency of the sample for the AG1 survey
are presented in figure 6.6. The total number of survey respondents that went
through the whole survey was 157. This was the initially used sample for the
analysis. 107 out of the 157 ( 68.15%) of the respondents reported that they had
experienced an event in their past where a message was lost due to submission
errors. This sub-sample was further used for the subsequent analyses.
In general, results showed that 48.6% of the respondents reported that they
chose to rewrite the message in the exact length, 30.8% rewrote a shorter message and 5.6% skipped writing the message altogether. This result is indicative
of a probable violation for the maxim of quantity. There was also 15% of individuals that skipped answering this question and their answers were omitted for
subsequent tests.
Most respondents reported that their message quality remained about the
same (45.8%). Sixteen-point-eight percent rated that their messages had a higher
quality, while 28.9% rated that their message was of lesser quality. This is another
indication for a probable violation of the maxim of quality. While participants
may have been truthful in their second message, omitting important arguments
and information may have been part of the process that led to a perceived lower
quality message. An additional 8.4% of participants skipped answering this question and were omitted for the subsequent statistical analyses.
A chi-square test was used to assess the H1 . A statistically significant effect
was not found for the variables of time and action taken by the respondents,
χ2 (2) = 2.773, p > .05, V = .175. However, due to violations of assumptions
for the chi-square test, a Fisher’s exact test was used to obtain an accurate significance value. The result remained non-significant, p > .05. Using a Bayesian
analysis with the ctable function I obtained a Bayes factor that favored the nullhypothesis, BF01 = 2.58. While both results have produced evidence for rejecting
H1 , the evidence was not strong.
In order to assess H2 , I converted the ten-point self-reported scale for the importance of the message to a five-point scale and developed a contingency table
for the action variable and the newly created variable. The chi-square test showed
no statistically significant dependency χ2 (8) = 9.445, p > .05, V = .228. However, assumptions for chi-square were violated and Monte Carlo algorithm was
used with confidence level at 95% and 100,000 samples which produced a nonsignificant value, p > .05. To assess the likelihood of a null hypothesis against
H2 , Bayesian analysis using ctable produced a Bayes factor favoring the null hypothesis, BF02 = 28.34. This provided strong evidence that the importance of
the message does not affect the action that will be taken in the occurrence of a
submission error.
A similar procedure was followed for testing the H3 hypothesis for the variable of time and the five-point scale for the importance of a message. A chisquare test found a statistically significant dependency between the two variables, χ2 (4) = 13.403, p < 0.01, V = .356. A strong Bayes factor was also produced in support of H3 , BF30 = 26.74. While it is difficult to determine causality
between the two variables, this is considered strong evidence that whenever the
99
message was important, respondents had time at their disposal. Figure 6.9 depicts
the dependency for the two variables.
Figure 6.9: Bar chart depicting the ten-point scale for importance of a message
and time availability.
A Spearman’s correlation for the variables of the quality of message and the
ten-point scale for the importance of message produced a statistically significant
result, rs = −.285, p < 0.01. The result indicated that as the importance of the
message got higher, the message quality lessened. Therefore, H4 was supported
but in a highly concerning way.
Furthermore, time was found to have a significant effect on the quality of the
message, χ2 (4) = 12.668, p < .05, V = .360. Due to violations of assumptions
I used Monte Carlo algorithms (95% confidence level using 100,000 samples) to
obtain a significance value which reaffirmed the initial result, p < .01. This was
further reassessed using Bayesian statistics using the ctable function which produced a Bayes factor favoring the H5 hypothesis, BF50 = 7.32.
The final hypothesis stated that a relation between the action taken and the
quality of the message may exist. A chi-square between the two variables returned a result that approached significance, χ2 (8) = 14.174, p = .077, V = .287.
However, assumptions for chi-square were violated and so an attempt to obtain
a more accurate p value using Monte Carlo sampling (95% confidence level using
100,000 samples) returned a similar result, p = .076. The Bayesian ctable function
also returned a Bayes factor against H6 hypothesis, BF06 = 16.58. Results of the
analyses for H6 showed evidence that a dependency between these two variables
did not exist.
100
6.4.4 Discussion
The results for this part of the study were illuminating as well as surprising. Support was provided for hypotheses H3 , H4 and H5 . These findings indicate a strong
triangle of relation between the time available for participants, the importance of
their message and the quality of the new message that will be written. One of
the most troubling results was the case between the quality of a message and
the importance of the message. There seems to be a negative dependency which
suggests that as the importance of a message rises the quality of the newly rewritten message falls. In organizations where key messages are exchanged between
employees this loss in quality may have drastic implications in the efficiency of
the work. Maxims of quality are genuinely affected which could be the cause for
implicatures and even worse lead to aggression at the workplace. This effect also
applies to other social media services as well. The more participants perceive a
message as important and they have a limited amount of time to submit it, the
more likely it is that the message will be of lesser quality. On the positive side of
things however, a positive effect was found between time and importance which
suggests that usually there is enough time available when the message is important.
On the other hand, expected hypotheses that aimed to find a dependency between the variables and the action that a user will take were not found. Specifically, there is no evidence to suggest that the importance of the message will affect
the action taken by a participant. Put simply, participants have equal chances of
not writing or rewriting a message in respect to the importance of a message; a
troubling finding. Additional dependencies between the time available and action taken by a user as well as between the quality of a message and the action
taken by a user were not found. However, the evidence was not conclusive on
these dependencies and so caution is advised.
Figure 6.10 depicts the model using all the variables studied based on the
results. There is a dependency between the two variables of time available and
the importance of a message and in turn, these two variables affect the quality of
the new message. However, their effects are reverse. Further, dependencies with
inconclusive evidence were also included in the model so that further studies can
establish whether there is a genuine effect between these variables.
Recommendations
The results have direct implications for social media design. Submission errors
are unacceptable from a design standpoint given the technologies available. This
study demonstrated that they can be the cause for implicatures which can result
to more severe consequences such as aggression. Designers should take steps to
ensure that the information on social media is preserved for all forms, especially
the ones where messages are exchanged. Failing to do so threatens the social experience of users and effectively their efficiency of communication and collaboration.
101
Figure 6.10: The error submission model based on the results.
Limitations
This study attempted to develop a working model through which one can understand the effects of a submission error on the overall communication for social
media. However, there are limitations when using a survey to determine these effects. Respondents were required to recall from memory what happened in their
past, which may affect the results. Additionally, this method also relied on their
perception of the incident and the actions taken.
102
C HAPTER 7
T HE S OCIAL I NTERACTION D ESIGN F RAMEWORK
The previous chapters presented research on groupthink and aggression that had
been conducted in order to demonstrate how features on social media software
can affect these phenomena. Based on these results, I argue that social behavior is
not only affected by interfaces but one can proactively design interfaces to engineer behavior. This chapter helps bind together, all previous information and results based on the two showcase phenomena used in the study and demonstrates
how interfaces affect behavior. This is achieved by providing a new definition
for social interaction design based on social experience goals and a framework
through which one can design and develop software. It accounts for desirable
and undesirable social behaviors, discusses methods through which one can obtain relevant results for such behaviors and implements these results, to help produce better social media software that take social experiences into account.
7.1 A New Look on Interaction Design
In the introduction chapter, I have already discussed the benefits provided by interaction design as well as of social interaction design which is a relatively new
concept. However, goals and ideas of social interaction design are not clearly established on the overall existing framework of interaction design and humancomputer interaction. Having already asserted that social behavior can be altered
through the means of social media design, implementing this into a framework
that can be adopted and used in software design is essential.
Prior to demonstrating this, two significant issues have to be considered. As I
pointed out earlier on in the introduction chapter, J. Brown (1997) discussed that
in part, the fault for dysfunctional systems falls in HCI specialists that do not
properly convey their results to software engineers. Results should not be presented in an abstract way but concisely and specifically. The second issue is that
by introducing more terminologies into a scientific domain, this can clutter and
confuse the people involved within it. If social interaction design is to be taken seriously and be implemented into the work process of social media development
teams, it needs to be implemented in a seamless and painless manner. These two
points served as guides for the proposal that I am presenting in this chapter.
103
7. T HE S OCIAL I NTERACTION D ESIGN F RAMEWORK
7.1.1 Defining Social Interaction Design within Interaction Design
There are three key characteristics in the interaction design process: (1) focus
on users, (2) usability and user experience goals and, (3) iteration (Rogers et al.,
2011). In their book, Rogers et al. (2011) described that two main goals are essential for interaction designers. These were defined as usability goals and user experience goals. Both types of goals are not just mere white noise in the background
but an essential part for effective software design. Strategies such as needfinding
are an essential part of identifying development opportunities and providing a
dynamic platform for design (Patnaik & Becker, 2010). By identifying the needs
of the clients and stakeholders, one ends up defining the goals of the system. Usability and user experience goals help produce a fine-tuned product. These goals
become the centerpiece for the core activities of interaction design: needfinding,
conceptual and physical models, interactive models and evaluation.
However, the above framework does not take into account the social dimension found in social media software. While user experience goals describe certain psychological aspects for an interaction between a human and a system
(e.g., computer, software, video game) it does not take into account the interconnectivity that exists between different users through social media. Individuals
are interacting with the software as much as they interact with each other through
it. Through these interactions behaviors emerge. As demonstrated by this study,
social behaviors can be altered by the software design in a proactive manner.
According to the above, I provide a new definition for social interaction design, which in light of this evidence, forms the base for the framework presented
in this chapter.
Social interaction design is a set of principles, models, methodologies
and other aids that are used in the proactive design and development
of systems which involve social environments, in order to satisfy a
system’s social experience goals. In turn, social experience goals are
a set of goals derived from the social dimension of a system and describe social behaviors, interactions, attitudes and phenomena, which
are identified by designers in order to meet system and user explicit
as well as implicit overall goals.
I proceed by presenting the social interaction design framework that builds
upon the current interaction design framework. As such, all current models that
are currently in use by interaction designers can still be used in this version of
interaction design that encompasses social interaction design. To achieve this I
developed an example that will demonstrate how goals for phenomena such as
groupthink and aggression, can determine social behavior and should concern
the decisions that designers make about features in social media. Below, I present
the example that is going to be used to demonstrate the practical aspects of the
framework.
A government organization is asked to develop a social media platform called UrCity to encourage an active involvement of people in
their communities. They want to empower communities, states and
104
the country as a whole to understand issues, find solutions and help
notify authorities faster about these problems. Additionally, this project
helps to also evaluate the efficiency of certain authorities. Authorities
that are not willing to listen get a lower rating based on how long it
takes them to resolve a problem. The public not only has the power
to report and find solutions for problems, but also has the power to
assess whether high rank officials are effectively doing their jobs.
There are many important issues that need to be resolved from a software
engineering and design standpoint but for the purposes of demonstrating how
social behavior can be engineered through the design of software, I focus only on
the social interaction design aspects. I attempt to demonstrate three major activities of interaction design and how they relate to the design of social experiences:
(a) needfinding, (b) prototyping, conceptual and physical models, and (c) evaluation.
7.2 Needfinding: Identifying Social Experience Goals
During needfinding designers should make sure that they understand all of the
social needs and dangers of the project. These are social behaviors of users while
interacting with others and can have an impact on the general mission of the
product. The process of identifying requirements in a project in terms of social
needs does not significantly differ from the process used to identify other functional or non-functional requirements of the project. The same strategies can be
used to account for the social dimension. Rogers et al. (2011) have discussed that
identifying environmental requirements for the social aspects of the environment
is particularly important. For example, one has to think about whether communication will be synchronous or asynchronous, or if all data is going to be shared.
However, this can be taken a step further. A general identification of requirements does not need to be confined in terms of how people will communicate or
collaborate. Social goals can also define social behaviors, interactions, attitudes
and phenomena in social media. These behaviors can later be classified as desirable, undesirable or neutral. This classification is not something to be taken
lightly. Since many social behaviors may come in conflict with each other one has
to determine the desirable behaviors that are necessary for an efficient interaction
between users and those that are undesirable in the eyes of a designer. Additionally, there may be behaviors classified as neutral. Finally, all these goals can also
be assigned different severity.
I define the new set of goals as social experience goals. These goals aim to
identify all aspects of social interaction, behavior, attitudes and phenomena between humans through the system that is to be designed. Humans are social creatures by nature and as argued earlier on in this dissertation, sociological needs are
the force that drives technological development. It therefore becomes appropriate
that such needs will finally be taken into account in the design process of social
media software.
Designers can set these goals based on user needs, which is a fundamental basis for interaction design work. However, identifying user needs is not sufficient
105
for identifying social experience goals. For example, most users will vouch for an
aggression-free environment however, few if any, will actually mention groupthink as an undesired behavior. Groupthink is a goal that is likely to be identified
only through the eyes of stakeholders or the designers. Therefore, social interaction design takes into account both user requirements as well as system requirements and expects that these will be adapted in order to complete each other.
The approach for classifying social experience goals is also important. While
everyone can agree on some nearly universal set of usability and user experience
goals, social experience goals are context-dependent to the intent of the software.
As an example, this paper has treated aggression as a negative aspect on social
media communities. On the other hand, Wallace (2001) has also pointed out that
there is a belief by some that venting out anger may provide them with catharsis.
While this remains a controversial topic, it is still worth noting that it may be an
option for a software designer (e.g., one that may be asked to develop a social
media platform where people come to vent their anger towards others). Finally,
there are certain social phenomena that exist in social media but may be of no
interest for the software designers. For example, one may notice that people exhibit signs of altruism online, but the community would not benefit, nor would
be affected negatively by such behavior. Or, it may be the case that such behavior
is acceptable at moderate levels while if it falls or rises above a certain threshold
it becomes either desirable or undesirable.
This process of identifying desirable, undesirable and neutral social experience goals will help designers in their further development of designs. The result
should help produce a comprehensive list that would cover most of the important social experience goals that are of direct interest and concern for designers.
Additionally, these will be classified under certain categories that better help designers address cost and benefit issues. Finally, each one of these goals can also
be assigned a severity to establish priorities.
Based on the above, I present a revised diagram describing usability, user experience goals and social experience goals based on the original diagram created
by Preece, Rogers, and Sharp (2002); Rogers et al. (2011). The new model still has
usability goals as a centerpiece, while user experience goals remain in the outer
rim. However, I added an extra layer that describes social experience goals that
builds on top of the other set of goals. Additionally, while it may seem that the
most important goals for a product are usability goals, that should not be the case.
User experience and social experience goals are just as important for avoiding
faulty designs and costly mistakes. For example one can build a highly efficient
product that does not provide an enjoyable experience, or an efficient product
that however leads to groupthink behavior. While the diagram is designed in
such a way that will help align the priorities for the design and development
teams, it is important to note that all goals are important. Through the iterative
process of design, implementation and evaluation, the team should constantly
aim to assess if goals are satisfied and to what degree.
Additionally, this model also demonstrates how important is that these goals
be identified early on in a project’s lifecycle. The cost for changes in the system
increases over time. This may be especially true for social experience goals where
106
Figure 7.1: Social Interaction Design Goals adapted from the Interaction Design
goals developed by (Rogers et al., 2011)
they need to be identified correctly and early on. As an example, a shift for a
groupthink goal from undesirable to desirable is easier to achieve early on in the
design stages, than later on during the development stages. This is an important
difference between social experience goals and user experience or usability goals.
While in the latter, changes may not be costly, the former will produce significantly more costs for changes later on in a project’s lifecycle.
To identify social experience goals one can use a variety of methods. Some
methods can be on-site observation, interviews and participant observation that
can help develop different perspectives for software needs. Questionnaires and
focus groups can also be an effective way (Rogers et al., 2011). Most methods can
be efficient for identifying social experience goals, however there is a tremendous
potential for methods that will produce real world data. As such, participant and
on-site observations may be more ideal in determining what social behaviors occur, how severe they are, and whether they should be classified as desirable, undesirable or neutral.
Having identified social experience goals, these can further be used in the development of personas. Personas are highly valuable and popular in interaction
design because of their ability to demonstrate behaviors (D. M. Brown, 2010;
Cooper, Reimann, & Cronin, 2012). However, personas do not have to be lim107
ited in devising scenarios between a single user and a system. These can also be
used for evaluating designs by creating scenarios between personas in order to
demonstrate a social behavior of interest and how this can be satisfied by features. Put simply, multiple personas can be used as puppets in social scenarios
to demonstrate how they exhibit social behaviors through a system and how the
system satisfies social experience goals.
7.2.1 Example: Social Experience Goals in UrCity
In the example of the UrCity project, one can identify two separate behaviors.
People are expected to collaborate with others about various topics. Some of them
may be controversial topics and disagreements may quickly evolve into flaming
wars. As such, the first identified social experience goal is aggression. Furthermore, one can imagine that since users will have to discuss problems as well as
propose their own solutions and vote for them, one has to evaluate how effective
is the decision-making process. Groupthink is a serious phenomenon that can affect decision-making and has been the cause of many historical fiascoes (Rose,
2011). As such, a second social experience goal is groupthink.
The second step involves classifying these behaviors. Both are classified as undesirable based on the descriptions of the previous paragraph. These two however, do not necessarily have the same severity. When having wicked problems
one can expect flaming wars in social media software. Low levels may be considered acceptable. On the other hand, groupthink is a problematic issue that can
affect the efficiency of the decision-making process and as such it should be assigned a high severity. Through needfinding (which is an iterative process in its
nature) one not only identifies the social experience goals derived by desirable,
undesirable or neutral social behaviors but also assigns severity to each goal that
will significantly affect the decisions that will be later taken in to the design.
7.3 Prototyping, Conceptual Models and Physical Models
Storyboarding, sketching, or wizard of oz prototyping are great ways for showing
how social experience goals are satisfied by various models. They also put more
focus on the needs rather than on the design of the product itself which is particularly helpful because of the complex nature of social behaviors. Identifying
opportunities for alternative designs can be helpful when creating low-fidelity
prototyping. It also helps to put into perspective, when there is an opportunity
for a new design as well as to establish how interactive versions of the design will
affect users.
Early on, ideas for prototypes that are expected to have an effect towards satisfying certain goals can be taken from published research studies or a designer’s
expert knowledge. In many cases, such research may require the development
of high-fidelity prototypes which can be more expensive. It becomes essential
that designers have exhausted all possible options and the designs selected to be
evaluated are the ones suspected to be more efficient. Additionally, when conceptual designs are created it may be beneficial to develop plus or minus scenarios
108
(Rogers et al., 2011, p. 292). This would help develop the most positive and most
negative results of a design on social experiences. Such an approach can also help
see if there is a balance between social experience goals and if in general they are
in accordance with the product’s mission goal and user social needs.
When designing a conceptual model that incorporates social goals, a use of
multiple methods such as sketching out ideas, storyboarding, describing possible scenarios, and prototyping aspects of the proposed behavior of the system is
important. This should reflect scenarios where the social needs can be put to the
test. For example, a storyboard can demonstrate a frustrated user using the software to send a message to all users. Prototyping or video prototyping can then
show how versions of the design can have a preventive effect on these interactions. New opportunities for exploration may be discovered during this process.
Wizard of oz prototyping can also be extremely helpful in observing social behavior while having individuals within the same observable space.
Transitioning between a physical design and a conceptual design does not
produce additional challenges in respect to social experience goals. Some social
experience goals may be able to be satisfied early on when prototyping while for
other goals a designer may be able to identify opportunities later on when dealing
with the physical design. For example, affective computing research has shown
that the interface can have a positive effect on reducing frustration (Prendinger &
Ishizuka, 2005) and in turn aggression in social media. Certain features can have
a positive effect on individuals but only be considered when dealing with the
actual physical design. At this point, many different approaches can be invented
so that specific goals such as the one provided in this example can be satisfied.
When it is time for interactive versions of the design to be developed, one
must have a relative certainty that social experience goals are likely to be satisfied.
A theoretically grounded approach is essential. Features that need to be changed
at this point due to the violation of these goals can be more costly for the design
team. For example, imagine designing a social media software platform meant
to be used by a community for anger management. One can start creating conceptual models, create storyboard and prototypes. One of the social experience
goals classified with high severity as undesirable is aggression. However, during
the prototyping phase, one has not considered thoroughly that anonymity states
may affect inhibition. Later on, once an interactive design has been created the
opportunity of using various anonymity states and the effects on aggression is
identified. However, at this later point in the design, most of the features of the
system that relate to the anonymity state used need to be reevaluated in accordance to the usability, user experience and social experience goals.
A final consideration involves the long established use of metaphors in design. While useful in interaction design it may have a limited power for developing new paradigms. Preece et al. (2002) have described that it limits “the designer’s imagination in conjuring up new paradigms and models” (p. 59). Designing social experiences in an online world is an opportunity in itself to achieve
things that could not be achieved otherwise. For example, using masks to hide
one’s anonymity is not the same with being online and use pseudonyms or even
be completely anonymous. Being restricted to having new ideas may not be ben109
eficial for identifying features that can assist in designing social interaction.
7.3.1 Example: Prototyping and Modeling for UrCity
Since we identified the main goals, a better understanding of the factors involved
is necessary. After having read the literature on the topic one can start developing
ideas that will satisfy these goals. Some may be concrete such as the randomization of the solutions for each user while others may not be as easy to determine.
For example, one may suspect that anonymity states can affect aggressive behavior and as such create various designs that display different anonymity states and
the way individuals communicate.
In early designs, several versions were created in using different anonymity
states and their combinations. It was suspected that anonymity had a tremendous
effect on aggression but also on groupthink by allowing users to discuss alternatives. Another suspicious feature was the dynamic voting indicator that may
affect voting and be a factor for groupthink. Further, a feature that was added
was a pro/con list representation of arguments for each solution. It was thought
that potentially could be better than textual collaborative interface for increasing
alternatives and reducing groupthink. Notes were sent to software developers
to ensure that information submitted through forms is not lost and also provide
alternatives for any errors that may appear in the system. Designs were created
for having dynamic voting indicators for the voting processes for solutions. Furthermore, two separate designs were created in which leaders could be identified to vote publicly, whether they are administrators or moderators. Finally, two
designs where created, displaying solutions in a randomized and standardized
manner.
At this early stage, most of these design features have neutral values for satisfying the social experience goals for the UrCity project. Early prototypes are
helpful and will reduce the chances of having to redesign costly elements. Several evaluations of the early designs will help to move toward the direction of
developing interactive versions that can formally be evaluated and establish on
whether features satisfy social experience goals.
7.4 Evaluating Social Interaction Designs
Evaluation can be used throughout the previously described activities. Evaluations were defined by Rogers et al. (2011) as formative and summative. These
evaluations can help assess that the needs are satisfied throughout the course of
the design. The same applies for social experience goals. Evaluation is important
so that designers will know that they are on the right track. Evaluation of social
behavior while can be more complex, it does not significantly differ from evaluation for usability goals or user experience goals.
Evaluation can happen in a number of ways. Rogers et al. (2011) suggested a
couple of evaluation paradigms such as “quick and dirty,” usability testing, field
studies, predictive. When exploring social experience goals most approaches can
be efficient. For example, the "quick and dirty" paradigm was used in this dis110
sertation to compare argument production for pro/con list interfaces and textual
collaborative interfaces. Usability testing was also used extensively but aimed not
to measure usability goals but social behavior goals. Field studies can also be an
efficient way for getting in touch with the users and finding out how the design
may affect behavior. However, this technique limits designs which affect individuals without the individuals realizing that this is happening; something that has
been observed during the course of this dissertation research. Moreover, the predictive paradigm, while effective for usability goals can be limiting for exploring
social experience goals. What needs to be observed, is never observed and it is
assumed. However, there are certain aspects of the software that given enough
literature one can be fairly certain of the expected behavior. Additionally, there is
a need that the experts involved have a background not only in HCI, but also in
psychology and sociology. Finally, a combination of all these paradigms can also
be an efficient strategy. For example, “quick and dirty” techniques produce qualitative data that can be later transformed through content analysis to quantitative
data and be statistically evaluated.
Exploring social behavior affected by the interface can be easier said than
done. During most of my dissertation research, I decided to establish alternative designs and measure quantitatively or qualitatively specific aspects. While
some remarks were drawn from literature, in most of the cases I had to use actual
users. At times I would ask them open-ended questions so that I can get a better
understanding of how things work. But even this was not sufficient enough information to determine social behavior. For example, during pilot testing I asked
people why they behaved the way that they did and some of the responses went
along the lines of “because I felt this way.” In other words, their response stated
that the interface had no effect on the way that they reacted in certain situations
through different interface designs. However, in many cases, quantitative results
suggested that the interface had an effect on them. A good example is the research
on anonymity states and aggression. As such, if my decision would be to just ask
the users and especially with faulty yes or no designs (i.e., "Do you feel that the
interface affects your reactions") then my results would have been completely
different and inaccurate.
Finally, the evaluation of the data is also an issue that needs to be addressed.
While NHST analysis is quite popular and has been used extensively in HCI,
many argue that it is insufficient or that it may lead to misinterpreting results
(Kaptein & Robertson, 2012). Ethnographic research can also be extremely affective, but others may still call upon its reliance on a researcher’s subjectivity. There
are always benefits and tradeoffs. Alternative methods should always be considered as a researcher needs to understand the limits of an analysis method and its
implications on the final results and conclusions.
7.4.1 Example: Evaluation for UrCity’s Features of Interest
Several features were evaluated in order to establish if some of the designs could
satisfy some of the social experience goals set forth by social interaction design.
These were evaluated using several quantitative as well as qualitative methods.
Results established decisions that need to be made by designers.
111
Pseudonymity was found to play a significant role in increasing the chances
for aggression while complete anonymity did not seem to have an effect on aggression. On the other hand, complete anonymity seems to have produced the
highest potential for reducing groupthink compared to pseudonymity. However,
from a practical perspective, designing UrCity in a way that everyone will be
anonymous is not feasible for practical reasons. As such, a decision was made for
users to appear with their real names, but the option of complete anonymity will
also be provided when posting messages.
The feature of dynamic voting did not produce significant differences for results (Tsikerdekis, in press-a). As such it is rendered as safe and users will be
allowed to participate publicly in voting processes using this mechanism. It has
a limited effect on groupthink.
Pro/con lists will be employed for all collaborative sessions on the software as
they were found to increase unique arguments. However, an additional decision
was made to use a textual collaborative interface for a narrative and discussion
on each topic along side with pro/con lists. As such, it was expected that more
alternatives would be proposed and more information would be produced for
users to make better judgments on the topics.
Error messages were designed to be exact, specific and short. Additionally, alternative solutions were designed for finding the information faster and had even
provided options for automatically looking at Internet archives to find information that may be lost. An additional option was given to users to ask for help from
the community on the forums of the website. These elements were identified as
critical for achieving a higher comprehension and helpfulness for error messages,
therefore decreasing unnecessary frustration for users.
Steps have also been taken to ensure that messages are not lost due to submission errors. Analysis has found that lost messages can critically affect the efficiency of communication on the community and as such, be the cause for implicatures and misunderstandings. Taking measures to always try and attempt to
store information in forms on the server will significantly improve both user and
social experience.
Solution lists for all the voting mechanisms will use algorithms that randomize them for every user. While an alternative design was made for delivering solutions in a stratified manner, the nature of the project did not necessitate a post-hoc
evaluation of voting procedures. Therefore, the chosen design was rendered an
acceptable choice.
Finally, some of the designs had administrators, moderators and community
organizers distinguished graphically from the rest of the users. Since people can
make their votes public if they wish to vote using their real names, people holding such positions in the community will be distinguished from the rest of the
voters. According to the evaluation of these designs, no effect was found by leaders influencing individual votes especially when all information for the topic is
available to users. As such, it was concluded that this is a good feature in the
design and would likely not affect decision-making processes, decision outcomes
and be the cause for groupthink.
112
7.5 Additional Aspects for Consideration
7.5.1 Social Cognition
Cognition is an important part for interaction design and the creation of mental
models. Rogers et al. (2011) have shown the importance of having good knowledge of all the processes involved with cognition such as attention, perception
and recognition, memory, learning, reading, speaking, listening, problem solving, planning, reasoning and decision-making. This knowledge of the physical
world can be applied in the design of the digital world.
However, cognition cannot sufficiently help a designer understand its users in
terms of social experience goals. For this, social cognition needs to be included in
the work of interaction designers, which is used in social psychology to explain
attitudes, attribution and groups dynamics. It “broadly includes the cognitive
processes used to decode and encode the social world” (Beer & Ochsner, 2006, p.
98). Many have studied social cognition and identified its components (Adolphs,
Tranel, & Damasio, 2001; Kihlstrom & Cantor, 2000; Kihlstrom & Klein, 1994).
Beer and Ochsner (2006) used all this information to develop a better definition
for social cognition. They stated that, social cognition describes perception of others, perception of self and interpersonal knowledge. Social stimuli are perceived
in various degrees of complexity and along with contextual knowledge people
conjure representations of responses to a situation (Beer & Ochsner, 2006). As
such, the way people respond in varying settings in social media is not dependent
only on a single individual, but also on the others involved in the “immediate”
surroundings as well as the context in which a situation unfolds.
Social cognition can help create mental models or information processing models the same why that cognition is used. In the example presented here, social
cognition processes can help identify gaps in the design that relate to undesired
goals. For example, highly controversial topics are known to attract people coming from the opposing sides of debates and that may be biased towards others.
Beer and Ochsner (2006) have discussed that in the process of perceiving others,
biases due to stereotyping may lead to people seeing things that are not present
in the behavior of others. Consequently, this is also similar to the solipsistic introjection described by Suler (2012). The result in circumstances where bias is
introduced, may be a well-founded anticipated aggressive collision. This realization can help create mental models that a designer can employ for finding ways
around such problems. In this case, attempts to reduce the bias regarding perceptions of others may be the best course of action.
7.5.2 Distributed Cognition Framework
An existing effective framework for interaction design when dealing with collaboration and communication that deals with the interaction of multiple individuals is the distributed cognition framework. This mainly investigates how individuals pass information through media (Rogers et al., 2011). However, it can also be
expanded to incorporate and take into account social behavior. This way, the emphasis can be placed on the social aspects of cognition as well. Rogers et al. (2011)
113
have described that distributed cognition involves examining several aspects of
a system and its individuals. I further expand the list of aims by adding one additional aspect worthy of examination that could help interaction designers take
into account social interactions and behaviors. A socially distributed cognition
framework should aim to examine:
•
The distributed collaboration or communication that takes place
(including the way people work together to solve a problem).
•
The role of verbal and non-verbal behavior.
•
The various coordinating mechanisms that are used (e.g., rules,
procedures).
•
The various communicative pathways that take place as an activity progresses.
•
How knowledge is shared and accessed.
•
How the behavior of individuals affected by the design as well
as exterior factors (e.g., the social environment) could influence
the processes in a system.
While using such a framework, a designer can make use of storyboards and
prototypes to display behavior relative to the social experience goals. For example, according to an undesirable groupthink goal, participants willing to follow a
majority opinion due to compliance pressures can be seen following a leader’s
opinion. This can be implemented as a behavior on the distributed cognition
model. In turn, a designer can identify issues on the outcome of the process violating the goal and attempt to find ways to minimize the impact of such a behavior.
A way for working around the issue may be for the program to ask participants
if they have enough information or if they would like to go through the information one more time before voting. Representation of arguments can also prove to
be an important way for inhibiting such a behavior.
114
C HAPTER 8
C ONCLUSION
8.1 Reflecting on the Dissertation
The work conducted in this dissertation and further work associated with it, has
shown great promise in not only answering key questions discussed earlier on in
chapter 1, but has also revealed how potentially vast the effect of software design
is in social media. Many social media features have been found to have an effect
on aggression such as, (a) anonymity states and specifically pseudonymity being
a factor for aggression (Tsikerdekis, 2011, 2012), (b) message submission errors
could severely impact communication, be the cause of implicatures and potentially lead to aggression, (c) quality of error messages in terms of helpfulness and
comprehension can reduce frustration for users and in turn aggression. Similarly,
features have been found to affect the potential for developing groupthink such
as, (a) anonymity states empower users to contribute their alternative solutions to
a discussion (Tsikerdekis, in press-b), (b) pro/con lists produce more unique arguments as compared to textual collaborative interfaces (Tsikerdekis, in press-c),
(c) publicly identified leaders do not affect decision-making voting processes, (d)
item randomization in voting processes has great potential for reducing groupthink as opposed to standardization, and (e) dynamic voting status indicators do
not affect voting outcomes (Tsikerdekis, in press-a). These results were put into
developing what is now social interaction design for social media and how the
framework provided can ensure the development of social media services that
satisfy social experience goals. Through the work of this dissertation interaction
design has been reevaluated as a field that should incorporate the social dimension of social systems. Finally, the process of actively researching and developing
social interaction design for this dissertation has clearly helped to identify some
of its core elements that can positively impact the process and make it successful.
I identify these as, (a) contextual awareness, (b) multidimensional apprehension,
(c) creativity, (d) scientific application, and (e) multimethodology.
A broad variety of multiple social behaviors, attitudes, interactions and phenomena are affected by system features choices when designing and developing social media services. Since social media services describe multiple classifications, this calls for social interaction designers to be contextually aware and
have a good understanding of the system as well as its social dimension and user
115
8. C ONCLUSION
population. Further, it is now evident that many system elements can potentially
impact not only one but multiple social behaviors, attitudes, interactions and phenomena on social media. A great example for this is the evidence provided by this
dissertation work to demonstrate that the choice of anonymity states can impact
two major social phenomena, aggression and groupthink. Hence, social interaction design not only requires a process necessitating contextual awareness but
also multidimensional apprehension. Additionally, as it is common in processes
involving design, creativity is an important factor for identifying social software
features that impact the way users communicate, collaborate and interact with
one another. For example, one not only needs to be creative in the way information disseminates and reaches individuals when designing social media services,
but also to understand how this information is perceived and what system factors may affect this process. Because the process where information is perceived
by users is invisible to the naked eye, creativity not only requires production of
diverse alternatives but also reaching for ideas “out of the box.” Moreover, as this
dissertation has shown, intuitive predictions for an effect rarely coincide with
actual data. The complexity of sociotechnical systems is such that only through
scientific application one can reveal the true effects hidden between interactions.
However, scientific application should not be confined in a narrow set of methodologies and methods. A multimethodology approach is required to provide social interaction designers with a deeper understanding of how elements affect
processes and outcomes as well as answering the questions as to why that is.
The benefits for this can be seen in many of the social media elements that were
investigated using a variety of methods in this dissertation.
8.2 Looking into the Future
In a broader perspective, in this research I attempted to successfully demonstrate
that digital environments just as environments in the real world affect human
behavior. The results have shown that not only this effect exists but also that it has
drastic implications on the interaction of people that are online. Through social
interaction design one can engineer these interactions. Furthermore, the idea of
increasing collaboration and decreasing conflict sets a standard for how future
online interactions are going to evolve.
About a decade ago, Zlatuška (1997) talked about the nature of Information
Society in terms of a synergy between technological and social changes. Today,
these changes are here but they are not over yet. Society is still facing some anxiety because of these changes, the infinite possibilities and the fear of the unknown. What we do in the present could ripple throughout our future. The behavioral phenomena studied in this dissertation are not a looming threat for future generations. They have been part of our societies long before the technological advances that we came to see the past decade. The threat to our future is simply ignorance. Ignorance in not identifying the opportunities that are presented
by Information Technologies, as well as the conscious ignorance for not trying
to improve the new virtual reality that we create. As Neil deGrasse Tyson put it,
“Where Ignorance lurks, so too do the frontiers of discovery and imagination.”
116
8. C ONCLUSION
In a future not far from now, HCI specialists will have at their disposal, a
database which describes system features and their effects on social interaction,
attitudes, behaviors, and phenomena. Decisions will be made not in terms of
qualitative likelihood for satisfying goals, but pure probabilities. In a future world,
people will gather at a virtual spot to discuss matters and the software will ensure that an effective decision will take place. Frustrated users may be distanced
from the group or become distracted so that the group can work on their collaborative task. Users which may have a weak understanding of a topic will be
detected and assisted in order to help lead to a more unbiased voting process. The
software will automatically adjust its features, so that all important discussion elements will be heard and evaluated by all users. The experience will be tailored
to every individual’s social, psychological and physiological state.
It may sound farfetched, but Clarke’s three laws for predicting the future have
so far indicated otherwise (Clarke, 1962). Venturing through the impossible can
help one achieve the greatest breakthroughs to the point of advanced technology
becoming indistinguishable from magic. As we move our lives more into the digital world, decisions on how this world is designed affects the new reality which
we will create for ourselves. Research defines what it is to be human, how we
interact in the real world and how different we interact online. As the consciousness of a future generation moves to an online world, they can always look back
at how things used to be once upon a time ago. If we understand enough about
our interactions today, this online consciousness may end up living in a safer and
more effective virtual world. A digital world that serves humanity’s needs better.
A social world built by humans, meant for humans, and designed for humans!
117
R EFERENCES
Abelson, R. P. (1981). Psychological status of the script concept. American
Psychologist , 36 (7), 715–729. Retrieved from http://www.bm.ust.hk/
~mark551/readings/T3cR6_abelson1981.pdf
Adolphs, R., Tranel, D., & Damasio, H. (2001). Emotion recognition from
faces and prosody following temporal lobectomy. Neuropsychology,
15 (3), 396–404. Retrieved from http://www.scopus.com/inward/
record.url?eid=2-s2.0-0034820118&partnerID=40&md5=
9d922ff2d0ba73fcbc014e12d0c4112c
Ahlfinger, N. R., & Esser, J. K. (2001). Testing the groupthink model: Effects
of promotional leadership and conformity predisposition. Social Behavior and Personality an international journal , 29 (1), 31–41. Retrieved
from
http://openurl.ingenta.com/content/xref?genre=
article&issn=0301-2212&volume=29&issue=1&spage=31
doi:
10.2224/sbp.2001.29.1.31
Ahlstrom, D., & Wang, L. C.
(2009).
Groupthink and France’s defeat
in the 1940 campaign. Journal of Management History, 15 (2), 159–
177. Retrieved from http://www.emeraldinsight.com/10.1108/
17511340910943804 doi: 10.1108/17511340910943804
Ainsworth, S., Gelmini-Hornsby, G., Threapleton, K., Crook, C., O’Malley, C., &
Buda, M. (2011). Anonymity in classroomvoting and debating. Learning
and Instruction, 21 (3), 365–378. doi: 10.1016/j.learninstruc.2010.05.001
Albert, J. (2009). Bayesian Computation with R (No. 3). New York, NY:
Springer. Retrieved from http://www.springerlink.com/index/10
.1007/978-0-387-92298-0 doi: 10.1007/978-0-387-92298-0
Anderson, C. A. (1983). Imagination and Expectation : The Effect of Imagining
Behavioral Scripts on Personal Intentions. Journal of Personality and Social
Psychology, 45 (2), 293–305. Retrieved from http://content.apa.org/
journals/psp/45/2/293 doi: 10.1037/0022-3514.45.2.293
Anderson, C. A., Berkowitz, L., Donnerstein, E., Huesmann, L. R., Johnson,
J. D., Linz, D., . . . Wartella, E. (2003). The influence of media violence on youth. Psychological Science in the Public Interest , 4 (3), 81–110.
Retrieved from http://psi.sagepub.com/lookup/doi/10.1111/j
.1529-1006.2003.pspi_1433.x doi: 10.1111/j.1529-1006.2003.pspi\
_1433.x
118
8. C ONCLUSION
Anderson, C. A., Buckley, K. E., & Carnagey, N. L. (2008). Creating your own hostile environment: a laboratory examination of trait aggressiveness and the
violence escalation cycle. Personality and Social Psychology Bulletin, 34 (4),
462–473. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/
18340032
Anderson, C. A., & Bushman, B. J. (2002, November). HUMAN AGGRESSION.
Annual Review of Psychology, 53 (1), 27–51. doi: 10.1146/annurev.psych
.53.100901.135231
Anderson, C. A., Bushman, B. J., & Groom, R. W. (1997). Hot years and serious
and deadly assault: Empirical tests of the heat hypothesis. Journal of
Personality and Social Psychology, 73 (6), 1213–1223. Retrieved from
http://psycnet.apa.org/index.cfm?fa=buy.optionToBuy&id=
1997-42257-006%&CFID=6830969&CFTOKEN=51209674
doi:
10.1037/0022-3514.73.6.1213
Anderson, C. A., & Godfrey, S. S. (1987). Thoughts about actions: The effects
of specificity and availability of imagined behavioral scripts on expectations about oneself and others. Social Cognition, 5 (3), 238–258. Retrieved
from http://dx.doi.org/10.1007/978-0-387-74118-5_25 doi:
10.1007/978-0-387-74118-5\_25
APA. (2004). Violence in the Media - Psychologists Help Protect Children from
Harmful Effects. Retrieved May 19, 2012, from http://www.apa.org/
research/action/protect.aspx
Archer, J. (2009). Does sexual selection explain human sex differences in aggression? Behavioral and Brain Sciences, 32 (3-4), 249–66; discussion 266–311.
Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/19691899
doi: 10.1017/S0140525X09990951
Aronson, E., Wilson, T. D., Akert, R. D., & Akert, R. M. (2009). Social Psychology.
Prentice Hall.
Asch, S. E. (1958). Effects of Group Pressure upon the Modification and Distortion
of Judgments. In E. E. Macoby, T. M. Newcomb, & E. L. Hartley (Eds.),
Readings in social psychology (3rd ed., pp. 174–183). New York, NY: Holt,
Rinehart, and Winston.
Asch, S. E. (1992, November). Opinions and Social Pressure. In E. Aronson (Ed.),
Readings about the social animal (6th ed., pp. 13–22). San Fransisco, CA:
Freeman.
Atkinson, R. (2008). Societal effects of cyberbullying: the dark side of building
bridges with technology. In South atlantic philoshopy of education society
yearbook. (p. 0).
119
8. C ONCLUSION
Bainbridge, W. S. (2007). The scientific research potential of virtual worlds.
Science, 317 (5837), 472–6. Retrieved from http://www.ncbi.nlm.nih
.gov/pubmed/17656715 doi: 10.1126/science.1146930
Bandura, A. (1978). Social learning theory of aggression. The Journal of communication, 28 (3), 12–29. Retrieved from http://www.ncbi.nlm.nih
.gov/pubmed/690254
Bandura, A. (2001a). Social cognitive theory: an agentic perspective. Annual Review of Psychology, 52 (1), 1–26. Retrieved from http://www.ncbi.nlm
.nih.gov/pubmed/11148297
Bandura, A.
(2001b).
Social Cognitive Theory of Mass Communication.
Media Psychology, 3 (3), 265–299.
Retrieved from
http://www.informaworld.com/openurl?genre=article&doi=
10.1207/S1532785XMEP0303_03&magic=crossref
doi:
10.1207/S1532785XMEP0303\_03
Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Mechanisms
of moral disengagement in the exercise of moral agency. Journal of Personality and Social Psychology, 71 (2), 364–374. Retrieved from http://doi
.apa.org/getdoi.cfm?doi=10.1037/0022-3514.71.2.364 doi:
10.1037//0022-3514.71.2.364
Barnes, S. B. (2006). A privacy paradox: Social networking in the United States.
First Monday, 11 (9), 11–15. Retrieved from http://firstmonday.org/
issues/issue11_9/barnes/index.html
Baron, R. A., & Richardson, D. R. (2004). Human Aggression. Springer.
Bateson, P. P. G., & Klopfer, P. H. (1989). Whither Ethology?, Volume 8. Springer.
Beer, J. S., & Ochsner, K. N. (2006). Social cognition: A multi level analysis. Brain
Research , 1079 (1), 98–105. doi: 10.1016/j.brainres.2006.01.002
Benzeghiba, M., Demori, R., Deroo, O., Dupont, S., Erbes, T., Jouvet, D.,
. . . Ris, C. (2007). Automatic speech recognition and speech variability: A review. Speech Communication, 49 (10-11), 763–786. Retrieved from http://linkinghub.elsevier.com/retrieve/pii/
S0167639307000404 doi: 10.1016/j.specom.2007.02.006
Berkowitz, L. (1989). Frustration-aggression hypothesis: examination and reformulation. Psychological Bulletin, 106 (1), 59–73. Retrieved from http://
www.ncbi.nlm.nih.gov/pubmed/2667009
Berkowitz, L. (1990). On the formation and regulation of anger and aggression.
A cognitive-neoassociationistic analysis. American Psychologist , 45 (4),
120
8. C ONCLUSION
2186678 doi: 10.1037/0003-066X.45.4.494
Berkowitz, L. (1993a). Aggression: Its causes, consequences, and control. Mc
Graw Hill.
Berkowitz, L. (1993b). Pain and aggression: Some findings and implications.
Motivation and Emotion, 17 (3), 277–293. Retrieved from http://www
.springerlink.com/index/10.1007/BF00992223
doi: 10.1007/
BF00992223
Berkowitz, L., & LePage, A. (1967). Weapons as aggression-eliciting stimuli.
Journal of Personality and Social Psychology, 7 (2p1), 202–207. Retrieved
from http://psycnet.apa.org/journals/psp/7/2p1/202/ doi:
10.1037/h0025008
Bernal, J. (2009). Web 2.0 and Social Networking for the Enterprise: Guidelines
and Examples for Implementation and Management Within Your Organization. Pearson Education.
Bishop, G. F., & Smith, A. E. (1997). RESPONSE-ORDER EFFECTS IN PUBLIC
OPINION SURVEYS. In annual conference of the american association for
public opinion research norfolk va also published in american statistical association the (pp. 1041–1046). Retrieved from http://www.amstat.org/
Sections/Srms/Proceedings/papers/1997_179.pdf
Björkqvist, K., Österman, K., & Lagerspetz, K. M. J. (1994). Sex differences
in covert aggression among adults. Aggressive Behavior, 20 (1), 27–33.
Retrieved from http://www3.interscience.wiley.com/journal/
112595147/abstract doi: 10.1002/ab
Bjtirkqvist, K. (1994). Sex differences in physical, verbal, and indirect aggression: A review of recent research.
Sex Roles, 30 (3-4), 177–188.
Retrieved from http://www.springerlink.com/index/10.1007/
BF01420988 doi: 10.1007/BF01420988
Black, P., Delaney, H., & Fitzgerald, B. (2007). Legal Issues for Wikis: The Challenge of User-Generated and Peer-Produces Knowledge, Content and Culture. eLaw Journal , 14 .
Bongers, I. L., Koot, H. M., Van Der Ende, J., & Verhulst, F. C. (2008). Predicting
young adult social functioning from developmental trajectories of externalizing behaviour. Psychological Medicine, 38 (7), 989–999. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/18047767
Bono, J., & Ilies, R. (2006). Charisma, positive emotions and mood contagion.
The Leadership Quarterly, 17 (4), 317–334. Retrieved from http://
linkinghub.elsevier.com/retrieve/pii/S104898430600035X
doi: 10.1016/j.leaqua.2006.04.008
121
8. C ONCLUSION
Boulos, M. N. K., Maramba, I., & Wheeler, S. (2006, January). Wikis, blogs and
podcasts: a new generation of Web-based tools for virtual collaborative clinical practice and education. BMC medical education, 6 (1), 41. Retrieved
from http://www.pubmedcentral.nih.gov/articlerender.fcgi
?artid=1564136&tool=pmcentrez&rendertype=abstract doi: 10
.1186/1472-6920-6-41
Bowles, C., & Box, J. (2011). Undercover User Experience Design. Berkeley, CA:
New Riders.
Boyd, D. M. (2003). Reflections on Friendster , Trust and Intimacy. In Ubiquitous
computing (ubicomp 2003), workshop application for the intimate ubiquitous computing workshop.
Boyd, D. M. (2004). Friendster and publicly articulated social networking. In Chi
’04 extended abstracts on human factors in computing systems (Vol. 2, pp.
1279—-1282). New York, NY, USA: ACM Press. Retrieved from http://
portal.acm.org/citation.cfm?doid=985921.986043
doi: 10
.1145/985921.986043
Boyd, D. M., & Ellison, N. B. (2007). Social Network Sites: Definition, History, Scholarship. Journal of Computer-Mediated Communication, 13 (1), 1–
19. Retrieved from http://jcmc.indiana.edu/vol13/issue1/boyd
.ellison.html
Brady, K. P., & Conn, K. (2006). Bullying Without Borders: The Rise of Cyberbullying in America’s Schools. School Business Affairs, 72 (9), 8. Retrieved from
http://www.preventionworksct.org/docs/Best_Practices
_Conference/ASBO%_Oct06_SBA_Article_Cyberbullying.pdf
Brams, S. J. (1991). Alternative Voting Systems. In L. S. Maisel (Ed.), Political pareties and elections in the united states: An encyclopedia (pp. 23–31).
New York: Garland.
Breslin, J. G., Passant, A., & Decker, S. (2009). Social Semantic Web. Springer.
Brooke, J., & Tyler, T. (2011). BOARD DIVERSITY AND CORPORATE PERFORMANCE: FILLING IN THE GAPS: DIVERSITY AND CORPORATE PERFORMANCE: A REVIEW OF THE PSYCHOLOGICAL LITERATURE. NCL
Rev., 89 , 715–1055.
Brown, D. M. (2010). Communicating Design: Developing Web Site Documentation for Design and Planning. New Riders.
Brown, J. (1997). HCI and Requirements Engineering - Exploring HumanComputer Interaction and Software Engineering Methodologies for the Creation of Interactive Software. SIGCHI Bulletin, 29 (1). Retrieved from
http://sigchi.org/bulletin/1997.1/brown.html
122
8. C ONCLUSION
Brown, J., Broderick, A. J., & Lee, N. J. (2007). Word of mouth communication within online communities: conceptualizing the online social network.
Journal of Interactive Marketing, 21 (3), 2–20. Retrieved from http://
dx.doi.org/10.1016/10.1002/dir.20082
Bryman, A. (2006). Mixed methods (No. v. 1). London: SAGE.
Bureau of Labor Statistics, U. D. o. L. (2011). Occupational Outlook Handbook,
2010-11 Edition, Computer Software Engineers and Computer Programmers. Retrieved January 26, 2012, from http://www.bls.gov/oco/
ocos303.htm
Burgess, J., & Green, J. (2009). YouTube: Online Video and Participatory Culture
(Vol. 1; J. Burgess & J. Green, Eds.) (No. 2). Polity. Retrieved from http://
www.amazon.com/dp/0745644791 doi: 10.1386/jdtv.1.1.121/4
Bush, V. (1945). As we may think. Atlantic monthly, 176 (1), 101–108.
Bushman, B., & Huesmann, L. (2001). Effects of televised violence on aggression.
In D. Singer & J. Singer (Eds.), Handbook of children and the media (pp.
223–254). Thousand Oaks, CA: Sage.
Bushman, B. J., & Anderson, C. A. (2001). Is it time to pull the plug on the hostile
versus instrumental aggression dichotomy? Psychological Review, 108 (1),
11212630
Buss, D. M., & Duntley, J. D. (2006). The evolution of aggression. In M. Schaller,
J. A. Simpson, & D. T. Kenrick (Eds.), Evolution and social psychology (pp.
263–285). Fronteers of Social Psychology.
Cairns, R. B., Cairns, B. D., Neckerman, H. J., Gest, S. D., & Gariepy, J.-l. (1988).
Social networks and aggressive behavior: Peer support or peer rejection?
Developmental Psychology, 24 (6), 815–823. Retrieved from http://doi
.apa.org/getdoi.cfm?doi=10.1037/0012-1649.24.6.815 doi:
10.1037/0012-1649.24.6.815
Camenisch, J., Bner, S. F.-H., & Rannenberg, K. (2011). Privacy and Identity
Management for Life. Springer.
Campbell, A. (1999). Staying alive: Evolution, culture, and women’s intrasexual aggression. Behavioral and Brain Sciences, 22 (02), 203–252.
Retrieved from http://www.journals.cambridge.org/abstract
_S0140525X99001818 doi: 10.1017/S0140525X99001818
Campbell, A. (2005). Aggression. In D. M. Buss (Ed.), The handbook of evolutionary psychology (pp. 628–652). John Wiley & Sons.
123
8. C ONCLUSION
Campbell, M. A. (2005). Cyber Bullying: An Old Problem in a New Guise?
Australian Journal of Guidance and Counselling, 15 (1), 68–76. Retrieved
from http://www.atypon-link.com/AAP/doi/abs/10.1375/ajgc
.15.1.68 doi: 10.1375/ajgc.15.1.68
Canny, J. (2006). The Future of Human-Computer Interaction. Queue, 4 (6), 24–32.
doi: 10.1145/1147518.1147530
Capaldi, D. M., & Crosby, L. (1997). Observed and reported psychological and
physical aggression in young, at-risk couples. Social Development , 6 (2),
184–206. Retrieved from http://doi.wiley.com/10.1111/j.1467
-9507.1997.tb00101.x doi: 10.1111/j.1467-9507.1997.tb00101.x
Card, N. A., Stucky, B. D., Sawalani, G. M., & Little, T. D. (2008). Direct and
indirect aggression during childhood and adolescence: a meta-analytic review of gender differences, intercorrelations, and relations to maladjustment. Child Development , 79 (5), 1185–1229. Retrieved from http://
Carroll, J. M. (1997). Human-computer interaction: psychology as a science
of design. Annual Review of Psychology, 48 (1), 61–83. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/15012476
Carroll, J. M. (2012). Human Computer Interaction (HCI). In M. Soegaard &
R. F. Dam (Eds.), Encyclopedia of human-computer interaction.
Castle, T., & Hensley, C. (2002). Serial killers with military experience: applying
learning theory to serial murder. International Journal of Offender Therapy
and Comparative Criminology, 46 (4), 453–465. Retrieved from http://
ijo.sagepub.com/cgi/doi/10.1177/0306624X02464007
Causes. (2012). About. Retrieved April 12, 2012, from http://www.causes
.com/about
Ceaparu, I., Lazar, J., Bessiere, K., Robinson, J., & Shneiderman, B. (2004).
Determining Causes and Severity of End-User Frustration.
International Journal of Human-Computer Interaction, 17 (3), 333–356.
Retrieved from http://www.informaworld.com/openurl?genre=
article&doi=10.1207/s15327590ijhc1703_3&magic=crossref
doi: 10.1207/s15327590ijhc1703\_3
Chan, A. (2006). Social Interaction Design Case Study : MySpace. Retrieved August 12, 2012, from http://www.gravity7.com/G7_SID
_case_myspace_v2.pdf
Chao, J. (2007). Student Project Collaboration Using Wikis. In 20th conference
on software engineering education training cseet07 (pp. 255–261). Washington, DC: IEEE. Retrieved from http://ieeexplore.ieee.org/stamp/
124
8. C ONCLUSION
stamp.jsp?tp=&arnumber=4271613 doi: 10.1109/CSEET.2007.49
Chapanis, A. (1965). Man-machine engineering. Wadsworth Pub. Co.
Chapman, J. (2006). Anxiety and defective decision making: an elaboration of the groupthink model. Management Decision, 44 (10), 1391–
00251740610715713 doi: 10.1108/00251740610715713
Chen, K.-T., Huang, C.-Y., Huang, P., & Lei, C.-L. (2006). An empirical evaluation
of TCP performance in online games. Proceedings of the 2006 ACM SIGCHI
international conference on Advances in computer entertainment technology ACE 06 , 5. Retrieved from http://portal.acm.org/citation
.cfm?doid=1178823.1178830 doi: 10.1145/1178823.1178830
Chen, Z., Lawson, R. B., Gordon, L. R., & McIntosh, B. (1996). Groupthink: Deciding with the Leader and the Devil. Psychological Record , 46 (4), 581–591.
Cheng, X., Dale, C., & Liu, J. (2008). Statistics and Social Network of
YouTube Videos. 2008 16th Interntional Workshop on Quality of Service,
229–238. Retrieved from http://ieeexplore.ieee.org/lpdocs/
epic03/wrapper.htm?arnumber=4539688 doi: 10.1109/IWQOS.2008
.32
Choi, D., & Kim, J. (2004). Why people continue to play online games: in search
of critical design factors to increase customer loyalty to online contents.
Cyberpsychology and Behavior, 7 (1), 11–24. Retrieved from http://
Clarke, A. C. (1962). Profiles of the future: an inquiry into the limits of the possible. Harper & Row.
Cogburn, D. L. (2003). HCI in the so-called developing world: what’s in it for
everyone. Interactions, 10 (2), 80–87. doi: 10.1145/637848.637866
Coleman, J. S. (1988). Social Capital in the Creation of Human Capital. American Journal of Sociology, 94 (1988), S95–S120. Retrieved from http://
www.jstor.org/stable/2780243 doi: 10.1086/228943
Cooper, A., Reimann, R., & Cronin, D. (2012). About Face 3: The Essentials of
Interaction Design. John Wiley & Sons.
Cordell, V. V. (1996, March). Application of Group Decision Support Systems in
Marketing Education. Journal of Marketing Education, 18 (1), 60–70. doi:
10.1177/027347539601800108
Cox, D. R. (1958). Planning of experiments. Wiley.
Daft, R. L., & Lengel, R. H. (1986). Organizational Information Requirements,
125
8. C ONCLUSION
Media Richness and Structural Design. Management Science, 32 (5), 554–
571. Retrieved from http://mansci.journal.informs.org/cgi/
doi/10.1287/mnsc.32.5.554 doi: 10.1287/mnsc.32.5.554
Davenport, D. (2002). Anonymity on the Internet: why the price may be too high.
Commun. ACM , 45 (4), 33–35. doi: 10.1145/505248.505267
DeSanctis, G., & Gallupe, B. (1984). Group decision support systems: a new frontier. Data Base, 16 (2), 3–10. Retrieved from http://portal.acm.org/
citation.cfm?id=1040688.1040689 doi: 10.1145/1040688.1040689
DeSanctis, G., & Gallupe, B. (1985, December). GDSS: a brief look at a new concept in decision support. In Proceedings of the twenty-first annual conference on computer personnel research - sigcpr ’85 (pp. 24–28). New York,
New York, USA: ACM Press. doi: 10.1145/16687.16690
DeWall, C. N., & Anderson, C. A. (2011). The General Aggression Model. In
M. Mikulincer & R. P. Shaver (Eds.), Understanding and reducing aggression, violence, and their consequences (pp. 15–33). Washington, DC: American Psychological Association.
DeWall, C. N., Anderson, C. A., & Bushman, B. J. (2011). The general aggression
model: Theoretical extensions to violence. Psychology of Violence, 1 (3),
245–258. Retrieved from http://doi.apa.org/getdoi.cfm?doi=10
.1037/a0023842 doi: 10.1037/a0023842
Diaper, D. (1989). Task Analysis for knowledge Descriptions (TAKD): the method
and example. In D. Diaper (Ed.), Task analysis for humancomputer interaction (pp. 108–159). E. Horwood ; Halsted Press.
Dickey, M. (2004). The impact of web-logs (blogs) on student perceptions of
isolation and alienation in a web-based distance-learning environment.
Open Learning The Journal of Open and Distance Learning, 19 (3), 279–291.
article&doi=10.1080/0268051042000280138&magic=crossref
doi: 10.1080/0268051042000280138
DiNucci, D.
(1999).
Fragmented Future.
Print , 53 (4), 32.
Retrieved
from
http://ezproxy.lib.ryerson.ca/login?url=http://
proquest.umi.com/pqdweb?did=44542139&Fmt=7&clientId=
10120&RQT=309&VName=PQD
Dollard, J. (1989). Frustration and aggression. Yale University Press.
Dollard, J., Doob, L., Miller, N., Mowrer, O., & Sears, R. (1939). Frustration and
Aggression. New Haven, CT: Yale Univ. Press.
Dooley, J. J., Pyżalski, J., & Cross, D. (2009). Cyberbullying Versus Face-to-Face
126
8. C ONCLUSION
Bullying. Zeitschrift für Psychologie Journal of Psychology, 217 (4), 182–
188. Retrieved from http://psycontent.metapress.com/openurl
.asp?genre=article&id=doi:10.1027/0044-3409.217.4.182
doi: 10.1027/0044-3409.217.4.182
Downton, A. (1993). Engineering the Human-Computer Interface. McGraw-Hill.
Driscoll, J. P., & King, J. A. (1988). An Empirical Experiment Evaluating the
Effectiveness of Group Decision Support Systems (GDSS). Master’s, Naval
Postgraduate School Monterey CA.
Duffy, P., & Bruns, A. (2006). The Use of Blogs, Wikis and RSS in Education: A
Conversation of Possibilities. In Proceedings online learning and teaching
conference 2006 (pp. 31–38). Brisbane: Queensland University of Technology.
Dummett, M. A. E. (1984). Voting Procedures. Clarendon Press.
Durrant, R.
(2011).
Collective violence: An evolutionary perspective.
Aggression and Violent Behavior, 16 (5), 428–436.
Retrieved from http://linkinghub.elsevier.com/retrieve/pii/
S1359178911000668 doi: 10.1016/j.avb.2011.04.014
Ebersbach, A., Glaser, M., Heigl, R., & Warta, A. (2008). Wiki: Web Collaboration
(Google eBook). Springer.
Ebner, M., & Schiefner, M. (2008). Microblogging - more than fun? In Procceding
of iadis mobile learning conference 2008 (pp. 155–159). IADIS.
Edwards, P.
(2007).
ness(September).
Managing Wikis in Business.
University Busi-
Ehrlich, K., & Shami, S. (2010). Microblogging inside and outside the workplace.
In Icwsm ’10: Proc. of the int. conf. on weblogs and social media. AAAI.
Ellison, N. B., Steinfield, C., & Lampe, C. (2007). The Benefits of Facebook
“Friends:” Social Capital and College Students’ Use of Online Social Network Sites. Journal of Computer-Mediated Communication, 12 (4), 1143–
1168. Retrieved from http://blackwell-synergy.com/doi/abs/
10.1111/j.1083-6101.2007.00367.x doi: 10.1111/j.1083-6101.2007
.00367.x
English, W. K., Engelbart, D. C., & Berman, M. L.
(1967).
DisplaySelection Techniques for Text Manipulation.
Ieee Transactions
On Human Factors In Electronics, HFE-8 (1), 5–15.
Retrieved
from
http://sloan.stanford.edu/MouseSite/Archive/
AugmentingHumanIntellect62/Display1967.html
doi:
10.1109/THFE.1967.232994
127
8. C ONCLUSION
Erdem, F. (2003). Optimal trust and teamwork : from groupthink to teamthink.
Work Study, 52 (5), 229–233. doi: 10.1108/00438020310485958
Esser, J. (1998). Alive and Well after 25 Years: A Review of Groupthink Research. Organizational Behavior and Human Decision Processes, 73 (2/3),
9705799 doi: 10.1006/obhd.1998.2758
Esser, J., & Lindoerfer, J. (1989). Groupthink and the space shuttle Challenger accident: toward a quantitative case analysis. Journal of Behavioural Decision
Making, 2 , 167–178.
Facebook. (2012). Statistics. Retrieved January 27, 2012, from http://www
.facebook.com/press/info.php?statistics
Farkas, C., Ziegler, G., Meretei, A., & Lörincz, A. (2002). Anonymity and
accountability in self-organizing electronic communities. In Proceedings
of the 2002 acm workshop on privacy in the electronic society (pp. 81–
90). New York, NY: ACM. Retrieved from http://portal.acm.org/
citation.cfm?doid=644527.644536 doi: 10.1145/644527.644536
Ferguson, C. J.
(2009).
An effect size primer: A guide for clinicians
and researchers. Professional Psychology Research and Practice, 40 (5),
.1037/a0015808 doi: 10.1037/a0015808
Ferguson, C. J., & Dyck, D.
(2012).
Paradigm change in aggression research: The time has come to retire the General Aggression
Model.
Aggression and Violent Behavior, 17 (3), 220–228.
Retrieved from http://www.sciencedirect.com/science/article/
pii/S1359178912000195?v=s5 doi: 10.1016/j.avb.2012.02.007
Ferguson, C. J., Rueda, S. M., Cruz, A. M., Ferguson, D. E., Fritz, S., & Smith,
S. M. (2008). Violent Video Games and Aggression: Causal Relationship or Byproduct of Family Violence and Intrinsic Violence Motivation?
Criminal Justice And Behavior, 35 (3), 311–332. Retrieved from http://
cjb.sagepub.com/cgi/doi/10.1177/0093854807311719 doi: 10
.1177/0093854807311719
Ferris, S. P., & Wilder, H. (2006). Uses and Potentials of Wikis in the Classroom.
Innovate, 2 (5), 8. Retrieved from http://www.innovateonline.info/
index.php?view=article&id=258
Field, A. (2009). Discovering Statistics Using SPSS (Vol. 58) (No. 3). London:
SAGE.
Flippen, A. R. (1999). Understanding groupthink from a self-regulatory perspective. Small Group Research , 30 (2), 139–165. Retrieved from http://
128
8. C ONCLUSION
sgr.sagepub.com/cgi/doi/10.1177/104649649903000201
10.1177/104649649903000201
doi:
Flowers, M. L. (1977). A laboratory test of some implications of Janis’s groupthink hypothesis. Journal of Personality and Social Psychology, 35 (12),
888–896. Retrieved from http://content.apa.org/journals/psp/
35/12/888 doi: 10.1037/0022-3514.35.12.888
Fodor, E. M., & Smith, T. (1982). The power motive as an influence on group
decision making. Journal of Personality and Social Psychology, 42 (1), 178–
185. Retrieved from http://content.apa.org/journals/psp/42/
1/178 doi: 10.1037/0022-3514.42.1.178
Fogel, J., & Nehmad, E. (2009). Internet social network communities: Risk taking,
trust, and privacy concerns. Computers in Human Behavior, 25 (1), 153–
160. Retrieved from http://linkinghub.elsevier.com/retrieve/
pii/S0747563208001519 doi: 10.1016/j.chb.2008.08.006
Follingstad, D. R., Rutledge, L. L., Berg, B. J., Hause, E. S., & Polek, D. S.
(1990). The role of emotional abuse in physically abusive relationships.
Journal of Family Violence, 5 (2), 107–120. Retrieved from http://www
.springerlink.com/index/10.1007/BF00978514
doi: 10.1007/
BF00978514
Freedman, J. L. (2002). Media Violence and Its Effect on Aggression: Assessing
the Scientific Evidence. University of Toronto Press.
Fujihara, T., Kohyama, T., Andreu, J. M., & Ramirez, J. M. (1999). Justification of interpersonal aggression in Japanese, American, and Spanish students. Aggressive Behavior, 25 (3), 185–195. Retrieved from
http://doi.wiley.com/10.1002/(SICI)1098-2337(1999)25:
3<185::AID-AB3>3.0.CO;2-K doi: 10.1002/(SICI)1098-2337(1999)25:
3<185::AID-AB3>3.0.CO;2-K
Fuller, S. R., & Aldag, R. J. (1998). Organizational Tonypandy: Lessons
from a quarter century of the groupthink phenomenon.
Organizational Behavior and Human Decision Processes, 73 (2/3), 163–184. Retrieved from http://linkinghub.elsevier.com/retrieve/pii/
S0749597898927601 doi: 10.1006/obhd.1998.2760
Gackenbach, J. (2007). Psychology And the Internet: Intrapersonal, Interpersonal,
And Transpersonal Implications. Academic Press.
Gauntlett, D. (2005). Moving Experiences: Media Effects And Beyond. Indiana
University Press.
Gavish, B., & Gerdes, J. H. (1997). Voting mechanisms and their implications
in a GDSS environment. Annals of Operations Research , 71 , 41 – 74. doi:
129
8. C ONCLUSION
10.1023/A:1018931801461
Geen, R. G. (2001). Human Aggression. Open University.
George, J. M. (1995). Leader Positive Mood and Group Performance: The Case
of Customer Service. Journal of Applied Social Psychology, 25 (9), 778–
794. Retrieved from http://doi.wiley.com/10.1111/j.1559-1816
.1995.tb01775.x doi: 10.1111/j.1559-1816.1995.tb01775.x
Gillin, P. (2007). The new influencers: a marketer’s guide to the new social media. International Journal of Advertising, 26 (4), 567–569. Retrieved from
http://portal.acm.org/citation.cfm?id=1526140
Giustini, D. (2007). Web 3.0 and medicine. (Vol. 335) (No. 7633). BMJ Publishing Group Ltd.
Retrieved from http://www.pubmedcentral
.nih.gov/articlerender.fcgi?artid=2151149&tool=
pmcentrez&rendertype=abstract
Goffman, E. (1959). The Presentation of Self in Everyday Life. Doubleday.
Goldstein, A. P. (1994). The Ecology of Aggression. Springer.
Goldstein, N. E., Arnold, D. H., Rosenberg, J. L., Stowe, R. M., & Ortiz, C. (2001).
Contagion of Aggression in Day Care Classrooms as a Function of Peer and
Teacher Responses. Journal of Educational Psychology, 93 (4), 708–719. Retrieved from http://eric.ed.gov/ERICWebPortal/Home.portal
?_nfpb=true&_pageLabel=RecordDetails&ERICExtSearch
_SearchValue_0=EJ640596&ERICExtSearch_SearchType_0=
eric_accno&objectId=0900000b8007b53a
doi: 10.1037//
0022-O663.93.4.708
Grice, P. (1989). Studies in the Way of Words. Harvard University Press.
Gross, R., Acquisti, A., & III, H. J. H. (2005). Information revelation and
privacy in online social networks. Human Factors, 0707 (November),
71. Retrieved from http://portal.acm.org/citation.cfm?doid=
1102199.1102214 doi: 10.1145/1102199.1102214
Groupthink. (2012). Merriam-Webster Online Dictionary. Merriam-Webster.
Retrieved August 14, 2012, from http://www.merriam-webster.com/
dictionary/groupthink
Guillen, M. F., & Suarez, S. L. (2005). Explaining the Global Digital Divide: Economic, Political and Sociological Drivers of Cross-National Internet Use.
Social Forces, 84 (2), 681–708. Retrieved from http://www.jstor.org/
stable/3598474 doi: 10.1353/sof.2006.0015
Gumbrecht, M.
(2004).
Blogs as “Protected Space.
Retrieved August 13,
130
8. C ONCLUSION
2012, from http://psychology.stanford.edu/~mgumbrec/Blogs
_as_Protected_Space.pdf
Gundry, J. (1992). Understanding collaborative learning in networked organisations. In A. R. K. Ed (Ed.), Collaborative learning through computer conferencing (pp. 167–178). Berlin: Springer-Verlag.
Gundry, J. (2006). Web 0.0 Social Media. Retrieved April 15, 2012, from http://
www.knowab.co.uk/socialmedia.html
Haenlein, M., & Kaplan, A. M. (2009). Flagship Brand Stores within Virtual
Worlds : The Impact of Virtual Store Exposure on Real-Life Attitude toward
the Brand and Purchase Intent. Recherche, 24 .
Hara, N., & Kling, R.
(1999).
Students’frustrations with a Webbased distance education course.
First Monday, 4 (12), 5.
Retrieved from http://firstmonday.org/htbin/cgiwrap/bin/ojs/
index.php/fm/article/viewArticle/710/620
Hargittai, E. (2008). Whose Space? Differences Among Users and Non-Users
of Social Network Sites. Journal of Computer-Mediated Communication,
13 (1), 276–297. Retrieved from http://blackwell-synergy.com/
doi/abs/10.1111/j.1083-6101.2007.00396.x doi: 10.1111/j.1083
-6101.2007.00396.x
Harman, C. M., Klopf, D. W., & Ishii, S. (1990). Verbal aggression among Japanese
and American students. Perceptual and motor skills, 70 (3 Pt 2), 1130.
Harper, R., Rodden, T., Rogers, Y., & Sellen., A. (2008). Being Human: HumanComputer Interaction in the Year 2020. (Tech. Rep.). Cambridge: Microsoft
Research Ltd.
Hart, P. (1991). Irving L. Janis’ Victims of Groupthink. Political Psychology, 12 (2),
247–278. Retrieved from http://www.jstor.org/stable/3791464
Hart, P.
(1998).
Preventing Groupthink Revisited: Evaluating and
Reforming Groups in Government.
Organizational Behavior
and Human Decision Processes, 73 (2/3), 306–326.
Retrieved
from
http://www.sciencedirect.com/science/article/
B6WP2-45JB1M6-N/2/4a8855a2d89c4cbba2c2d41c22be820a doi:
10.1006/obhd.1998.2764
Hendler, J. (2008). Web 3.0: Chicken Farms on the Semantic Web (Vol. 41)
(No. 1).
IEEE.
Retrieved from http://ieeexplore.ieee.org/
lpdocs/epic03/wrapper.htm?arnumber=4445616 doi: 10.1109/MC
.2008.34
Hendler, J.
(2009).
Web 3.0 Emerging (Vol. 42) (No. 1).
IEEE Computer
131
8. C ONCLUSION
Society.
Retrieved from http://ieeexplore.ieee.org/lpdocs/
epic03/wrapper.htm?arnumber=4755170 doi: 10.1109/MC.2009.30
Henningsen, D. D., Henningsen, M. L. M., Eden, J., & Cruz, M. G. (2006).
Examining the Symptoms of Groupthink and Retrospective Sensemaking. Small Group Research , 37 (1), 36–64. Retrieved from http://
sgr.sagepub.com/cgi/doi/10.1177/1046496405281772 doi: 10
.1177/1046496405281772
Hensher, D. A., Rose, J. M., & Greene, W. H. (2005). Applied Choice Analysis: A
Primer. Cambridge University Press.
Hensley, T. R., & Griffin, G. W. (1986). Victims of Groupthink. Journal of Conflict
Resolution, 30 (3), 497–531. Retrieved from http://jcr.sagepub.com/
content/30/3/497.abstract doi: 10.1177/0022002786030003006
Herring, S. (1999). Interactional Coherence in CMC. Journal of ComputerMediated Communication, 4 (4), 0. doi: 10.1111/j.1083-6101.1999.tb00106.x
Hewett, T. T., Baecker, R., Card, S., Carey, T., Gasen, J., Mantei, M., . . . Verplank, W. (1992). ACM SIGCHI Curricula for Human-Computer Interaction (Vol. 2008) (No. February 17). ACM. Retrieved from http://
portal.acm.org/citation.cfm?id=133967
Hiltz, S. R. (1985). Online Communities: A Case Study of the Office of the Future.
Intellect Books.
Hiltz, S. R. (1990). Evaluating the virtual classroom. In L. M. Harasim (Ed.), Online education perspective on a new environment (pp. 133–183). Praeger.
Hiltz, S. R., & Turoff, M. (1978). The Network Nation: Human Communication
Via Computer. MIT Press.
Hiltz, S. R., & Turoff, M. (1985). Structuring computer-mediated communication systems to avoid information overload. Communications of the ACM ,
28 (7), 680–689. Retrieved from http://portal.acm.org/citation
.cfm?doid=3894.3895 doi: 10.1145/3894.3895
Hinduja, S., & Patchin, J. W. (2010). Bullying, cyberbullying, and suicide.
Archives of suicide research official journal of the International Academy
for Suicide Research , 14 (3), 206–221. Retrieved from http://www.ncbi
.nlm.nih.gov/pubmed/20658375
Hines, D. A., & Saudino, K. J. (2003). Gender differences in psychological,
physical, and sexual aggression among college students using the revised
conflict tactics scales. Violence and victims, 18 (2), 197–217. Retrieved
from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
Retrieve&db=PubMed&dopt=Citation&list_uids=12816404
132
8. C ONCLUSION
Hoegg, R., Martignoni, R., Meckel, M., & Stanoevska, K. (2006). Overview of
business models for Web 2 . 0 communities. Communication, 2006 (3), 1–
17. Retrieved from http://www.alexandria.unisg.ch/export/DL/
31412.pdf
Howard, T. W. (2010). Design to Thrive: Creating Social Networks and Online
Communities That Last. Morgan Kaufmann.
Huang, K.-h., & Deng, Y.-s. (2008). Social Interaction Design in Cultural Context
: A Case Study of a Traditional Social Activity. International Journal , 2 (2),
81–96. Retrieved from http://www.ijdesign.org/ojs/index.php/
IJDesign/article/view/341/168
Huesmann, L. R. (1986). Psychological Processes Promoting the Relation Between Exposure to Media Violence and Aggressive Behavior by the Viewer.
Journal of Social Issues, 42 (3), 125–139. Retrieved from http://dx.doi
.org/10.1111/j.1540-4560.1986.tb00246.x doi: 10.1111/j.1540
-4560.1986.tb00246.x
Huesmann, L. R. (1998). The role of social information processing and cognitive
schema in the acquisition and maintenance of habitual aggressive behavior.
In R. G. Donnerstein & E. Geen (Eds.), Human aggression theories research
and implications for social policy (pp. 73–109). Academic Press. Retrieved
from http://psycnet.apa.org/psycinfo/1998-06391-003
Humphrey, W. S. (1988, May). The software engineering process: definition and
scope. ACM SIGSOFT Software Engineering Notes, 14 (4), 82–83. Retrieved
from http://dl.acm.org/citation.cfm?id=75111.75122 doi: 10
.1145/75111.75122
Hyman, I. (1995). Corporal punishment, psychological maltreatment, violence, and punitiveness in America: Research, advocacy, and public policy. Applied and Preventive Psychology, 4 (2), 113–130. Retrieved from http://linkinghub.elsevier.com/retrieve/pii/
S0962184905800848 doi: 10.1016/S0962-1849(05)80084-8
IBM. (2012). DeepQ Project - FAQ. Retrieved January 26, 2012, from http://
www.research.ibm.com/deepqa/faq.shtml
Jakobsson, M. (2006). Virtual Worlds & Social Interaction Design. Print Media.
Janis, I. L. (1971). Groupthink. Psychology Today, 5 (6), 43–46, 74–76.
Janis, I. L. (1972). Victims of groupthink: a psychological study of foreign-policy
decisions and fiascoes. Boston, MA: Houghton, Mifflin.
Janis, I. L. (1982). Groupthink: psychological studies of policy decisions and
fiascoes. Boston, MA: Houghton Mifflin.
133
8. C ONCLUSION
Janis, I. L. (1989). Crucial decisions: leadership in policymaking and crisis
management. Free Press. Retrieved from http://books.google.cz/
books?id=QyVHAAAAMAAJ
Janis, I. L., & Mann, L. (1977). Decision making: A psychological analysis of
conflict, choice, and commitment. New York, NY: Free Press.
Jeffreys, H. (1998). Theory of Probability. Clarendon Press.
Jessup, L. M., Connolly, T., & Galegher, J. (1990). The effects of anonymity on
GDSS group process with an idea-generating task. MIS Quarterly, 14 (3),
313–321. Retrieved from http://www.jstor.org/stable/248893
?origin=crossref doi: 10.2307/248893
Jobe, J. B., & Mingay, D. J. (1991). Cognition and survey measurement: History
and overview (Vol. 5) (No. 3). Retrieved from http://dx.doi.org/10
.1002/acp.2350050303 doi: 10.1002/acp.2350050303
Jobe, J. B., Tourangeau, R., & Smith, A. F. (1993). Contributions of Survey
Research to the Understanding of Memory. Applied Cognitive Psychology, 7 (7), 567–584. Retrieved from http://doi.wiley.com/10.1002/
acp.2350070703 doi: 10.1002/acp.2350070703
Joinson, A., & Gackenbach, J. (1998). Causes and implications of disinhibited
behavior on the Internet. In J. Gackenbach (Ed.), Psychology and the internet intrapersonal interpersonal and transpersonal implications (pp. 43–60).
Academic Press.
Joinson, A. N. (2001). Self-disclosure in computer-mediated communication: The
role of self-awareness and visual anonymity. European Journal of Social
Psychology, 31 (2), 177–192. Retrieved from http://doi.wiley.com/
10.1002/ejsp.36 doi: 10.1002/ejsp.36
Joinson, A. N. (2006). Disinhibition and the internet. In Psychology and the
internet (pp. 75–92). Academic Press.
Jones, B. F., & Olken, B. A. (2005). Do Leaders Matter? National Leadership and
Growth Since World War II. Quarterly Journal of Economics, 120 (3), 835–
864. Retrieved from http://www.mitpressjournals.org/doi/abs/
10.1162/003355305774268165 doi: 10.1162/003355305774268165
Kamau, C., & Harorimana, D. (2008). Does knowledge sharing and withholding
of information in organizational committees affect quality of group decision
making? In The 9th european conference on knowledge management (pp.
341–348). Academic Publishing. Retrieved from http://www.academic
-conferences.org/eckm/eckm2009/eckm08-proceedings.htm
Kaplan, A. M., & Haenlein, M.
(2007).
Consumers , Companies and
134
8. C ONCLUSION
Virtual Social Worlds : A Qualitative Analysis of Second Life. Advances in Consumer Research , 36 (1), 2006. Retrieved from http://
scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:
Consumers+,+Companies+and+Virtual+Social+Worlds+:
+A+Qualitative+Analysis+of+Second+Life#0
Kaplan, A. M., & Haenlein, M.
(2009).
Consumer Use and Business Potential of Virtual Worlds: The Case of “Second Life.
International Journal on Media Management , 11 (3), 93–101.
article&doi=10.1080/14241270903047008&magic=crossref
doi: 10.1080/14241270903047008
Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges
and opportunities of Social Media. Business Horizons, 53 (1), 59–68. Retrieved from http://www.sciencedirect.com/science/article/
pii/S0007681309001232 doi: 10.1016/j.bushor.2009.09.003
Kaplan, A. M., & Haenlein, M. (2011). The early bird catches the news:
Nine things you should know about micro-blogging. Business Horizons, 54 (2), 105–113. Retrieved from http://linkinghub.elsevier
.com/retrieve/pii/S0007681310001254 doi: 10.1016/j.bushor.2010
.09.004
Kaptein, M., & Robertson, J. (2012). Rethinking statistical analysis methods
for CHI. Proceedings of the 2012 ACM annual conference on Human
Factors in Computing Systems CHI 12 , 1105. Retrieved from http://
dl.acm.org/citation.cfm?doid=2207676.2208557 doi: 10.1145/
2207676.2208557
Karahanna, E., Straub, D. W., & Chervany, N. L. (1999). Information Technology
Adoption Across Time: A Cross-Sectional Comparison of Pre-Adoption Beliefs and Post-Adoption Beliefs. MIS Quarterly, 23 (2), 183–213.
Karray, F., Alemzadeh, M., Saleh, J. A., & Arab, M. N. (2008). Human-Computer
Interaction : Overview on State of the Art. International Journal , 1 (1),
137–159. Retrieved from http://scholar.google.com/scholar?hl=
en&btnG=Search&q=intitle:Human-Computer+Interaction+:
+Overview+on+State+of+the+Art#0
Kass, R. E., & Raftery, A. E. (1995). Bayes Factors. Journal of the American Statistical Association, 90 (430), 773–795. Retrieved from http://
www.jstor.org/stable/2291091 doi: 10.1038/ejhg.2010.17
Katz, J. E., Rice, R. E., & Aspden, P. (2001, November). The Internet, 1995-2000:
Access, Civic Involvement, and Social Interaction. American Behavioral
Scientist , 45 (3), 405–419. doi: 10.1177/0002764201045003004
135
8. C ONCLUSION
Keesom, K. A. (2004). Anonymity, accountability & John Doe. In Proceedings of
the 1st annual conference on information security curriculum development
(pp. 115–118). New York, NY: ACM.
Keltner, D., & Robinson, R. J. (1996). Extremism, Power, and the Imagined Basis of Social Conflict. Current Directions in Psychological Science, 5 (4),
101–105. Retrieved from http://cdp.sagepub.com/lookup/doi/10
.1111/1467-8721.ep11452765 doi: 10.1111/1467-8721.ep11452765
Kessler, R. C. (1990). The national comborbidity survey. DIS Newsletter, 7 (12),
1–2.
Khosrowpour, M. (2008). Innovative Technologies For Information Resources
Management. Hershey, PA: Idea Group Inc (IGI).
Kiesler, S., Siegel, J., & McGuire, T. W. (1984). Social psychological aspects of
computer-mediated communication. American Psychologist , 39 (10), 1123–
1134. Retrieved from http://content.apa.org/journals/amp/39/
10/1123 doi: 10.1037/0003-066X.39.10.1123
Kietzmann, J. H., Hermkens, K., McCarthy, I. P., & Silvestre, B. S. (2011).
Social media? Get serious! Understanding the functional building
blocks of social media.
Business Horizons, 54 (3), 241–251.
Retrieved from http://linkinghub.elsevier.com/retrieve/pii/
S0007681311000061 doi: 10.1016/j.bushor.2011.01.005
Kihlstrom, J., & Cantor, N. (2000). Social intelligence. In R. Sternberg (Ed.),
Handbook of intelligence (pp. 359–379). New York, NY, USA: University
Press.
Kihlstrom, J., & Klein, S. (1994). The self as a knowledge structure. In R. Wyer
& T. Srull (Eds.), Handbook of social cognition (pp. 153–208). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Kim, A. J. (2000). Community Building on the Web. Peachpit Press.
Klein, D. B., & Stern, C. (2009). Groupthink in Academia Majoritarian Departmental Politics and the Professional Pyramid. Independent Review, 13 (4),
585–600.
Koerber, C. P., & Neck, C. P. (2003, January). Groupthink and sports: an application of Whyte’s model. International Journal of Contemporary Hospitality
Management , 15 (1), 20–28. Retrieved from /journals.htm?issn=0959
-6119&volume=15&issue=1&articleid=867456&show=html doi:
10.1108/09596110310458954
Kraemer, K. L., & King, J. L. (1988, July). Computer-based systems for cooperative
work and group decision making. ACM Computing Surveys, 20 (2), 115–
136
8. C ONCLUSION
146. doi: 10.1145/46157.46158
Kramer, R. M. (1998). Revisiting the Bay of Pigs and Vietnam Decisions 25
Years Later: How Well Has the Groupthink Hypothesis Stood the Test
of Time? Organizational Behavior Human Decision Processes, 73 (2/3),
236–271.
Retrieved from http://ejournals.ebsco.com/Direct
.asp?AccessToken=8UOUOVM34NP4BZWBXJDIYW-I17VP90O0&Show=
Object doi: 10.1006/obhd.1998.2762
Kroon, M. B. R., Hart, P., & Van Kreveld, D. (1991). Managing Group Decision Making Processes: Individual Versus Collective Accountability and
Groupthink. International Journal of Conflict Management , 2 (2), 91–
eb022695 doi: 10.1108/eb022695
Krosnick, J. A., & Alwin, D. F. (1987). An Evaluation of a Cognitive Theory of
Response-Order Effects in Survey Measurement. Public Opinion Quarterly,
51 (2), 201–219. Retrieved from http://poq.oxfordjournals.org/
cgi/content/abstract/51/2/201 doi: 10.1086/269029
Krosnick, J. A., & Presser, S. (2010). Question and Questionaire Design. In
P. V. Marsden & J. D. Wright (Eds.), Handbook of survey research (pp. 263–
314). Emerald Group Publishing.
Kruk, M. R., Van Der Poel, A. M., Meelis, W., Hermans, J., Mostert, P. G., Mos,
J., & Lohman, A. H. (1983). Discriminant analysis of the localization
of aggression-inducing electrode placements in the hypothalamus of male
rats. Brain Research , 260 (1), 61–79. Retrieved from http://www.ncbi
.nlm.nih.gov/pubmed/6681724
Kussmaul, C., & Jack, R. (2009). Wikis for Knowledge Management: Business
Cases, Best Practices, Promises, & Pitfalls. Web 20 , 1–19.
Labrie, N., & Quell, C. (1997). Your Language, My Language or English? The
Potential Language Choice in Communication Among Nationals of the European Union. World Englishes, 16 (1), 3–26. Retrieved from http://
dx.doi.org.proxy-ub.rug.nl/10.1111/1467-971X.00043
Lange, P. G. (2008). Publicly Private and Privately Public: Social Networking on YouTube. Journal of Computer-Mediated Communication, 13 (1),
361–380. Retrieved from http://blackwell-synergy.com/doi/abs/
10.1111/j.1083-6101.2007.00400.x doi: 10.1111/j.1083-6101.2007
.00400.x
Lapidot-Lefler, N., & Barak, A. (2012). Effects of anonymity, invisibility, and lack
of eye-contact on toxic online disinhibition. Computers in Human Behavior,
28 (2), 434–443. Retrieved from http://linkinghub.elsevier.com/
retrieve/pii/S0747563211002317 doi: 10.1016/j.chb.2011.10.014
137
8. C ONCLUSION
Lassila, O., & Hendler, J. (2007). Embracing "Web 3.0". IEEE Internet Computing,
11 (3), 90–93. Retrieved from http://ieeexplore.ieee.org/lpdocs/
epic03/wrapper.htm?arnumber=4196182 doi: 10.1109/MIC.2007.52
Lawrence, C. L. (2002). Communication Technology to Develop Collaborative
Skills. Journal of Educational Technology Systems, 31 (2), 191–204.
Lawson-Borders, G., & Kirk, R. (2005). Blogs in Campaign Communication.
American Behavioral Scientist , 49 (4), 548–559. Retrieved from http://
apps.isiknowledge.com/full_record.do?product=WOS&search
_mode=GeneralSearch&qid=2&SID=N14OIeD5ne2NO2Ailh5&page=
3&doc=126 doi: 10.1177/0002764205279425
Lea, M., Spears, R., & de Groot, D. (2001, May). Knowing Me, Knowing You:
Anonymity Effects on Social Identity Processes within Groups. Personality
and Social Psychology Bulletin, 27 (5), 526–537. Retrieved from http://
psp.sagepub.com/cgi/doi/10.1177/0146167201275002 doi: 10
.1177/0146167201275002
Leach, H. S., Komo, L. W., & Ngugi, I. K. (2012). Engaging Consumers
Through Company Social Media Websites. Computing and Information Systems, 16 (2), 5–7. Retrieved from http://cis.uws.ac.uk/
research/journal/CISJ-v16n2.pdf
Leana, C. R.
(1985).
A Partial Test of Janis’ Groupthink Model: Effects of Group Cohesiveness and Leader Behavior on Defective Decision Making.
Journal of Management , 11 (1), 5–17.
Retrieved
from
http://ezproxy.lib.ucf.edu/login?URL=http://
search.ebscohost.com/login.aspx?direct=true&db=buh&AN=
5692774&site=ehost-live
Lee, C., & Warren, M. (2010). Micro-blogging in the Workplace. In C. Bolan (Ed.),
Proceedings of 8th australian information security management conference
(pp. 42–48). Perth, Australia: SECAU - Security Research Centre.
Lee, G. B. (1996). Addressing anonymous messages in cyberspace. Journal of
Computer-Mediated Communication, 2 (1), 0. doi: 10.1111/j.1083-6101.1996
.tb00183.x
Lee Michael D, & Eric-Jan Wagenmakers.
(2010).
A Course in
Bayesian Graphical Modeling for Cognitive Science. Convergence, 7 ,
2009. Retrieved from http://users.fmg.uva.nl/ewagenmakers/
BayesCourse/BayesBook.pdf
Lenhart, A., & Madden, M. (2007). Teens, privacy and online social networks:
How teens manage their online identities and personal information in the
age of MySpace. Pew Internet and American Life Project. Retrieved from
http://apo.org.au/?q=node/16750
138
8. C ONCLUSION
Levine, J. M., & Moreland, R. L. (1994). Group Socialization: Theory and Research. European Review of Social Psychology, 5 (1), 305–336. doi: 10.1080/
14792779543000093
Lewin, C. (2005). Understanding and Describing Quantitative Data. In B. Somekh
& C. Lewin (Eds.), Research methods in the social sciences (pp. 220–230).
London: Sage.
Li, H. (2012). Weibo’s Influence on Collective Action Participation in China.
Unpublished doctoral dissertation, Masaryk University. Retrieved from
http://is.muni.cz/th/380536/fss_m/
Li, Z. (2003). Design and evaluation of a voting tool in a collaborative environment. Phd thesis, New Jersey Institute of Technology.
Libert, B. D. (2010). Social Nation: How to Harness the Power of Social Media
to Attract Customers, Motivate Employees, and Grow Your Business. John
Wiley & Sons.
Licklider, J. C. R. (1960). Man-Computer Symbiosis. Ire Transactions On
Human Factors In Electronics, HFE-1 (1), 4–11. Retrieved from http://
groups.csail.mit.edu/medg/people/psz/Licklider.html doi:
10.1109/THFE2.1960.4503259
Lin, K.-Y., & Lu, H.-P.
(2011).
Why people use social networking
sites: An empirical study integrating network externalities and motivation theory. Computers in Human Behavior, 27 (3), 1152–1161. Retrieved from http://linkinghub.elsevier.com/retrieve/pii/
S0747563210003766 doi: 10.1016/j.chb.2010.12.009
Liu, H., & Maes, P.
(2005).
InterestMap: Harvesting Social Network
Profiles for Recommendations.
Network , 54–59.
Retrieved from
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10
.1.1.107.488&rep=rep1&type=pdf#page=54
Lofgren, E. T., & Fefferman, N. H. (2007). The untapped potential of virtual game
worlds to shed light on real world epidemics. The Lancet infectious diseases, 7 (9), 625–629. Retrieved from http://www.ncbi.nlm.nih.gov/
pubmed/17714675
Lomas, W. (2009). Conflict , Violence , and Conflict Resolution in Hunting and
Gathering Societies. Journal of Anthropology, 17 (1).
Lorenz, K. (2002). On Aggression. Routledge.
Maglio, P. P., & Kandogan, E. (2004). Error Messages: What’s the Problem?
Queue, 2 (8), 50.
139
8. C ONCLUSION
Maier, M. (2002). Ten Years after A Major Malfunction...: Reflections on
"The Challenger Syndrome". Journal of Management Inquiry, 11 (3), 282–
292. Retrieved from http://jmi.sagepub.com/cgi/doi/10.1177/
1056492602113013 doi: 10.1177/1056492602113013
Majchrzak, A., Wagner, C., & Yates, D. (2006). Corporate wiki users. In
D. Riehle & J. Noble (Eds.), Proceedings of the 2006 international symposium on wikis - wikisym ’06 (Vol. pp, p. 99).
New York, NY:
ACM Press. Retrieved from http://portal.acm.org/citation.cfm
?doid=1149453.1149472 doi: 10.1145/1149453.1149472
Marreiros, G., Ramos, C., & Neves, J. (2004). Emotion and Group Decision Making in Artificial Intelligence. Engineering, 1–6.
Marsh, R. L., Hicks, J. L., & Bink, M. L. (1998). Activation of completed, uncompleted, and partially completed intentions. Journal Of Experimental
Psychology. Learning Memory And Cognition, 24 (2), 350–361. Retrieved
from http://doi.apa.org/getdoi.cfm?doi=10.1037/0278-7393
.24.2.350 doi: 10.1037/0278-7393.24.2.350
Maslow, A. H. (1943). A theory of human motivation. Psychological Review,
50 (4), 370–396. doi: 10.1037/h0054346
McCall, G. S., & Shields, N. (2008). Examining the evidence from smallscale societies and early prehistory and implications for modern theories
of aggression and violence. Aggression and Violent Behavior, 13 (1), 1–
pii/S1359178907000377 doi: 10.1016/j.avb.2007.04.001
McCarthy, J., & Wright, P. (2004). Technology as experience (Vol. 11) (No. 5).
MIT Press. Retrieved from http://www.amazon.ca/exec/obidos/
redirect?tag=citeulike09-20&path=ASIN/0262134470
McCauley, C. (1989). The nature of social influence in groupthink: Compliance
and internalization. Journal of Personality and Social Psychology, 57 (2),
.1037/0022-3514.57.2.250 doi: 10.1037/0022-3514.57.2.250
McCauley, C. (1998). Group Dynamics in Janis’s Theory of Groupthink: Backward
and Forward. Organizational Behavior and Human Decision Processes,
73 (2-3), 142–162. Retrieved from http://linkinghub.elsevier.com/
retrieve/pii/S0749597898927595 doi: 10.1006/obhd.1998.2759
Mcellistrem, J. (2004). Affective and predatory violence: A bimodal classification
system of human aggression and violence*1. Aggression and Violent Behavior, 10 (1), 1–30. Retrieved from http://linkinghub.elsevier.com/
retrieve/pii/S1359178903000521 doi: 10.1016/j.avb.2003.06.002
140
8. C ONCLUSION
McEnery, K. W. (1995). The Internet, World-Wide Web, and Mosaic: an overview.
Ajr American Journal Of Roentgenology, 164 (2), 469–473. Retrieved
from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=
Retrieve&db=PubMed&dopt=Citation&list_uids=7839991
McLeod, I., Zhang, Y., & Yu, H. (2003). Multiple-Choice Randomization. Journal
of Statistics Education, 11 (1), 0.
McQuail, D. (2010). McQuail’s Mass Communication Theory. SAGE Publications
Ltd.
Mika, P. (2007). Social Networks and the Semantic Web (Google eBook). Springer.
Millard, D. E., & Ross, M. (2006). Web 2.0: hypertext by any other name?
In U. K. Wiil, P. J. Nürnberg, & J. Rubart (Eds.), Structure (pp. 27–30).
ACM. Retrieved from http://portal.acm.org/citation.cfm?id=
1149941.1149947 doi: 10.1145/1149941.1149947
Mingay, D. J., & Greenwell, M. T. (1989). Memory Bias and Response-order Effects. Journal of Official Statistics, 5 (3), 253–263.
Miranda, S. M. (1994). Avoidance of Groupthink: Meeting Management
Using Group Support Systems.
Small Group Research , 25 (1), 105–
136. Retrieved from http://sgr.sagepub.com/cgi/doi/10.1177/
1046496494251007 doi: 10.1177/1046496494251007
Miranda, S. M., & Saunders, C. (1995). Group Support Systems: An Organization
Development Intervention to Combat Groupthink. Public Administration
Quarterly, 19 (2), 193–216.
Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological Review, 80 (4), 252–283. Retrieved from http://
Mischel, W. (1999). Personality coherence and dispositions in a cognitive-affective
personality system (CAPS) approach. In D. Cervone & Y. Shoda (Eds.), The
coherence of personality socialcognitive bases of consistency variability and
organization (pp. 37–60). The Guilford Press.
Mischel, W., & Shoda, Y. (1995). A Cognitive-Affective System Theory of Personality: Reconceptualizing Situations, Dispositions, Dynamics, and Reconceptualizing Situations, Dispositions, Dynamics and Invariance in Personality
Structure. Psychological Review, 102 (April), 246–268.
Mitchell, D. H., & Eckstein, D. (2009). Jury Dynamics and Decision-Making:
A Prescription for Groupthink. International Journal of Academic Research , 1 (1), 163–169. Retrieved from http://www.ijar.lit.az/pdf/
1/2009(1-24).pdf
141
8. C ONCLUSION
Moorhead, G., Ference, R., & Neck, C. P. (1991, June). Group Decision Fiascoes Continue: Space Shuttle Challenger and a Revised Groupthink
Framework. Human Relations, 44 (6), 539–550. Retrieved from http://
hum.sagepub.com/cgi/doi/10.1177/001872679104400601 doi:
10.1177/001872679104400601
Morris, M. G., & Venkatesh, V. (2000). Age Differences in Technology Adoption
Decisions: Implications for a Changing Work Force. Personnel Psychology,
53 (2), 375–403. Retrieved from http://doi.wiley.com/10.1111/j
.1744-6570.2000.tb00206.x doi: 10.1111/j.1744-6570.2000.tb00206.x
Mullen, B., Anthony, T., Salas, E., & Driskell, J. E. (1994). Group Cohesiveness and
Quality of Decision Making: An Integration of Tests of the Groupthink Hypothesis. Small Group Research , 25 (2), 189–204. Retrieved from http://
.1177/1046496494252003
Muller, M. N., & Wrangham, R. W. (2004). Dominance, aggression and testosterone in wild chimpanzees: a test of the ’challenge hypothesis’. Animal Behaviour, 67 (1), 113–123. Retrieved from http://linkinghub
.elsevier.com/retrieve/pii/S0003347203003981 doi: 10.1016/
j.anbehav.2003.03.013
Murchu, I. O., Breslin, J. G., & Decker, S. (2004). Online Social and Business
Networking Communities. Proceedings of ECAI 2004 Workshop on Application of Semantic Web Technologies to Web Communities, 5458 , 241–
267. Retrieved from http://www.deri.org/fileadmin/documents/
DERI-TR-2004-08-11.pdf doi: 10.1007/978-1-4419-7142-5
Murugesan, S.
(2007).
Understanding Web 2.0 (Vol. 9) (No. 4).
IEEE.
Retrieved from http://ieeexplore.ieee.org/lpdocs/
epic03/wrapper.htm?arnumber=4287373 doi: 10.1109/MITP.2007
.78
Myers, B. A. (1998). A brief history of human-computer interaction technology.
Interactions, 5 (2), 44–54. doi: 10.1145/274430.274436
Nastri, J., Peña, J., & Hancock, J. T. (2006). The Construction of Away Messages:
A Speech Act Analysis. Journal of Computer-Mediated Communication,
11 (4), 1025–1045. doi: 10.1111/j.1083-6101.2006.00306.x
Neck, C. P.
(1995, May).
Groupthink Remodeled: The Importance of
Leadership, Time Pressure, and Methodical Decision-Making Procedures. Human Relations, 48 (5), 537–557. Retrieved from http://
hum.sagepub.com/cgi/doi/10.1177/001872679504800505 doi:
10.1177/001872679504800505
Network, N. E. C. C. R. (2004). Trajectories of physical aggression from tod142
8. C ONCLUSION
dlerhood to middle childhood: predictors, correlates, and outcomes. Monographs of the Society for Research in Child Development , 69 (4), vii, 1–129.
Retrieved from http://www.blackwellpublishing.com
Nielsen, J. (1994a). Enhancing the explanatory power of usability heuristics.
In Proceedings of the sigchi conference on human factors in computing
systems: celebrating interdependence (pp. 152–158). New York, NY, USA:
ACM.
Nielsen, J. (1994b). Heuristic evaluation. In J. Nielsen & R. L. Mack (Eds.), Usability inspection methods (Vol. 17, pp. 25–62). John Wiley & Sons. Retrieved
from http://portal.acm.org/citation.cfm?id=189200.189209
doi: 10.1089/tmj.2010.0114
Nippert-Eng, C. E. (2010). Islands of Privacy. University of Chicago Press.
Norman, D. A. (1983). Design principles for human-computer interfaces. In
A. Janda (Ed.), Proceedings of the sigchi conference on human factors in
computing systems chi 83 (pp. 1–10). ACM Press. Retrieved from http://
portal.acm.org/citation.cfm?id=800045.801571 doi: 10.1145/
800045.801571
Norman, D. A. (1986). Cognitive Engineering. In D. A. Norman & S. W. Draper
(Eds.), User centered system design (Vol. 31-61, pp. 31–61). Lawrence Erlbaum Associates. Retrieved from http://www.ncbi.nlm.nih.gov/
pubmed/21864441
Norman, D. A. (2002). The Design of Everyday Things. Basic Books.
Nunamaker, J., Vogel, D., & Konsynski, B. (1989). Interaction of task and technology to support large groups. Decision Support Systems, 5 (2), 139–
pii/0167923689900031 doi: 10.1016/0167-9236(89)90003-1
Nunamaker Jr., J. F., Applegate, L. M., & Konsynski, B. R. (1987). Facilitating Group Creativity: Experience with a Group Decision Support System.
Journal of Management Information Systems, 3 (4), 5–19. Retrieved from
http://dl.acm.org/citation.cfm?id=33279.33280
O’Donnell, C. R. (1995). Firearm deaths among children and youth. American
Psychologist , 50 (9), 771–776.
O’Reilly, T. (2005). What Is Web 2.0 (Vol. 30) (No. 1). San Francisco:. Retrieved
from http://oreilly.com/web2/archive/what-is-web-20.html
doi: 10.1186/1472-6947-8-54
Palfrey, J., & Gasser, U. (2010). Born Digital: Understanding the First Generation
of Digital Natives. Basic Books.
143
8. C ONCLUSION
Papargyris, A., & Poulymenakou, A. (2005). Learning to fly in persistent digital worlds: the case of Massively Multiplayer Online Role Playing Games.
ACM SIGGROUP Bulletin, 25 (1), 41–49. Retrieved from http://portal
.acm.org/citation.cfm?id=1067699.1067706
Park, W.-w. (1990, October). A Review of research on Groupthink. Journal of Behavioral Decision Making, 3 (4), 229–245.
Retrieved from
http://doi.wiley.com/10.1002/bdm.3960030402 doi: 10.1002/
bdm.3960030402
Park, W.-W.
(2000, December).
A comprehensive empirical investigation of the relationships among variables of the groupthink model.
Journal of Organizational Behavior, 21 (8), 873–887.
Retrieved from
http://doi.wiley.com/10.1002/1099-1379(200012)21:
8<873::AID-JOB56>3.0.CO;2-8
doi: 10.1002/1099-1379(200012)21:
8<873::AID-JOB56>3.0.CO;2-8
Parrott, D. J. (2008). A theoretical framework for antigay aggression: review
of established and hypothesized effects within the context of the general
aggression model. Clinical Psychology Review, 28 (6), 933–951. Retrieved
from http://www.pubmedcentral.nih.gov/articlerender.fcgi
?artid=3049314&tool=pmcentrez&rendertype=abstract
Patnaik, D., & Becker, R. (2010, June). Needfinding: The Why and How of
Uncovering People’s Needs. Design Management Journal (Former Series) , 10 (2), 37–43. Retrieved from http://doi.wiley.com/10.1111/j
.1948-7169.1999.tb00250.x doi: 10.1111/j.1948-7169.1999.tb00250.x
Patterson, D., & Fox, A. (2012). Engineering Long-Lasting Software: An Agile
Approach Using SaaS and Cloud Computing, Alpha Edition. Strawberry
Canyon LLC.
Paulus, P. B. (1998). Developing Consensus about Groupthink after All These
Years. Organizational Behavior and Human Decision Processes, 73 (2/3),
9705809
Peña, J., & Hancock, J. T. (2006). An Analysis of Socioemotional and Task Communication in Online Multiplayer Video Games. Communication Research ,
33 (1), 92–109. doi: 10.1177/0093650205283103
Pfitzmann, A., & Hansen, M. (2010). A terminology for talking about privacy by data minimization: Anonymity, Unlinkability, Undetectability, Unobservability, Pseudonymity, and Identity Management. Retrieved April
28, 2012, from http://dud.inf.tu-dresden.de/literatur/Anon
_Terminology_v0.34.pdf
Phelps, R. P. (2005, October). Are U.S. Students the Most Heavily Tested on
144
8. C ONCLUSION
Earth? Educational Measurement: Issues and Practice, 15 (3), 19–27. Retrieved from http://doi.wiley.com/10.1111/j.1745-3992.1996
.tb00819.x doi: 10.1111/j.1745-3992.1996.tb00819.x
Picard, R. W. (2000). Affective Computing. Mit Press.
Pike, G. H. (2007). Google, YouTube, Copyright, and Privacy. Information Today(April), 15–16.
Popkin, H. A. (2007). Survive your inevitable online humiliation - Technology
& science - Tech and gadgets - Technotica - msnbc.com. Retrieved May
19, 2012, from http://www.msnbc.msn.com/id/20611439/#.T7eN
-rT9mXt
Poynter, R. (2010). The Handbook of Online and Social Media Research: Tools
and Techniques for Market Researchers. John Wiley & Sons.
Preece, J., Rogers, Y., & Sharp, H. (2002). Interaction design: beyond humancomputer interaction. J. Wiley & Sons.
Prendinger, H., & Ishizuka, M. (2005). The Empathic Companion: A CharacterBased Interface That Addresses User’s Affective States. Applied Artificial Intelligence: An International Journal , 19 (3-4), 267–285. doi: 10.1080/
08839510590910174
Rains, S. A. (2007). The Impact of Anonymity on Perceptions of Source Credibility and Influence in Computer-Mediated Group Communication: A Test
of Two Competing Hypotheses. Communication Research , 34 (1), 100–
125. Retrieved from http://crx.sagepub.com/cgi/doi/10.1177/
0093650206296084 doi: 10.1177/0093650206296084
Ramesh, G., & Bhattiprolu, R. (2006). Software Maintenance: Effective Practices
for Geographically Distributed Environments. Tata McGraw-Hill Education.
Ramírez, J. M., & Andreu, J. M. (2006). Aggression, and some related psychological constructs (anger, hostility, and impulsivity); some comments from
a research project. Neuroscience & Biobehavioral Reviews, 30 (3), 276–291.
Reeves, B., & Read, J. L. (2009). Total Engagement: Using Games and Virtual
Worlds to Change the Way People Work and Businesses Compete (Google
eBook) (Vol. 2009). Harvard Business Press.
Reynolds, C., & Picard, R. W. (2001). Designing for affective interactions.
Media, 499 – 503. Retrieved from http://citeseerx.ist.psu.edu/
viewdoc/download?doi=10.1.1.16.4688&rep=rep1&
type=pdf
145
8. C ONCLUSION
Roberts, L. D., & Parks, M. R. (1999). The social geography of gender-switching
in virtual environments on the Internet. Information Communication Society, 2 (4), 521–540. Retrieved from http://www.informaworld.com/
index/16DAP1G86YETQQ71.pdf doi: 10.1080/136911899359538
Robertson, J., & Good, J. (2005). Story creation in virtual game worlds. Communications of the ACM , 48 (1), 61–65. Retrieved from http://sro.sussex
.ac.uk/27518/
Roediger, H. L., & Marsh, E. J. (2005). The positive and negative consequences
of multiple-choice testing. Journal of experimental psychology Learning
memory and cognition, 31 (5), 1155–1159.
Rogers, Y., Sharp, H., & Preece, J. (2011). Interaction Design: Beyond HumanComputer Interaction. John Wiley & Sons.
Rose, J. D. (2011). Diverse Perspectives on the Groupthink Theory – A Literary Review. Emerging Leadership Journeys, 4 (1), 37–57. Retrieved
from http://www.regent.edu/acad/global/publications/elj/
vol4iss1/Rose_V4I1%_pp37-57.pdf
Rosenblum, D. (2007). What Anyone Can Know: The Privacy Risks of Social Networking Sites (Vol. 5) (No. 3). IEEE Computer Society. Retrieved
from http://doi.ieeecomputersociety.org/10.1109/10.1109/
MSP.2007.75 doi: 10.1109/MSP.2007.75
Rosenthal, J. M., & Okie, S. (2005). White Coat, Mood Indigo — Depression
in Medical School. New England Journal of Medicine, 353 (11), 1085–
1088. Retrieved from http://www.nejm.org/doi/full/10.1056/
NEJMp058183 doi: 10.1056/NEJMp058183
Rovio, E., Eskola, J., Kozub, S. A., Duda, J. L., & Lintunen, T. (2009). Can
High Group Cohesion Be Harmful?: A Case Study of a Junior Ice-Hockey
Team. Small Group Research , 40 (4), 421–435. Retrieved from http://
.1177/1046496409334359
Ryu, H. (2007). Walkthrough in Web Usability: Cognitive, Activity, and Heuristic
Walkthrough. In P. Zaphiris & S. Kurniawan (Eds.), Human computer interaction research in web design and evaluation (pp. 229–256). Idea Group
Inc (IGI).
Schank, R. C., & Abelson, R. P. (1977). Scripts, Plans, Goals and Understanding: An Inquiry into Human Knowledge Structures (L. E. Associates, Ed.).
Michigan University.
Schnettler, S. (2009). A structured overview of 50 years of small-world research.
Social Networks, 31 (3), 165–178. Retrieved from http://linkinghub
146
8. C ONCLUSION
.elsevier.com/retrieve/pii/S0378873308000695 doi: 10.1016/
j.socnet.2008.12.004
Schoffstall, C. L., & Cohen, R. (2011). Cyber Aggression: The Relation between
Online Offenders and Offline Social Competence. Social Development ,
20 (3), 587–604. Retrieved from http://doi.wiley.com/10.1111/j
.1467-9507.2011.00609.x doi: 10.1111/j.1467-9507.2011.00609.x
Schroeder, R. (2002). The Social Life of Avatars: Presence and Interaction in
Shared Virtual Environments Google eBook). Springer.
Schroth, C., & Janner, T.
(2007).
Web 2.0 and SOA: Converging Concepts Enabling the Internet of Services (Vol. 9) (No. 3).
IEEE.
Retrieved from http://ieeexplore.ieee.org/xpls/abs_all.jsp
?arnumber=4216107 doi: 10.1109/MITP.2007.60
Schwarz, N., & Hippler, H. J. (1991). Response alternatives: The impact of their
choice and presentation order. In P. P. Biemer, R. M. Groves, L. E. Lyberg, N. A. Mathiowetz, & S. Sudman (Eds.), Measurement errors in surveys (pp. 41–56). John Wiley & Sons, Inc. Retrieved from http://
nbn-resolving.de/urn:nbn:de:0168-ssoar-67257 doi: 10.1002/
9781118150382.ch3
Scobee, F. R. D., Smith, M. J., Onizuka, E. S., Resnik, J. A., McNair, R. E.,
McAuliffe, S., & Jarvis, G. B. (1986). Report to the President: By the PRESIDENTIAL COMMISSION on the Space Shuttle Challenger Accident (Tech.
Rep.). NASA History Office. Retrieved from http://history.nasa
.gov/rogersrep/genindex.htm
Sheeks, M. S., & Birchmeier, Z. P. (2007). Shyness, sociability, and the use
of computer-mediated communication in relationship development. Cyberpsychology and Behavior, 10 (1), 64–70. Retrieved from http://www
.ncbi.nlm.nih.gov/pubmed/17305450
Shiller, R. (2008). Challenging the crowd in whispers, not shouts.
Shirani, A. I., & Lee, S. C. (2008). COLLECTIVE INFORMATION PROCESSING AND COMPUTER MEDIATED COMMUNICATION: AN OVERVIEW
OF MAJOR THEORETICAL FRAMEWORKS. Retrieved 25/04/12, from
http://cisnet.baruch.cuny.edu/friedman/rm/277-8204.pdf
Shirky, C. (2009). Here Comes Everybody: How Change Happens When People
Come Together. Penguin.
Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunications. Wiley.
Shuen, A. (2008). Web 2.0: A Strategy Guide. O’Reilly Media, Inc.
147
8. C ONCLUSION
Sicilia, M., & Palazón, M. (2008). Brand communities on the internet: A case
study of Coca-Cola’s Spanish virtual community. Corporate Communications An International Journal , 13 (3), 255–270. Retrieved from http://
www.emeraldinsight.com/10.1108/13563280810893643 doi: 10
.1108/13563280810893643
Siegel, J., Dubrovsky, V., Kiesler, S., & McGuire, T. W. (1986). Group processes in computer-mediated communication.
Organizational Behavior and Human Decision Processes, 37 (2), 157–187.
Retrieved
from
http://www.sciencedirect.com/science/article/
B6WP2-4CYG35D-F6/2/7223207765560ed23d488140c40c680d
Slonje, R., & Smith, P. K. (2008). Cyberbullying: Another main type of bullying? Scandinavian Journal of Psychology, 49 (2), 147–154. Retrieved from
http://dx.doi.org/10.1111/j.1467-9450.2007.00611.x
Smith, C. B., McLaughlin, M. L., & Osborne, K. K. (1997). Conduct Control on
Usenet. Journal of Computer-Mediated Communication, 2 (4), 0. doi: 10
.1111/j.1083-6101.1997.tb00197.x
Smith, D. H. (1967). A Parsimonious Definition of “Group:” Toward Conceptual Clarity and Scientific Utility*. Sociological Inquiry, 37 (2), 141–
168.
Retrieved from http://dx.doi.org/10.1111/j.1475-682X
.1967.tb00645.x doi: 10.1111/j.1475-682X.1967.tb00645.x
Smith, S., & Jan, N. (1985). Groupthink and the Hostage Rescue Mission. British
Journal of Political Science, 15 (1), 117–123. Retrieved from http://www
.jstor.org/stable/193750
Somit, A. (1990). Humans, chimps and bonobos: the biological bases of aggression, war and peacemaking. Journal of Conflict Resolution, vol. 34 (no. 3),
pp. 553–582.
Squire, K. D., & A., C. S. (2006). Generating cyberculture/s: The case of Star Wars
galaxies. In D. Gibbs & K.-L. Krause (Eds.), Cyberlines 2.0: Languages and
cultures of the internet (p. 310).
Staub, E. (1989). The Roots of Evil: The Origins of Genocide and Other Group
Violence. Cambridge University Press.
Stewart, M. G.
(2010).
Margaret Gould Stewart: How YouTube thinks
about copyright (Ted video).
Retrieved 010-10-12 12:51:31, from
http://www.ted.com/talks/margaret_stewart_how_youtube
_thinks_about_copyright.html
Stone, C., Hansen, M., Kooperberg, C., & Troung, Y. K. (1997). Polynomial
splines and their tensor products in extended linear modeling. Annals of
Statistics, 25 (4), 1371–1470. Retrieved from http://www.jstor.org/
148
8. C ONCLUSION
stable/2959054
Subrahmanyam, K., & Smahel, D. (2011). Digital youth: The role of media in
development. Retrieved from http://scholar.google.com/scholar
?hl=en&btnG=Search&q=intitle:Digital+Youth#1
Suler, J. R. (2000, April). Psychotherapy in Cyberspace: A 5-Dimensional Model
of Online and Computer-Mediated Psychotherapy. CyberPsychology & Behavior, 3 (2), 151–159. doi: 10.1089/109493100315996
Suler, J. R. (2004a). The online disinhibition effect. Cyberpsychology and Behavior, 7 (3), 321–326. Retrieved from http://www.ncbi.nlm.nih.gov/
pubmed/15257832
Suler, J. R. (2004b). The Online Disinhibition Effect. CyberPsychology and Behavior, 7 (3), 321–327. Retrieved from http://www.liebertonline.com/
doi/abs/10.1089/1094931041291295
Suler, J. R. (2004c). The psychology of text relationships. In R. Kraus, J. S. Zack,
& G. Stricker (Eds.), Online counseling: A handbook for mental health professionals (pp. 19–50). CA: Elsevier.
Suler, J. R. (2012). Psychology of Cyberspace - The Online Disinhibition Effect. Retrieved May 20, 2012, from http://users.rider.edu/~suler/
psycyber/disinhibit.html
Sutherland, I. E. (1964). Sketch pad a man-machine graphical communication system. Simulation, 2 (5), R–3–R–20. Retrieved from http://
sim.sagepub.com/cgi/doi/10.1177/003754976400200514 doi:
10.1145/800265.810742
Sy, T., Côté, S., & Saavedra, R. (2005). The contagious leader: impact of
the leader’s mood on the mood of group members, group affective tone,
and group processes. (Vol. 90; Tech. Rep. No. 2). College of Business Administration, California State University, Long Beach, CA 90815,
USA. [email protected]. Retrieved from http://www.ncbi.nlm.nih.gov/
pubmed/15769239
Tanis, M., & Postmes, T. (2007). Twofaces of anonymity: Paradoxical effects of
cues to identity in CMC. Computers in Human Behavior, 23 (2), 955–970.
doi: 10.1016/j.chb.2005.08.004
Taylor-Bonds, D. (2009, April). When obsession turns deadly: Men against
Women - Detroit Top News | Examiner.com.
Retrieved May 19,
2012, from http://www.examiner.com/article/when-obsession
-turns-deadly-men-against-women
Tedeschi, J. T., & Felson, R. B. (1994). Violence, Aggression, & Coercive Actions.
149
8. C ONCLUSION
American Psychological Association.
Thurlow, C., Lengel, L. B., & Tomic, A. (2004). Computer Mediated Communication: Social Interaction and the Internet. SAGE.
Travers, J., & Milgram, S. (1969). An Experimental Study of the Small World
Problem. Sociometry, 32 (4), 425–443. Retrieved from http://www.jstor
.org/stable/2786545?origin=crossref doi: 10.2307/2786545
Tremblay, R. E. (2000). The development of aggressive behaviour during childhood: What have we learned in the past century? International Journal
of Behavioral Development , 24 (2), 129–141. Retrieved from http://jbd
.sagepub.com/cgi/doi/10.1080/016502500383232 doi: 10.1080/
016502500383232
Troyer, L. (2003). Incorporating theories of group dynamics in group decision
support system (GDSS) design. In Proceedings international parallel and
distributed processing symposium (p. 8). IEEE Comput. Soc. doi: 10.1109/
IPDPS.2003.1213215
Tsikerdekis, M. (2011). Engineering anonymity to reduce aggression online. In
K. Blashki (Ed.), Iadis international conference - interfaces and human computer interaction (pp. 463–467). IADIS.
Tsikerdekis, M.
(2012).
The choice of complete anonymity versus
pseudonymity for aggression online.
eMinds International Journal on Human-Computer Interaction, 2 (8), 35–57.
Retrieved
from
http://www.eminds.uniovi.es/index.php?journal=
eminds&page=article&op=viewFile&path[]=106&path[]=67
Tsikerdekis, M. (in press-a). Dynamic Voting Interface in Social Media: Does it
Affect Individual Votes? In Sofsem 2013: Theory and practice of computer
science. Springer-Verlag.
Tsikerdekis, M. (in press-b). The Effects of Perceived Anonymity and Anonymity
States on Conformity and Groupthink in Online Communities: A Wikipedia
Study. Journal of the American Society for Information Science and Technology.
Tsikerdekis, M. (in press-c). Pro/Con Lists and their Use in Group Decision Support Systems for Reducing Groupthink. INFOCOMP Journal of Computer
Science.
Turner, D.
(2004).
Extending the Wiki paradigm for use in the classroom.
International Conference on Information Technology Coding and Computing 2004 Proceedings ITCC 2004 , 1 , 255–259.
Retrieved
from
http://ieeexplore.ieee.org/lpdocs/epic03/
wrapper.htm?arnumber=1286462 doi: 10.1109/ITCC.2004.1286462
150
8. C ONCLUSION
Turner, M. E., & Pratkanis, A. R.
(1998).
Theoretical Perspectives
on Groupthink: A Twenty-Fifth Anniversary Appraisal.
Organizational Behavior and Human Decision Processes, 73 (2-3), 103–104. Retrieved from http://www.sciencedirect.com/science/article/
B6WP2-45JB1M6-B/2/4e491e56fa7537615c0537bdc7e8cdde doi:
10.1006/obhd.1998.2768
Turner, M. E., Pratkanis, A. R., Probasco, P., & Leve, C. (1992). Threat, cohesion,
and group effectiveness: Testing a social identity maintenance perspective
on groupthink. Journal of Personality and Social Psychology, 63 (5), 781–
796. Retrieved from http://search.ebscohost.com/login.aspx
?direct=true&db=pdh&AN=psp-63-5-781&site=ehost-live doi:
10.1037/0022-3514.63.5.781
Valkenburg, P. M., Peter, J., & Schouten, A. P. (2006). Friend Networking Sites
and Their Relationship to Adolescents’ Well-Being and Social Self-Esteem.
CyberPsychology Behavior, 9 (5), 584–590. Retrieved from http://www
.liebertonline.com/doi/abs/10.1089/cpb.2006.9.584 doi: 10
.1089/cpb.2006.9.584
Van Der Linden, W. J., & Hambleton, R. K. (1997). Item Response Theory: Brief
History, Common Models, and Extensions. Handbook of modern item response theory(1), 0.
Vitiello, B., Behar, D., Hunt, J., Stoff, D., & Ricciuti, A. (1990). Subtyping aggression in children and adolescents. The Journal of neuropsychiatry and
clinical neurosciences, 2 (2), 189–192.
Vredenburg, K., Mao, J.-Y., Smith, P. W., & Carey, T. (2002). A survey of
user-centered design practice. Proceedings of the SIGCHI conference on
Human factors in computing systems Changing our world changing ourselves CHI 02 , 40 (1), 471. Retrieved from http://portal.acm.org/
citation.cfm?doid=503376.503460 doi: 10.1145/503376.503460
Wadhwa, S., & Bhattacharya, S. (2000, July). Reengineering the Group DecisionMaking Process. Vision: The Journal of Business Perspective, 4 (2), 29–35.
doi: 10.1177/097226290000400204
Wagenmakers, E.-J., Lee, M. D., Lodewyckx, T., & Iverson, G. (2008). Bayesian
versus frequentist inference. In H. Hoijtink, I. Klugkist, & P. A. Boelen
(Eds.), Bayesian evaluation of informative hypotheses in psychology (pp.
181–207). New York: Springer.
Wahlster, W., Schwarzkopf, E., Sauermann, L., Roth-Berghofer, T., Pfalzgraf, A.,
Kiesel, M., . . . Sintek, M. (2006). Web 3.0: Convergence of Web 2.0 and the
Semantic Web. Technology Radar, II , 0.
Wallace, P. (2001). The Psychology of the Internet (No. 1). Cambridge: Cambridge
151
8. C ONCLUSION
University Press.
Wang, F.-Y., Carley, K. M., Zeng, D., & Mao, W. (2007). Social Computing: From
Social Informatics to Social Intelligence. IEEE Intelligent Systems, 22 (2), 79–
83. Retrieved from http://ieeexplore.ieee.org/lpdocs/epic03/
wrapper.htm?arnumber=4136863 doi: 10.1109/MIS.2007.41
Webber, L., & Webber, F. (2007). IT Project Management Essentials, 2007. Aspen
Publishers Online.
Weinshenker, N. (2002). Bimodal classification of aggression: affective defense and predatory attack. Aggression and Violent Behavior, 7 (3), 237–
pii/S1359178901000428 doi: 10.1016/S1359-1789(01)00042-8
West, J. A., & West, M. L. (2009). Using wikis for online collaboration: The power
of the read–write web. San Francisco, CA: John Wiley & Sons.
Whyte, W. H. (1952). Groupthink.
Wood, L. (2005). Blogs & Wikis: Technologies for Enterprise Applications?
The Gilbane Report , 12 (10), 1–24. Retrieved from http://scholar
.google.com/scholar?hl=en&btnG=Search&q=intitle:BLOGS+
%&+WIKIS+:+TECHNOLOGIES+FOR+ENTERPRISE+APPLICATIONS#0
Wu, J., & Boulevard, C. B. (2008). Why they enjoy virtual game worlds? An
empirical investigation. Journal of Electronic Commerce Research , 9 (3),
219–230. Retrieved from http://www.csulb.edu/journals/jecr/
issues/20083/paper4.pdf
Yen, J.-Y., Yen, C.-F., Wu, H.-Y., Huang, C.-J., & Ko, C.-H. (2011). Hostility in the real world and online: the effect of internet addiction, depression, and online activity. Cyberpsychology behavior and social networking, 14 (11), 649–55. Retrieved from http://www.ncbi.nlm.nih.gov/
pubmed/21476897 doi: 10.1089/cyber.2010.0393
YouTube. (2010). Classic Tony48219 Videos. Retrieved May 19, 2012, from
http://www.youtube.com/watch?v=wrP9XEjX028
Youtube. (2012a). ContentID. Retrieved April 11, 2012, from http://www
.youtube.com/t/contentid
Youtube. (2012b). Statistics. Retrieved April 2, 2012, from http://www
.youtube.com/static?template=pressstatistics
Zhao, D., & Rosson, M. B. (2009). How and why people Twitter: the role that
micro-blogging plays in informal communication at work. In Proceedings
of the acm 2009 international conference on supporting group work (pp.
152
8. C ONCLUSION
243–252). New York, NY, USA: ACM.
Zillmann, D. (1983). Arousal and aggression. In R. G. Geen & E. I. Donnerstein
(Eds.), Aggression: Theoretical and empirical reviews (pp. 75–102). New
York: Academic Press.
Zimmerman, D. H., & West, C. (1975). Sex roles, interruptions and silences in conversation. In B. Thorne & N. Henley (Eds.), Language
and sex difference and dominance (Vol. 105, pp. 105–129). Newbury
House. Retrieved from http://www.stanford.edu/~eckert/PDF/
zimmermanwest1975.pdf
Zlatuška, J. (1997). Stepping Stones to an Information Society (Vol. FIMU-RS-97;
Tech. Rep.). Brno, Czech Republic: Masaryk University.
153
S ELECTED P UBLICATIONS OF M ICHAIL T SIKERDEKIS
International Peer-Reviewed Publications
Tsikerdekis, M. (2011). Engineering anonymity to reduce aggression online. In
K. Blashki (Ed.), Iadis international conference - interfaces and human computer interaction (pp. 463–467). IADIS.
Tsikerdekis, M.
(2012).
The choice of complete anonymity versus
pseudonymity for aggression online.
eMinds International Journal on Human-Computer Interaction, 2 (8), 35–57.
Retrieved
from
http://www.eminds.uniovi.es/index.php?journal=
eminds&page=article&op=viewFile&path[]=106&path[]=67
Tsikerdekis, M. (in press-a). Dynamic Voting Interface in Social Media: Does it
Affect Individual Votes? In Sofsem 2013: Theory and practice of computer
science. Springer-Verlag.
Tsikerdekis, M. (in press-b). The Effects of Perceived Anonymity and Anonymity
States on Conformity and Groupthink in Online Communities: A Wikipedia
Study. Journal of the American Society for Information Science and Technology.
Tsikerdekis, M. (in press-c). Pro/Con Lists and their Use in Group Decision Support Systems for Reducing Groupthink. INFOCOMP Journal of Computer
Science.
Lectures and Presentations
Tsikerdekis, M. (2011). Social media and social networks: An introduction [pdf
document]. “PA159 Net-Centric Computing I” course, Faculty of Informatics, Masaryk University [Guest Lecture]. Retrieved September 20, 2012,
from http://tsikerdekis.wuwcorp.com/images/ISM.pdf.
Tsikerdekis, M.
(2012a, September).
Introduction to social media and social networks [pdf document].
SitSem 2012 [Invited Seminar Presentation]. Retrieved September 20, 2012, from
http://tsikerdekis.wuwcorp.com/images/ISM.pdf.
Tsikerdekis, M.
(2012b, September).
Social interaction design (sxid) for social media [pdf document].
SitSem 2012
[Seminar Presentation]. Retrieved September 20, 2012, from
http://tsikerdekis.wuwcorp.com/images/SIxDnSM.pdf.
154

Social Interaction Design for Social Media: The Case of Groupthink

Transcription

Similar documents

Groupthink

APA Cheat Sheet

PDF (Author`s version) - OATAO (Open Archive Toulouse Archive

Using an Online Game to Teach about Survival During Conflict

Globalize Your School.

View/Open

Three Processes of Memory

Washington Elementary School 1608 East Locust Davenport, IA

Layperson Definitions Of Aggression

Where Does Your Tax Money Go? - National Tax Research Center