PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/CIRC/DateDue.indd-p.1

 

THE CREATION AND VALIDATION OF A PERCEIVED ANONYMITY SCALE
BASED ON THE SOCIAL INFORMATION PROCESSING MODEL
AND ITS NOMOLOGICAL NETWORK TEST
IN AN ONLINE SOCIAL SUPPORT COMMUNITY

By

Haejin Yun

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Telecommunication, Information Studies and Media

2006

ABSTRACT

THE CREATION AND VALIDATION OF A PERCEIVED ANONYMITY SCALE
BASED ON THE SOCIAL INFORMATION PROCESSING MODEL
AND ITS NOMOLOGICAL NETWORK TEST

IN AN ONLINE SOCIAL SUPPORT COMMUNITY

By

Haejin Yun

The rapid development of communication technologies has increased attention to
the research construct of anonymity. This study redeﬁned anonymity as perceived
anonymity based on research gaps in three major theories of computer-mediated
communication (CMC) and empirical studies in Social Identity of Deindividuation
(SIDE) theory and group decision support systems (GDSS) research. The redeﬁned
construct of perceived anonymity adopted the Social Information Processing (SIP)
model’s approach to CMC. Two competing models of perceived anonymity affecting
online public disclosure — a deindividuation model and a SIP-based model -- were built
and compared for the predictive validity test.

The scale validation and the nomological network test were performed with data
from a real online social support community, MissyUSA, an online community for
married Korean women living in the USA. A total of 301 members completed the

online questionnaire. The data was analyzed with structural equation modeling.

The results showed that the perceived anonymity construct has a three-
dimensional hierarchical structure, consisting of self-anonymity (SA), other-anonymity
(OA), and discursive anonymity (DA). The SIP-based model was supported with
perceived anonymity negatively affecting online public disclosure. Need for social
support negatively affected all three sub—dimensions of PA, and increased online public
disclosure. A multiple group analysis by group identiﬁcation (GI) revealed that the
sizes of path coefﬁcients were comparable across the groups, which suggested that there
was no interaction effect of group identiﬁcation. A secondary analysis demonstrated
that perceived anonymity was not bounded by technical anonymity (deﬁned as nominal
anonymity), supporting the notion that technological conditions do not determine the
mental state of online community members. Although anonymity perceptions decreased
evaluation concern, the latter did not mediate between perceived anonymity and online
pubic disclosure. The data was also tested for reverse paths from online public
disclosure to the three sub-dimensions of perceived anonymity. Online public
disclosure decreased SA and DA, but not OA.

The present study attested to the theoretical applicability and predictability of the
SIP model over other CMC theories. First, it was conﬁrmed that the model, which
usually has been applied in the interpersonal or small group context, can be expanded to
embrace large group CMC, that is, members’ public communication in online
communities. Second, only the SIP model was able to predict the negative relationship
between perceived anonymity and online public disclosure. Theoretical and practical

implications of the study follow, together with limitations of the study.

Copyright by
HAEJIN YUN
2006

ACKNOWLEDGEMENTS

I wish to express my special gratitude to my advisor, Professor Robert LaRose.

He continuously challenged me regarding theoretical thinking and provided insightful
comments. Working with him was a signiﬁcant learning experience. I am thankful for
his remote advising by email while I worked on this dissertation in Korea. I also am most
grateful for having an exceptional doctoral committee and wish to thank Professors Mark
Levy, Kelly Morrison and Dan Jong Kim for their outstanding direction. I also appreciate
that they allowed me to defend my dissertation via teleconference, a most meaningful
experience for me as a telecommunication major.

Words are inadequate to express my sincere appreciation to Professor Bella Mody,
my former advisor, for her unchanging support and encouragement. Until, and even after,
she left Michigan State University, she has been a source of new perspectives and
practical solutions in my doctoral study. She taught me how to balance my life by sharing
with me her life. My husband Minsung and I cherish memories of being her neighbors in
her downstairs apartment. I also owe a special note of gratitude to Dean Charles Salmon
for having served on my guidance committee and for his continuous encouragement.

My family provided me the strength and persistence crucial to complete this work.
My husband deserves my deepest thanks and respect for his unparalleled understanding
and patience. He never failed to show his conﬁdence in me, cheering me and standing by
me through good times and bad. My mother Heasung sacriﬁced her six months in the
United States and another six in Korea to care for my newborn Daniel. Her help was the

most critical ingredient for me to become a doctoral candidate and mother in one and the

same year. My brother Kyungjin provided his expertise in database programming and
website design for data collection through online survey, remotely from Korea on demand.
My father Ilro encouraged me to maintain my enthusiastic attitude from Korea, gladly
bearing the lonely six-month absence of my mother. All other family members prayed for
my spiritual and physical health throughout the writing of this dissertation.

I extend many thanks to my colleagues and friends; Jaemin Cha, Mi Kyung Kim
and Hee-Jung Kim for their Christian companionship, especially during my late months of
pregnancy; Junghyun Kim and Hyeeun Lee for their coordinating my defense-at-a-
distance on my behalf; members of Lansing Korean United Methodist Church and
Reverend Borin Cho for their spiritual support; Michael Morris for affording me his
professional editing service; and the Department of Telecommunication, Information
Studies and Media staff and Nancy Ashley at the dean’s ofﬁce for assisting me with
administrative work necessary for completing my doctoral program. This research was
partially funded by the Graduate Student Research Enhancement Award at Michigan State
University.

Most of all, 1 thank God for His Greatest Love and Grace. I pursued my academic
and professional goal during the last eight years, away from my home country. He was
with me continually although I could not always realize His Presence. He gradually
unveiled His Plan, nourishing me with what I needed. Ending a chapter of my life and
looking forward to another chapter that He prepared for-me, I confess that only God is the

way to true knowledge, none and nothing else.

“The fear of the Lord is the beginning of knowledge. ” (Proverbs I : 7)

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix
LIST OF FIGURES ............................................................................................................. x
INTRODUCTION ............................................................................................................... 1
CHAPTER 1
REDEFINING ANONYMITY ............................................................................................ 4
Identifying Research Gaps ............................................................................................. 4
Research Gaps in Three Major Computer-Mediated Communication Theories ..... 4
Theoretical Gaps in Empirical Studies in Social Identity of Deindividuation
Theory and Group Discussion Support System Research ............................... 10
Redeﬁning Anonymity as Perceived Anonymity ........................................................ 18
Theoretical Signiﬁcance of Perceived Anonymity ...................................................... 26
CHAPTER 2
MODELING PERCEIVED ANONYMITY AND ONLIN E PUBLIC DISCLOSURE
IN AN ONLINE SOCIAL SUPPORT COMMUNITY .................................................... 31
Deﬁning Online Public Disclosure .............................................................................. 32
Self-Disclosure in General ..................................................................................... 33
Online Public Disclosure ....................................................................................... 35
Perceived Anonymity Affecting Online Public Disclosure ......................................... 38
Deindividuation Model .......................................................................................... 38
SIP-Based Model ................................................................................................... 41
The Social Information Processing Model Revisited ...................................... 41
SIP-Based Model, Respeciﬁcation of Deindividuation Model ....................... 45
CHAPTER 3
METHODS ........................................................................................................................ 52
Research Site .............................................................................. 52
Initial Item Generation and Reﬁnement ...................................................................... 54
Scale Validation and Nomological Network Test ....................................................... 59
Measures ................................................................................................................ 59
Data Collection ...................................................................................................... 60
Data Analysis ......................................................................................................... 62
CHAPTER 4
RESULTS .......................................................................................................................... 64
Exploratory Factor Analysis on Perceived Anonymity ............................................... 64
Conﬁrmatory Factor Analysis on Perceived Anonymity ............................................. 68
Perceived Anonymity Affecting Online Public Disclosure ......................................... 75
Multiple Group Analysis ............................................................................................. 80

vii

Secondary Analysis ....................................................................................................... 83

Comparison between Perceived Anonymity and Technical Anonymity .............. 83
Separate Paths from Self-, Other—, and Discursive Anonymity to
Online Public Disclosure and Evaluation Concern .......................................... 85
Reverse Paths from Online Public Disclosure to Perceived Anonymity ............... 85
CHAPTER 5
DISCUSSION .................................................................................................................... 88
Summary of Results ..................................................................................................... 88
Contributions of the Study ........................................................................................... 89
Theoretical Contributions ...................................................................................... 89
Practical Implication .............................................................................................. 95
Limitations and Future Research ................................................................................. 96
Perceived Anonymity Affecting or Affected by Online Public Disclosure? ......... 96
External Validity and Reliability ........................................................................... 96
Different Functions of Self-Disclosure .................................................................. 99
Explaining the Unexplained ................................................................................... 99
Conclusion .................................................................................................................. 101
APPENDICES .................................................................................................................. 103
REFERENCES ................................................................................................................. 146

viii

LIST OF TABLES

Table 1-1. Anonymity Deﬁnitions by CFO, SIDE and SIP ............................................. 9
Table 1-2. Examples of Anonymity Manipulations ....................................................... 16
Table 1-3. Types of Identity Information ....................................................................... 25
Table 2-1. Relational Developments in CFO, SIDE, SIP and Hyperpersonal

Perspective .................................................................................................... 44
Table 3-1. Candidate and Final Items by Concept ......................................................... 56
Table 3-2. Characteristics of Subjects ............................................................................ 61
Table 3-3. Correlations, Means and Standardized Deviations for Scales ...................... 63
Table 4-1. Factor Loadings — Five-F actor Solution ....................................................... 65
Table 4-2. Factor Loadings — Three-Factor Solution ..................................................... 66
Table 4-3. Fit Indices by Model ..................................................................................... 70
Table 4-4. Modiﬁcation Indices by Model ..................................................................... 70
Table 4-5. CFA Results .................................................................................................. 73
Table 4-6. Convergent and Discriminant Validity ......................................................... 73
Table 4-7. Final Items by CFA ......................................... ............................................. 75
Table 4-8. Results by Hypothesis ................................................................................... 80
Table 4-9. Multiple Group Analysis — Low vs. High GI ................................................ 81
Table 4-10. Mean Differences by Technical Anonymity ................................................. 84

ix

Figure 1-1.

Figure 2-1.

Figure 2-2.

Figure 2-3.
Figure 4-1.
Figure 4-2.
Figure 4-3.
Figure 4-4.
Figure 4-5.
Figure 4-6.

Figure 4-7.

Figure 4-8.

Figure 5-1.

LIST OF FIGURES

Hierarchical Dimension of Perceived Anonymity ........................................ 18

Deindividuation Model of Perceived Anonymity Affecting
Online Public Disclosure .............................................................................. 41

SIP-Based Model of Perceived Anonymity Affecting

Online Public Disclosure .............................................................................. 48
Reﬁned SIP-Based Model ............................................................................ 51
Scree Plot of First EFA ................................................................................. 67
Final CFA Model (Model CFA PA-2-4) ...................................................... 71
Deindividuation Model ................................................................................. 77
SIP-Based Model .......................................................................................... 78
Reﬁned SIP-Based Model ............................................................................ 79
Multiple Group Analysis by Group Identiﬁcation ........................................ 82

Separate Paths from Self-, Other-, and Discursive Anonymity to
Online Public Disclosure and Evaluation Concern ...................................... 86

Reverse Paths from Online Public Disclosure to
Perceived Anonymity .................................................................................... 87

Perceived Anonymity and Technical Anonymity ......................................... 91

INTRODUCTION

Thanks to the rapid development of communication technologies, scholarly
interest in research on anonymity or anonymous communication has developed
(Anonymous, 1998; Detweiller, 1993; Williams, 1998). Examples of anonymous
communication — for now, deﬁned as communication in which the identity of the source is
lacking -- abound in the history of face-to-face and traditionally-mediated communication:
anonymous letters to newspaper editors, anonymous authorship, anonymous reporting to
police departments, church confession, whistle blowing on governmental abuses, and the
like. All such examples show that people use anonymity in situations where revealing
some information is considered potentially damaging to themselves.

Anonymity is a primary component of several communication technologies,
including group decision support systems (GDSSs), anonymous electronic remailers,
computer-based bulletin board systems (BBSs), and Internet chat rooms. A sense of
anonymity brought by such technologies to users is regarded as one of the factors for a
distinctive communication behavioral pattern in computer-mediated communication
(CMC) l: People tend to be more disinhibited when communicating online (Suler, 2002;

J oinson, 1999). People say and do things in cyberspace that ordinarily they would not say
or do face-to-face. They express themselves more openly. Such a phenomenon is called
the “disinhibition” effect. The effect presents a double-edged sword: It can be extreme

ﬂaming, or unusual kindness in the online world.

 

1 The use of CMC as used in the present study refers to text-based computer technologies unless
otherwise speciﬁed.

On the positive side of the “disinhibition” effect is the ever-increasing number of
online social support communities (OSSCs) (Yun et al., 2004). Traditionally, face-to-face
mutual self-help groups such as Alcoholics Anonymous have emphasized the importance
of conﬁdentiality (not discussing the circumstances of another member without direct
consent) (The Self-Help Resource Centre, 2003). The identity of members should not be
revealed to outsiders. With the increased possibility of protecting privacy through
anonymous communication (Walther & Boyd, 2002), people now do not hesitate to gather
online and exchange support with similar sufferers. Anonymity differs from
conﬁdentiality in that in the former the identity of an anonymous source is not known to
anyone whereas in the latter the source is known to a limited number of others
(Anonymous, 1998). Online anonymity increases people’s willingness to reveal
themselves at deep levels, which is a major contributing factor to highly caring and
supporting relationships found in online social support communities (VanLear, Sheehan,
Withers, & Walker, 2005; Klaw, Heubsch, & Humphreys, 2000; Phillips, 1996). It has
been reported repeatedly that CMC can be characterized by high levels of self-disclosure
(Joinson, 2001b; McKenna & Bargh, 1998; Parks & Floyd, 1996). Tensions between
privacy and emotional closeness through self-disclosure seem considerably reduced in
cyberspace (Ben-Ze’ev, 2003). Via reciprocal self-disclosure, those afﬂicted by illnesses,
addiction, or other traumatic events can build intimate and supportive relationships with
strangers online (Radin, 2001; Tichon & Shapiro, 2003).

Despite many studies examining perceived similarities and differences in
communication features between CMC and face-to-face (F tF ) interactions, few studies
attempted to explicate the concept of anonymity itself, and to test it empirically. Further,

although there is anecdotal evidence that anonymity in CMC seems to encourage self-

disclosure in online social support communities (OSSC), the connection between
anonymity and self-disclosure in OSSCs has been assumed, but not questioned in terms of
what theory can explain the phenomenon.

This study sought to advance the current understanding of anonymity in CMC and
to examine its role in online social support communities (OSSC). The study ﬁrst
addresses theoretical and empirical research gaps in previous studies of anonymity in
CMC. The construct of anonymity is redeﬁned based on identiﬁed research gaps and
tested with data from real OSSC participants. For the predictive validity of the redeﬁned
construct, a nomological network model is built that explains the relationship between
anonymity and self-disclosure in OSSCs. The literature review, therefore, consists of two
parts: redeﬁning anonymity, and modeling anonymity and online self-disclosure in
OSSCs. The guiding theoretical framework is the Social Information Processing (SIP)
model (Walther, 1992). The literature review details how the SIP model contributes to
redeﬁning of the constructand to modeling anonymity and online self-disclosure in an

online social support community.

CHAPTER 1

REDEFINING ANONYMITY

What is anonymity? The present study deﬁnes anonymity as a perceived lack of
identity information that would help communicators recognize each other. This section
presents the explication of the anonymity concept based on gaps in previous anonymity
research. First, a theoretical guide on how to redeﬁne the concept is provided by
reviewing three major theoretical perspectives of CMC, the Cues-Filtered-Out (CFO)
approach, Social Identiﬁcation of Deindividuation (SIDE) theory, and the Social
Information Processing (SIP) model. Then, a more detailed analysis of previous empirical
studies, especially SIDE and GDSSs studies, develops a framework on which the
explication is based. Finally, the concept of perceived anonymity is further reﬁned via

ﬁve types of identity information and two anonymity sub-dimensions.

Identifying Research Gaps

Research Gaps in Three Major Computer-Mediated Communication Theories

There are three major theoretical perspectives to CMC: The Cues-Filtered-Out

(CFO) approach, Social Identiﬁcation of Deindividuation (SIDE) theory, and the Social

Information Processing (SIP) model. Only SIDE theory explicitly includes the concept of

anonymity in its research paradigm among the three perspectives. However, close

examinations of the other two theories enable us to infer how each theory would have
deﬁned the anonymity concept.

Theories that belong to the ﬁrst theoretical approach are the Social Presence model
(Short, Williams, & Christie, 1976), Media Richness Theory (Daft & Lengel, 1984, 1986),
and the Reduced Social Cues approach (Sproull & Kiesler, 1986). These theories all
focus on the reduction of non-verbal cues as the critical difference between CMC and
face-to-face (F tF ) channels. The effects of the medium are determined by its technical
features (i.e., bandwidth restrictions) and are believed inherent, constant and context
invariant. These theories concur that CMC is impersonal and appropriate for task-
oriented communication. Further, the lack of social context cues reduces the impact of
social norms, therefore leads to deregulated, antisocial behavior such as ﬂaming. This
group of theories represents an early theoretical perspective of CMC in the 1970s and
19805. Although these early CMC theories did not pay attention to the anonymity concept
itself, their focus on the absence of nonverbal social cues in CMC, as opposed to in F tF
interactions, might have led researchers to deﬁne anonymity as a lack of co-presence with,
or invisibility of, communication partners. The more CMC users can sense each other as
if they interact face-to-face, with assistance of communication technologies, the more they
perceive they can identify each other.

The second theoretical perspective, SIDE theory, arose in the 19908 partly as a
response to the CFO approach. The original framework of the theory was designed to
model social inﬂuence processes in crowds (Reicher, 1984, 1987), not for CMC. The
application of the theory to CMC started in the mid 1980s, and the ﬁrst study on group
polarization was published in 1990 (Spears, Lea, & Lee, 1990). Rather than focusing on

the effects of reduced social context cues, a group of European social psychologists in

C MC redirected their theoretical attentions to the concept of anonymity. They also
introduced another important factor to their CMC studies: group identity salience (Lea &
Spears, 1991; Spears & Lea, 1992; Spears et al., 1990). The main point is that whereas
CMC indeed may ﬁlter out many social context cues that individuate people, group
identity cues are delivered relatively independently of bandwidth restrictions, and shift
people’s self focus from personal to group identity, affecting their deﬁnition of the
communication situation. According to SIDE theory, anonymity refers to whether or not
C MC users can identify each other. The theory employed different forms of anonymity.
Visibility or physical anonymity concerns itself with whether or not CMC users are
separated from each other in different locations. Visual anonymity refers to whether or
not they are provided with visual channels such as real time video conferencing systems
or pictures on computer screens. Nominal anonymity is deﬁned as whether or not they
use their real names or usemames, or no personal identiﬁers are assigned. Biographical
anonymity refers to whether or not they receive detailed information about each other
such as gender, age, hobby, major, and so on. The differential effects of these varying
forms of anonymity are yet to be investigated.

There are two major differences between the CFO approach and SIDE theory.
First, SIDE theory argues that anonymity in CMC does not lead always to anti-normative,
disinhibited behaviors. When group identity is salient, anonymity functions to further
increase group identity by reducing attention to individual differences, the so-called
“depersonalization” effect. As a result, group normative behaviors increase. Anti-
normative behaviors occur only when personal identity is salient instead of group identity,
when a group norm is not clear, there is no consensus about the norm, or the norm is

perceived as an out-group’s, not an in-group’s (Postrnes, Spears, Lea, & Reicher, 2000).

The second difference is the level of communication contexts on which each
theory focuses. The CFO approach has been applied to interpersonal or group
communication contexts while SIDE theory has examined group communication
exclusively. Studies of the CFO approach examined both individual and group level
outcomes while SIDE theory focused on group outcomes such as conformity to group
norms and group coherence. Methodologically, CFO researchers employed both ﬁeld
studies and experiments while SIDE theorists preferred experiments.

The third theoretical perspective, the SIP model, was also introduced in the early
1990s (Walther, 1992). Like SIDE theory, the SIP model also criticized the CFO
approach’s deterministic viewpoint of CMC that bandwidth restrictions remove social
context cues, which makes CMC impersonal. Based on the impression formation
literature, the SIP model argues that bandwidth restrictions and reduced social context
cues in CMC delay, rather than remove, social information exchange. The crucial factor
is time. Over time, people learn how to verbalize social context cues that, ofﬂine, are
non-verbal. CMC users develop an interpersonal epistemology, which refers to a
distinctive representation of the communication partner. It is an individuating knowledge
gained through ongoing interaction. If there is sufﬁcient time, the differences between
CMC and face-to-face communication diminish. Anonymity was not the concept of
interest in the SIP model. However, its acknowledgement of interpersonal knowledge
increasing over time hints that anonymity might be deﬁned as the lack of identity
information exchanged between CMC users that would help them recognize each other.

Walther (1992) argued that impersonal effects of CMC may be limited to initial
interactions among unacquainted communicators, by pointing out inconsistent ﬁndings

between laboratory and ﬁeld studies in the CFO approach. Walther proposed longitudinal

experiments. SIP studies usually involved small group contexts, and focused on
individual outcomes in relational communication such as immediacy, trust and dominance
(Walther & Burgoon, 1992), and anti-social communication behavior (Walther, Anderson,
& Park, 1994).

The SIP model and SIDE theory agree with each other, arguing against the ﬁxed
effects of CMC implied by the CFO approach. There also are some differences between
the two. In terms of the central concept of interest, the SIP model focuses on the effects
of reduced social context cues like the CFO approach, while SIDE theory focuses on the
concept of anonymity resulting from reduced social context cues. In addition, whereas
SIDE researchers employed one-shot experiments, SIP researchers preferred longitudinal
experiments. A comparison of the three theoretical perspectives is presented in Table 1-1.

The present study agrees with SIDE theory and the SIP model’s argument that the
effects of CMC are not technologically determined. SIDE theory emphasizes the
importance of group salience for such argument whereas the SIP model focuses on
communicator adaptability to limited bandwidth. By deﬁning anonymity in terms of
objective technological features of CMC, however, the SIDE model still holds an attribute
of technological determinism. Anonymity manipulations, such as whether real names or
user names are used, whether or not biographic database is provided, and whether or not
video conferencing systems are equipped, are given to experimental and control groups,
and SIDE experiments do not pay attention to how people adjust to such conditions over
time. Contrarily, the SIP model underscores communicators’ adaptability to technical
features of CMC and maintains that people are not bound by such features. Adopting the
SIP model’s viewpoint, the present study deﬁnes anonymity as a perceptual variable.

Perceived anonymity refers to a perceived lack of identity information exchanged among

CMC users. Anonymity perceptions are not ﬁxed, but vary according to the degree to

which CMC users develop interpersonal epistemology about each other and to which

communicators adapt to CMC over time.

Table 1-1. Anonymity Deﬁnitions by CFO, SIDE, and SIP

 

 

 

 

 

 

 

 

 

 

 

CFO SIDE SIP
Main argument CMC leads to Reduced social cues Differences between
impersonal and anti- are not the only factor CMC and FtF lie in
normative affecting (group) rates of social
communication interaction in CMC; information
behavior. identity cues (group processing. Given
vs. personal) are also sufﬁcient time, the
important. differences diminish.
Anonymity in CMC
may lead to group
normative behaviors.
Focus of Interest Lack of Social Group Salience; Lack of Social
Context Cues; Anonymity Context Cues; People
Bandwidth adapt to bandwidth
restriction of medium restrictions
is inherent and
context invariant
Communication Interpersonal, Group Group Interpersonal, Group
Context
Outcome level Individual, Group Group Individual
Preferred Field studies, One-shot experiments Longitudinal
Method Experiments experiments
Deﬁnition of Lack of co-presence Visibility, Visual Perceived lack of
Anonymity anonymity, Nominal identity information
anonymity, exchanged
Biographical
anonymity

 

Note. CFO, Cue-Filtered-Out approach; SIDE, Social Identity of Deindividuation theory; SIP,

Social Information Processing model

 

Theoretical Gaps in Empirical Studies in the Social Identity of Deindividuation (SIDE)

theory and Group Discussion Support Systems (GDSS) Research

Although commentaries on Internet anonymity abound (Nissenbaum, 1999;
Wayner, 1999; Lee, 2005), the empirical scholarship on anonymity comes from SIDE
theory and GDSS research. By allowing anonymous communication, GDSSs were
expected to increase idea generation and improve the quality of decision making by
liberating participants from social evaluations (Postrnes & Lea, 2000). Contrarily, SIDE
theory draws attention to the role of anonymity in increasing the salience of the group.
From the viewpoint of SIDE theory, GDSS research adopted the CFO approach that
anonymity diminishes the social inﬂuence of the group over the individual. Despite such
difference, the two streams of research share common weaknesses in deﬁning anonymity.
These weaknesses mainly stem from their common methodological approach, but
generate conceptual drawbacks.

Depending on experimental manipulations, anonymity has been operationalized
as a dichotomous variable. Various experiments constrained anonymity to visibility (or
physical proximity), and visual or nominal anonymity (Barreto & Ellemers, 20%;
Douglas & McGarthy, 2001 , 2002; Lea, Spears, & de Groot, 2001; Postrnes, Spears, &
Lea, 2002; Reicher, Levine, & Gordijn, 1998; Sassenberg & Postmes, 2002). That is,
researchers manipulated it by not showing a communication partner’s picture, name or
usemame on the computer screen. Alternatively, subjects communicated with each other
from separate rooms or in one lab together (Siegel, Dubrovsky, Kiesler, & McGuire,

1986; Sia, Tan, & Wei, 2002; Connolly, Jessup, & Valacich, 1990; Jessup, Connolly, &

10

Galegher, 1990; Jessup & Tansik, 1991; McLeod, Baron, Marty, & Yoon, 1997; Sosik,
1997).

Recent theorizing on anonymity (Anonymous, 1998; Hayne & Rice, 1997, Marx,
1999; O’Sullivan, Rains, & Grabb, 2001; Pinsonneault & Heppel, 1997-1998) argues that
this experimental approach is limited in three aspects. First, it only allows researchers to
examine objective features of communication technology. Subjective perceptions of
anonymity by communicators have been ignored. Also, the absence or presence of
technological features only makes the concept dichotomous while the subjective
perceptual dimension of anonymity renders it a continuous variable. Hayne and Rice
(1997) distinguished social and technical anonymity. Technically anonymous CMC takes
place when communication technologies are set to remove identifying information about
sources from messages. Social anonymity is deﬁned in terms of the ability to use the
stylistic characteristics available in messages to make attribution of authorship. The latter
type of anonymity can change over the course of communication, and does not necessarily
correspond to technical anonymity. Social anonymity is a subjective experience of
technical conditions, which varies considerably according to individuals in the same
technical condition. SIDE research and GDSS studies implicitly equated the subjective
dimension with the technical dimension.

The subjective perception approach embraces partial anonymity as well as perfect-
or non-anonymity (Anonymous, 1998; Hayne & Rice, 1997; O'Sullivan et al., 2001).
Anonymous (1998) emphasized the importance of discursive anonymity over physical
anonymity in verbal communication. Discursive anonymity concerns whether a message
can be connected to its source whereas physical anonymity refers to conditions wherein

one is physically separated from a message source and, therefore, cannot sense presence

11

of the source. Discursive anonymity has two key dimensions which determine its degree:
source speciﬁcation and source knowledge. Source speciﬁcation refers to the extent to
which a message source is distinguished from other possible sources. Source
speciﬁcation would vary on a continuum between when a message can be attributed to a
speciﬁc person and when to a group of individuals. Source knowledge concerns the
degree of familiarity between the source and the receiver. It would range from complete
strangers to close friends. Partial anonymity exists when either a message source cannot
be speciﬁed individually or when there is a moderate to low level of knowledge about a
message source. For example, course evaluation from a large course is partially
anonymous because the professor knows the class from which the evaluation came
(source knowledge) but cannot attribute individual comments to individual students.

Source speciﬁcation is analogous to Valacich, Dennis and Nunamaker’s (1992)
content anonymity, and Licker’s (1992) source dissociation. Content anonymity refers to
the degree to which one can identify a message source by recognizing the author through
an identiﬁer embedded in the message. Source dissociation is deﬁned as a perception that
others cannot identity one as the source of speciﬁc messages. Social anonymity, source
speciﬁcation, content anonymity, and source dissociation all attend to the connection
between a source and a message, and entail varying degrees of subjective anonymity
perceptions.

Another overlooked possibility is that anonymity is a multi-dimensional concept
(McLeod, 1997). Pinsonneault and Heppel (1997-1998) suggested ﬁve components (i.e.
lack of identiﬁcation, diffused responsibility, proximity, knowledge of other group
members, and the conﬁdence group members have in the system). Marx (1999) speciﬁed

seven components (i.e. legal name, locatability, pseudonyms linked to name or action,

12

pseudonyms not linked to name or location, social categorization, pattern knowledge, and
symbols of eligibility/non eligibility) of the concept. Among the ﬁve components by
Pinsonneault and Heppel (1997-1998), lack of identiﬁcation corresponds to Anonymous’
(1998) source speciﬁcation, knowledge of other group members to source knowledge, and
proximity to physical anonymity. Diffused responsibility and system conﬁdence are
antecedents rather than components of anonymity. When individuals perceive that
responsibilities are diffused to all members of the group, and when they trust that the
technical system really guarantees anonymity, they feel more anonymous. Among the
seven dimensions by Marx (1999), pattern knowledge is worth mention. Pattern
knowledge refers to distinctive behaviors or communication styles that can be attributable
to a particular person without actual identity or locatability. He argues that “being
unnamed is not necessarily the same as being unknown” (p.101). Like those we may
encounter regularly on a commuter train, we do know something about those we meet
online from their patterns, styles, or tones of online communication, but do not know their
real names, appearance or any speciﬁc personal information. It parallels social anonymity
by Hayne and Rice (1997), which emphasizes stylistic characteristics or evaluative tones
in messages. Empirical studies in GDSS and SIDE research have controlled one, or two
at best, anonymity conditions (see Table 1-2). Multiple aspects of anonymity generate
inﬁnite degrees of anonymity that also correspond to the continuous, subjective
conceptualization.

Third, anonymity has two aspects according to who is unidentiﬁable to whom.
Self-anonymity concerns identiﬁability of self to others. Other-anonymity refers to
whether or not others are identiﬁable to self. SIDE theorists also pointed out this

possibility of confounding effects (Sassenberg & Postmes, 2002; Spears & Lea, 1994).

13

Self-anonymity is relevant when a message source perceives his or her own identity is
unknown while other anonymity is what a message receiver experiences when responding
to a message from an unidentiﬁable source (Anonymous, 1998). Although the two
aspects are related and co-present in most real-life situations, they are conceptually
distinctive and lead to different social inﬂuences in groups. First, other-anonymity
increases group salience by obscuring individual differences among group members,
which in turn increases group attraction and conformity to group norms. SIDE research
calls this process the cognitive dimension. When we cannot differentiate others based on
individuating characteristics, we tend to depend on commonalities and perceive others
representative of their group. Highlighted similarities between self and others as members
of the same group leads to higher group attraction and conformity. Second, the other
social inﬂuence process concerns self-anonymity. When people perceive that others
cannot recognize them, they sense lower accountability about their behaviors or comments.
Self-perception as a unique individual rather than as a group member decreases group
conformity (Lea, Spears, Watt, & Rogers, 2000). This latter process is called the
strategic dimension. The two processes have opposing effects on group outcomes and
counterbalance each other. There would be interactions between the two dimensions. For
example, identiﬁability of others tends to increase anonymity of self. That is, people tend
to perceive that they are more unidentiﬁable when others are visible to them than when
others are not. Knowing themselves to be more anonymous than others reduces the sense
of being a member of the group.

A theoretical implication of the self versus other distinction is important when it
is applied to online communities. It is known that a majority of online community

members are “lurkers” who read others’ messages, but do not contribute to their

14

community (Cummings, Butler, & Kraut, 2002; DeSanctis & Roach, 2002; Nonnecke,
2000; Yun et al., 2004). For them, the strategic process comes into play because of higher
self-anonymity than other-anonymity.

SIDE theory’s dichotomous experimental approach might have stemmed from its
group psychology orientation. Its experimental treatments are typical of those of social
identity or social categorization theory, such as emphasizing subjects’ experimental group
assignments to prime social identity salience (Turner, Hogg, Oakes, Reicher, & Wetherell,
19987). Such methodological approach might have rendered SIDE theory another
dichotomous approach to anonymity. In GDSS research from the management
information systems ﬁeld, anonymity is a ﬁxed technical feature of text-based computer-
conferencing tools to enhance group decision-making. Although the origins of the two
research traditions differ, their methodological approaches resemble each other. Efforts to
redeﬁne anonymity as a perceptual variable should include how to resolve the limitations
mentioned above. The present study further reﬁnes the concept by incorporating multiple
components of anonymity and the directional sub-dimensions between self and other into

a new conceptualization.

15

Table 1-2. Examples of Anonymity Manipulations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GDSS Self- or
or Other-

Studies SIDE Manipulations Anonymity
Reicher & SIDE Experiment 1: Visibility: Subjects are separated by a Mixed
Levine, 1994 screen on a round table, or not.

Experiment 2: Nominal Anonymity: Subjects provide Self-

their own names, or usemames or codes. Anonymity
Postmes, SIDE Study 1: Nominal and Visual Anonymity: Subjects are Mixed
Spears, & identiﬁed with initials and a group tag only, or ﬁrst
Lea, 2002 names, a group tag and pictures.

Study 2: Visual Anonymity: Subjects are identiﬁed Other-

with pictures and usemames, or usemames only. Anonymiy
Douglas & SIDE Study 1: Nominal and Geographic Anonymity: Subjects Self-
McGarty, use full names and countries of residence, or none. Anonymity
2002 Study 2: Nominal and Course Anonymity: Subjects use Self-

full names and course titles enrolled, or not. Also, Anonymity

subjects are told that their messages can be linked to

them personally, or not
Barreto & SIDE Experiment 1 & 2: Visual Anonymity: Subjects are told Self-
Ellemers, that their pictures will be displayed on computer Anonymity
2000b screens, or not. Subjects are also told that they will be

required to justify their responses at the end of

experiments.
Douglas & SIDE Study 1: Nominal Anonymity and Traceable Email Other-
McGarty, Address: Internet Newsgroup messages with real names Anonymity
2001 and email addresses, or aliases or no email addresses.

Study 2 and 3: Nominal and Geographic Anonymity:

Subjects provide their names and countries of Self-

residence, or not. Anomity
Sassenberg & SIDE Study 1 and 2: Visual Anonymity: Pictures of subjects Both, but
Postmes, 2002 are shown, or not. Self— and Other-Anonymity are not mixed

separatelmaniﬂlated.
Lea, Spears, SIDE Visual Anonymity: Text-based computer-based Mixed
& de Groot conferencing system only or supplemented with two-
2001 way real-time silent video.
Sia, Tan, & GDSS VisibilityzFace-to-face meeting, CMC meeting in the Mixed
Wei, 2002 same room, or CMC meetingin a separate cubicle
Taylor & SIDE Biographic Anonymity: An electronic biographic Mixed
McDonald, database of each group member is provided, or only
2002 usemames of other members are provided
Joinson, 2001 Neither Study 1:Visibility: Subjects in CMC arrive at different Mixed

times and are separated in separate cubicles, and

subjects in face-to-face arrive at the same time and are

seated together.

Study 2: Visual Anonymity: Subjects can see their Other-

discussion partner’s video, or not. Anonymity

 

l6

 

Table 1-2. (continued)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

GDSS Self— or
or Other-

Studies SIDE Manipulations Anonymity
Joinson, 2001 Neither Study 3: Visual Anonymity: Other-Anonymity: A Both, but

video-conferencing picture of others on computer not mixed

screen; Self-Anonymity: Subjects arrive in a darkened

corridor and are led to a cubicle with a blackened

window, or arrive in a well-lit corridor and are lead to a

cubicle with a clear window
Tanis & SIDE Visual and Biographic Anonymity: Portrait pictures of Other-
Postmes, 2003 the partner are available or not, and Biographic Anonymity

information of the partner is available or not
Siegel, et al., GDSS Visibility, Nominal and Social Anonymity: Participants Mixed
1986 are in different locations; Comments are labeled with

computer terminal numbers; Participants are ad-hoc

groups of students.
Hiltz, et al., GDSS Nominal and Social Anonymity: Participants are Mixed
1989 labeled with pseudonyms; Participants are corporate

employees.
Weisband, GDSS Nominal and Social Anonymity: Participants are Mixed
Schneider, labeled with pseudonyms, and are seated in the same
Connolly, room, but arranged not to see each other; Participants
1995 are ad-hoc groups of students.
Connolly, et GDSS Nominal Anonymity and Visibility: Contributions are Mixed
al., 1990 tagged to names, or not; Participants are introduced to

each other in advance and physicalﬂ co-present.
Jessup, et al., GDSS Nominal Anonymity and Visibility: Contributions are Mixed
1990 tagged to names. or not; Participants are physically co-

present.
Jessup & GDSS Nominal Anonymity and Visibility: Contributions are Mixed
Tansik, 1991 tagged to names, or not; Participants are physically co-

present.
Valacich, et GDSS Nominal Anonymity: Contributions are tagged to Mixed
al., 1992 names, or not: Participants are introduced to each other

in advance.
McLeod, et GDSS Nominal Anonymity and Visibility: Contributions are Mixed
al., 1997 tagged to names, or not; Participants are physically co-

present.
Cooper, et al., GDSS Nominal Anonymity: Contributions are tagged to Mixed
1998 names, or not.
Sosik, 1997 GDSS Visibility: Participants are physically proximate. Mixed
George, et al., GDSS Visibility, Nominal and Social Anonymity: Participants Mixed

Easton, 1990

 

 

are seated separately; Real names are not linked to
comments; Participants are ad-hoc groups of students.

 

 

Note. SIDE, Social Identity of Deindividuation theory; GDSS, Group Decision Support Systems

17

 

Redeﬁning Anonymity as Perceived Anonymity

The anonymity construct was redeﬁned as perceived anonymity based on research
gaps identiﬁed in the previous section. It is a perceived lack of identity information that
would help communicators to recognize each other. Perceived anonymity consists of two
sub-dimensions — self- and other-anonymity. The self and other distinction concerns the
direction of anonymity. Self-anonymity refers to a perceived lack of identity information
about the self known to others. Other-anonymity involves perceptions of how much
identity information about others the self can recognize. Perceived anonymity is a global
construct that encompasses self— and other-anonymity. Therefore, there exists a
hierarchical relationship between perceived anonymity and its two sub-dimensions as

shown in Figure 1-1.

Figure 1-1. Hierarchical Dimension of Perceived Anonymity

  

Perceived Anonymity

    

Other - Anonymity

   

Self - Anonymity

Then, what is “identity”? Identity is the distinguishing character or personality of
an individual. All information representing a mental image of a person comprises his or
her identity. A person’s online identity does not necessarily parallel his or her off-line

identity. Online identity is what a person selects to present from a variety of identity cues

18

about who he or she is off-line. In fact, it was noted that people do have multiple versions
of the self off-line as well (Goffman, 1959; Jung, 1953). People exercise more than one
mental image according to which facet of self is emphasized. People take on different
personae according to different roles they assume in different situations. Internet
communication made it easier to act on a particular persona, chosen from multi-faceted
aspects of self. For example, I present myself as an enthusiastic but sometimes stressed
doctoral student in PHinisheD, an online community for those who are working on a
dissertation. In MissyUSA, another online community for Korean married women living
in the United States, I act as someone’s wife who has a nine-month old with complaints
about in-laws in Korea. In PhinisheD, my ethnic and gender identity is not as important
as in MissyUSA. Contrary to off-line, people possess a control over what to be presented
about self online. Cyberspace is considered as a safe laboratory to experiment various
personae without fear of disapproval by those in an off-line circle (Turkle, 1995). It can
be an idealized version with desired qualities emphasized, or a hidden self that is not fully
expressed in social life (Bargh, McKenna, & Fitzsimons, 2002).

Online identity is different from off-line identity in that it is re-composed with an
emphasis on different aspects of self that are consciously or subconsciously selected
according to its owner’s social needs. This does not suggest that off-line identity is a true
identity, and online identity. is a false one. Rather, it is a matter of self-expression
facilitated by anonymity on the Internet (Bargh, McKenna, & Fitzsmions, 2002). The SIP
model also adopts this self-presentational viewpoint, stating that people try to manage
self-images due to various motivations such as afﬁliation motive and dominance drive
(Walther, 1992).

A person’s self-anonymity perception depends on how much identity information

19

that signals his or her off-line identity has been presented to others in online interactions.
There are two cases in self-representation — overrepresentation and misrepresentation.
Overrepresentation concerns having some elements of truth about self ampliﬁed and
others downplayed. Misrepresentation refers to providing false information about self. If
a person omits some information about the self, or presents him/herself with fake
information, his or her perceived self-anonymity is high. A person’s other-anonymity
perception is determined by how well he or she can discern online communication
partners based on who they claim to be, that is, online identity. A person may detect a
mismatch between online and off-line identity and suspect deception. It is a matter of
trust among members, not of other-anonymity perception. Online communities grow
based on trust among members that identity information exchanged is authentic. The
revelation of a member’s fabrication of an online persona greatly inﬂuences the sense of
community (Donath, 1999; Birchmeier, Joinson, & Dietz—Uhler, 2005).

Then, what kinds of information may signal a person’s identity in online
communication? This study categorizes identity information into ﬁve types, based on
Marx’s seven types of identity knowledge (Marx, 1999). The categorization is
complemented with other anonymity deﬁnitions in previous research (see Table 1-3).
Marx’s seven types of identity knowledge are legal name, locatability (e. g. a telephone
number, a mail or email address), traceable pseudonym (symbols or nicknames that can be
linked to legal names or locatability), non-traceable pseudonym (symbols or nicknames
that cannot be linked to a person or an address), pattern knowledge (reference to
distinctive behavior or communicative patterns that can be attributable to a particular
person without actual identity or locatability), social categorization (identity information

that does not differentiate the individual from others sharing them, e. g. gender, ethnicity,

20

religion, age, class, education, language and organizational memberships), and symbols of
eligibility (symbols that tell its possessors could be entitled to have corresponding
knowledge, skills and authorities and labels them as to be treated in a certain way -
example: titles, uniforms, certiﬁcation).

The ﬁve categories of identity information the current study proposes are: name
or pseudonym, locatability, biographic information, communication pattern and style, and
audio-visual information. The degree to which a person reveals or obtains pieces of
information in each type determines perceived anonymity. First, name or pseudonym
corresponds to what SIDE theory refers to as nominal anonymity (Barreto & Ellemers,
2000a; Lea, Spears, Watt, & Rogers, 2000). Nominal anonymity seems to be at ﬁrst
glance a dichotomous variable. However, the use of ﬁrst name only generates degrees of
anonymity perceptions signiﬁcantly different from the use of full name. Using a
nickname further masks a person’s nominal anonymity. In CMC, user names or screen
names also signify nominal identity. Knowing a person’s name or screen name does not
necessarily mean full identiﬁability. In other words, people may identify others without
knowing their names (Anonymous, 1998; Marx, 1999).

In interpersonal or small group CMC, nominal anonymity is a binary variable
according to whether a CMC system allows them to interact without revealing real names
or usemames. A small number of participants make it a matter of either-or-not conditions
to distinguish different names. In the context of online community where a large number
of members participate, however, even if a CMC system requires members to use names,
members can rarely relate all known names to speciﬁc members. The longer a member’s
prior experiences in the community is, the more names the member can recognize. In this

sense, nominal anonymity can be a continuous variable, especially in a large group CMC.

21

Anonymity deﬁnitions in GDSS studies such as lack of identiﬁcation (Pinsonneault &
Heppel, 1997-1998), source speciﬁcation (Anonymous, 1998), content anonymity
(Valacich, Dennis, Nunamaker, 1992), and source dissociation (Licker, 1992) deal with
nominal anonymity. That is, these deﬁnitions, commonly emphasizing a connection
between a source and a message, were operationalized whether comments are tagged with
names or pseudonyms.

Second, locatability is about answering a “where” question while nominal identity
gives an answer to a “who” question (Marx, 1999). Various pieces of information may
provide locatability cues in CMC. In addition to email addresses or homepage addresses
which are the most representative cyber-locality information, people often include their
mail addresses, telephone numbers and fax numbers in online signatures. Some
anonymous online bulletin boards provide Internet Protocol (IP) addresses to prevent
extremely ﬂaming comments. Since IP addresses link only to a computer, not to the user
of that computer, and may change whenever a new Internet connection is made when
using commercial Internet services, these addresses are not completely traceable, but only
provide temporary identiﬁcation information. Even so, it provides online communicators
with partial locatability. Domiciliary anonymity (lacking a traceable address) in SIDE
theory parallels locatability.

Third, biographic information that would help characterize communicators also
serves as identity information. Marx’s social categorization and symbols of eligibility are
included in this category. Some such information is demographic - gender, ethnicity, age
— which may be discernible audio-visually. Other biographic information such as
education, hobby and profession is not readily available through audio-visual information

without other cues. Online signatures usually include such information through

22

organizational membership. Domain names in email addresses also provide hints as to
what kind of organization the message sender belongs, or of which organization he or she
prefers to be perceived a member. As Marx himself mentioned, this category does not
provide individuating information, but, informs communicators of stereotypic
characteristics that the other party might possess with a hint of group membership.
However, as shown in SIDE theory, such information provides biographic details about
communicators, although not so individuating as names or email addresses. Biographic
information may categorize communicators, rather than individuate them, but it helps
build a mental representation of communication partners better than does no information.

Fourth, Marx’s pattern knowledge that also embraces behavioral pattern is
trimmed to communication pattern and style, since behaviors are displayed and delivered
through verbal communication in CMC. This category parallels source knowledge
(Anonymous, 1998), social anonymity (Hayne & Rice, 1997), and knowledge of other
group members (Pinsonneault & Heppel, 1997-1998). This information includes uses of
certain acronyms, recurring mentions of certain topics, expressive tones and writing styles.
Unlike the above categories of identity information, such information requires stable
participation on the message receiver side. It reveals the more subtle side of personality
unique to the message source.

The last type of identity information that Marx did not mention is audio-visual
information. Audio-visual information is not unique to FtF communication. Although
such information is limited in CMC, it is not impossible to have such information
available in CMC. Some online communities allow (or require) users to upload pictures
in biography sections so as to promote more intimate and trustful relationships among

them. Physical appearances may be revealed through such option even though users may

23

provide false visual identities or manage their own outlooks. Also, video cameras and/or
headsets mounted to personal computers deliver those pieces of information to
communication partners. An avatar is an example of self-presentational visual
information (Waldzus & Schubert, 2000; Kang & Yang, 2004).

Varying degrees in perceived anonymity are determined according to how much
identity information in each category communicators perceive about each other. The
more identity information they can recognize, the fewer the anonymity perceptions they

sense.

24

 

2sz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:cscaxuam
5:933: 59:5 55.253:—
_::m_> 53:3. .8585 .msm._>-o_::<
muzbm
wEEB “Em
3:2 9:32:me
.moES 58:8
309:2: mo 303:2:
93ch 95:63. Ebm
A5555 owns—>55— 85: :0 032255. 58.38% :5 Susan
Euom oesom 933.595.: 53.3 58:3 :0 mom: :o::u_::EEoU
mEmBQEoE
2533:: bﬁﬁwzo .m:c::~_:mw:0
2:32.300 .3 Eon—:xm 2.: 52382:
55.55:: :ccﬁtowuﬁu £265.85 corms—:85
2:9on 3.5m oEQEonmQ 92:29.5
59:3: can 5
59:3: 30::
$353 .3225 :32
2:38: .3225 A2
a wExum—V £363
55:30:: owamquI
bﬁzeEoQ 5:58:84 .3826 :25 b25333:
593::qu 2:35—22
«3392?:0: .5 6:8: :oobm
553836 £5.88: :ccwuwzoumm 5:35:52 $83.5: :SN 0232:; 65::sz
35cm 28:00 850m .3 x23 _::_EoZ JEmZ $34 .352 2::2
moi Roa— mac—-82
Si ..__w e .85: M32 .83: a 22:5
9335 dose—.3 Q 0:3: .mscEx:o:< :smo::om:i mam 25. .532 55553:— 9::

 

 

5:558:— b::o::o 39; .m; 2%.:

25

Theoretical Signiﬁcance of Perceived Anonymity

Anonymity is deﬁned as a perceived lack of identity information that would help
communicators recognize each other. The ﬁve types of identity information and the two
directional (i.e. self to other and other to self) sub-dimensions are incorporated into the
reﬁned conceptualization. What differences would the new deﬁnition of anonymity as a
perceptual concept make in previous studies that employed a technically deﬁned
dichotomous conceptualization of anonymity in relation to self-disclosure?

Since most anonymity experiments recruited subjects with zero acquaintance and
allowed interaction times of less than an hour, subjects were not given opportunities to
overcome technical anonymity. According to the SIP model, anonymity perceptions do
not depend on CMC systems. C MC users gradually decrease levels of perceived
anonymity through communication. The degree to which anonymity perceptions decrease
by adding the same amount of information exchanged about each other would be larger in
CMC than in FtF interactions because the baseline perception is much lower in CMC than
in FtF. Experimental conditions that were used to increase group salience (for example,
the information that subjects are students in the same university versus in different
universities) would function as additional (biographic) information that would lower
anonymity perceptions much more when they interact through a CMC system than when
they communicate face to face. It is suggested, therefore, that ﬁeld studies that compare
different anonymity conditions in the same CMC system examine the effect of perceived
anonymity more correctly.

One SIDE experiment extended interaction time to two weeks and recruited

subjects from a population of Internet users (Taylor & MacDonald, 2002). The

26

researchers examined main and interaction effects of identiﬁability (low or high) and
salience (individual or group). Identiﬁability was manipulated by providing a biographic
database containing details of each group member in the experimental group but not in the
control group. Subjects communicated through the same CMC system. Contrary to
predictions by SIDE and other online self-disclosure studies, it was found that the more
that was known about group members (i.e. high identiﬁability), the more subjects self-
disclosed. There was also an interaction effect. That is, subjects self-disclosed the most
in the high identiﬁability and group salience condition, the second most in the high
identiﬁability and individual salience condition, then in the low identiﬁability and
individual salience condition, and the last in the low identiﬁability and group salience
condition. It can be inferred that, over the two weeks, subjects’ anonymity perceptions
decreased the most in the high identiﬁability and group salience condition and the least in
the low identiﬁability and group salience. Interactions with identiﬁable members,
coupled with heightened group membership, increased the sense that subjects knew each
other. Group membership seemed to serve as additional information to other biographic
information provided in the identiﬁable condition. In the low identiﬁable condition, on
the contrary, group salience increased the impersonality of interaction that was already
triggered by the absence of biographic information. As a result, subjects in the high
identiﬁability and group salience condition perceived the lowest level of anonymity while
the highest level of anonymity perception was induced in the low identiﬁability and
individual salience condition.

This study suggests that anonymity perceptions technically imposed by CMC
systems and those that have been overcome through communication have different effects

on communication outcomes. Previous experimental studies on online self-disclosure

27

focused on technically generated anonymity perceptions, and did not examine differences
in self-disclosure after subjects overcame their initial anonymity perceptions. Joinson
(2001) manipulated only one type of technical anonymity, lack of co-presence in study 1
and visual anonymity in studies 2 and 3. The other types of identity information — name,
locatability, biographic information, and communication pattern and style —- were not
provided in either condition at the beginning of experiments. Considering that anonymity
perceptions are also determined by these different types of identity information, it can be
argued that subjects’ anonymity perceptions might not have paralleled the technical
conditions. Further, the fact that the amount of self-disclosure was higher in the CMC
condition than in the FtF condition and that the content of self-disclosure was biographic
(for example, “I’m a psychology student”) implied that subjects’ anonymity perceptions
decreased in the CMC condition more than in the FtF condition, over the experiments.
Through exchanging biographic information about each other, they should have decreased
levels of perceived anonymity, and subjects in the CMC condition decreased perceived
anonymity more than those in the FtF conditions. Unfortunately, the experiments stopped
short of exploring the effect of decreased anonymity perceptions on future self-disclosure.
It can be counter-argued that self-disclosure negatively affects anonymity
perceptions rather than anonymity perceptions decrease self-disclosure. In other words,
the direction of causation may be reversed. Contrary to previous studies’ argument that
anonymity increases self-disclosure via decreased public self-awareness (Joinson, 2001),
subjects in such experiments might have been motivated to exchange more identity
information to increase predictability about the behaviors of the self and the other, as
Uncertainty Reduction Theory suggests. That is, subjects in CMC or unidentiﬁability

conditions wanted to lower perceived anonymity because it brought to them uncertainty

28

about communication situations. Here, it should be noted that self-disclosure has a
dimension other than the amount (or breadth) dimension, namely the depth dimension.
Strangers are willing to share superﬁcial information about each other, but not inner
thoughts or feelings. People reveal such content when they can expect positive outcomes
as a reward (Altman and Taylor, 1973). The breadth of self-disclosure would increase
when CMC users meet for the ﬁrst time in order to reduce uncertainty brought by
perceived anonymity. It is not until they feel they know each other enough and are
convinced that the other will positively respond to their revelations of the inner self that
they start to talk about themselves at deeper levels. Therefore, the effect of perceived
anonymity on self-disclosure would differ according to which dimension is the focus of
interest.

Online social support communities are characterized by sympathetic messages.
Members disclose even negative aspects of the self because they believe other members
will respond supportively to their postings. Perceived anonymity is relatively low because
they all are members of the same community, with similar life experiences. Sometimes,
depending upon the individual contents to be revealed, members are uncertain how other
members may reply. As they accrue more information about each other, they are more
likely to become understanding of others’ situations. Perceived anonymity is expected to
decrease the tendency of online community members to reveal negative aspects of the self.

In the following chapter, two competing models of perceived anonymity are built
and compared, namely a deindividuation model and a SIP-based model. Online self-
disclosure is redeﬁned as online public disclosure in order to apply it to the context of
online social support communities. The deindividuation model represents arguments in

previous online self-disclosure studies maintaining that anonymity in CMC leads to

29

heightened self-disclosure compared to in FtF interactions. The SIP-based model, on the

contrary, predicts lowered self-disclosure by perceived anonymity.

30

CHAPTER 2

MODELING PERCEIVED ANONYMITY AND ONLINE PUBLIC DISCLOSURE IN

AN ONLINE SOCIAL SUPPORT COMMUNITY

In the previous chapter, the construct of anonymity has been redeﬁned as
perceived anonymity, based on the comparison of three major theories in CMC and the
analysis of empirical studies in SIDE and GDSS. Based on the SIP model’s argument that
CMC users adapt to bandwidth restrictions, it was proposed that anonymity also should be
redeﬁned as a subjective concept of varying degrees. Two sub-dimensions of perceived
anonymity and ﬁve types of identity information have been identiﬁed to redeﬁne
anonymity as a subjective and perceptual continuous variable.

The following section covers the second objective of this study, modeling
perceived anonymity and online self-disclosure in an online social support community.
Such modeling is an important part of scale validation. Cronbach and Meehl (1955)
maintained in their seminal article on construct validity that a construct should be tested
within a nomological network of antecedent and consequent variables in order to examine
the predictive ability of its scale. A nomological network should be guided and built by a
relevant theory. Modeling of perceived anonymity and online self-disclosure draws upon
Walther’s SIP model.

This chapter begins with deﬁning online self-disclosure as online public disclosure
in order to make the concept more relevant to the context of online social support
communities which are characterized by excessive self-disclosing messages, compared to

communities with other purposes (Campell, 2002; Preece, 1999; Radin, 2001). The

31

second section of the chapter discusses the ways in which previous studies on self-
disclosure in CMC might model perceived anonymity and online public disclosure
(“deindividuation model”). The third section revisits the social information processing

(SIP) model, and a competing model is speciﬁed, based on SIP (“SIP-based model”).

Deﬁning Online Public Disclosure

The present study builds a model of perceived anonymity and online public
disclosure in an online social support community. The reasons for focusing on online
social support communities are twofold. First, anonymity, which is the central construct
of the present study, functions as a safety net for those who seek social approval from
someone with similar distressing experiences. Those people, inevitably or voluntarily,
reveal sensitive and intimate facts about themselves, expecting social validation and
emotional relief in return (Campbell, 2002). Risking anticipated vulnerabilities, they also
engage themselves in self-impression management. This represents a dialectical tension
that all humans experience between expressiveness and protectiveness, disclosure and
privacy (Rawlins, 1992), openness and closedness (Baxter, 1990), and ambiguity and
clarity (O’Sullivan, 2000). People want both emotional closeness and personal
boundaries, especially when they are distressed. Anonymity in CMC reduces such
tension that is produced in pursuing the two contradictory goals, by providing self-
presentational opportunities.

The second reason for online social support communities is that these communities
are characterized by excessive self-disclosing messages, compared to communities with

other purposes (Campbell, 2002; Preece, I999; Radin, 2001). Self-disclosure seems to

32

function variously, to wit: response solicitation and emotional catharsis. Excessive self-
disclosure is a manifestation of the “disinhibition effect” in CMC (Suler, 2002).
Therefore, the relationship between anonymity and self-disclosure can be examined most
clearly in online social support communities. This section ﬁrst deﬁnes self-disclosure in

general from a social exchange perspective, then online public disclosure in particular.

Self-Disclosure in General

There have been inconsistencies in the conceptual deﬁnitions used in self-
disclosure research. Several earlier deﬁnitions exempliﬁed how inconsistent they are.
The most frequently cited deﬁnition of self-disclosure is “the act of making yourself
manifest, showing yourself so others can perceive you” (Jourard, 1971). Worthy, Gary
and .Kahn (1969) deﬁned the concept as “that which occurs when A knowingly
communicates to B information about A which is not generally known and is not
otherwise available to B.” Cozby (1973) deﬁned it as “any information about himself
which person A communicates verbally to person B” (p.73). Goodstein and Reinecker
(1974) restricted the use of the term to “verbal disclosures that are of a private nature and
selectively revealed under only special circumstances” (p.198). Jourard’s deﬁnition
includes both verbal and non-verbal disclosing about the self. Therefore, it suggests that
whenever people encounter another, they automatically disclose some aspects of
themselves. Worthy, Gary and Kahn’s deﬁnition emphasizes whether or not the discloser
consciously or intentionally reveals. Cozby’s deﬁnition limits the scope of empirical
inquiry to verbally transmitted information about the self. Lastly, Goodstein and

Reinecker’s deﬁnition puts a restriction on the term by adding the private nature.

33

These inconsistencies reﬂect that the concept is multi-dimensional, not
unidimensional. Chelune (1975) suggested ﬁve dimensions: amount or breadth, intimacy,
duration or rate, affectiveness, and ﬂexibility. Wheeless and Grotz (1976) proposed a
different set of ﬁve dimensions: intention, amount, valence (positive/negative),
honesty/accuracy, depth. More generally, as shown in Social Penetration Theory (Altman
& Taylor, 1973) and according to Derlega and Chaikin (1977), self-disclosure has two
dimensions: breadth and depth. Breadth refers to the number of topics covered, and depth
to the intimacy level of the disclosure. Omarzu (2000) added the third dimension,
duration, which refers to the amount of the disclosure. Empirical studies on self-
disclosure in any context should be aware of such multidimensional structure of the
concept and clarify in which dimension the studies are interested.

Self-disclosure is a medium of social exchange, which involves balancing beneﬁts
and costs (Foddy, 1984). Various functions that self-disclosures have, such as self
expression (catharsis), self-clariﬁcation, social validation, relationship development and
social control (Derelga & Grzelak, 1979), are rewards that self-disclosers expect. There
are also risks that people take into account when considering self-disclosure. Baxter and
Montgomery (1996) identiﬁed four risks of disclosing: rejection by the listener, reduction
of one’s autonomy and personal integrity, loss of control or self-efﬁcacy, and the
possibility of hurting or embarrassing the listener. Kelly and McKillop (1996) added a
distorted impression on the part of the listener. The present study deﬁnes self-disclosure
as an exchange relationship. People weigh what they will gain or lose as a result of
disclosing. It always involves an audience, whether individual or group. Such revelations

could be made verbally or non-verbally.

34

Online Public Disclosure

Several points can be clariﬁed in deﬁning online public disclosure from the above
discussion. First, the depth dimension is more vulnerable to interpersonal risks listed
above than is the breadth dimension (Omarzu, 2000). If people think that disclosing the
core self entails more risks than rewards, they choose to talk about superﬁcial information
at length (breadth and duration). The depth dimension, at the same time, is likely to bring
on more rewards than other dimensions, by the receiver’s incurring an obligation to
reciprocate (Rubin, 1975). In the present study, the depth dimension is of interest.

Privacy erosion is one of the major concerns in Internet environments. In order to
use online services such as e-commerce sites and discussion boards, prospective members
are required to disclose to the owner of the site personal information such as real name,
age, postal address, email address, and phone numbers, by ﬁlling in registration forms.
People often provide invalid information when asked on web sites due to privacy concerns
(Fox, 2000). This is related to the desire to keep personal information out of the hands of
others. DeCew (1997) calls it the informational dimension of privacy.

Another major risk, more relevant to the context of online social support
communities and the depth dimension of self-disclosure, is the possibility that the
discloser’s self-image is distorted on the part of the disclosure receiver. Prospective
members surrender some of their informational privacy to community owners in order to
access a cyberspace where they can freely express their self-identity without interference
from others. It is the expressive dimension of privacy, according to DeCew (1997).
Online community members are freed from evaluation concerns usually experienced off-

line when they try to express some parts of their self — the inner or core self (Bargh,

35

McKenna, & Fitzsimons, 2002). Expressive privacy is higher online than off-line (Ben-
Ze’ev, 2003). Parodoxically, online community members gradually lose their expressive
privacy as they build interpersonal relationships and the community becomes an important
part of their identity. As close relationships develop in an online community, members
become more concerned about other members’ opinions as well as expect more supportive
responses. The public nature of message exchange in online communities magniﬁes such
concerns. That is, a few members’ negative replies can be viewed as the majority opinion
of the community because all members have access to publicly posted messages.

In summary, online community members disclose some information about the self,
especially about the core self, expecting social support from other members in return.
They go through processes of balancing the rewards and risks associated with the
disclosing — (informational) privacy erosion and having self-images distorted.

Second, the functional approach suggests that different situations entail different
motivations. The “trait versus situation” debate is yet to be settled in self-disclosure
research. Personality traits such as competence and sociability have been found positively
related to self-disclosure (DeVito, 2000). Studies on gender differences in self-disclosure
support the trait position. Different personality traits in male versus female, whether
biologically determined or socially learned, lead to different patterns of self-disclosure
(Hatch & Leighton, 1986; Petronio, Martin, & Littleﬁeld, 1984; Winstead, Derlega, &
Wong, 1984). Researchers should be clear about which position they hold. The focus of
the present study is on situational differences in self-disclosure. The main situational
factor in this study is how much anonymity individuals perceive in CMC. The same
individual may perceive different levels of anonymity according to online communities in

which the individual is participating, or topics that he or she is covering.

36

Third, self-disclosure requires at least one individual other than the self. It needs
an audience. Most self-disclosure research assumes, but is not limited to, interpersonal
contexts. In the current study, the audience of self-disclosure is the discloser’s online
social support community as a whole. Rheingold (1993) in his deﬁnition of online
community claimed that online community members carry on “public” discussions.2
Because an online community develops through public 3 exchanges of messages, the
primary model of interaction is marked by its relation to the community as a whole
(Burnett, 2000). People develop interpersonal relationships through public interactions
(Parks & Floyd, 1996; Radin, 2001; Utz, 2000; Zhang & Hiltz, 2003). Public exchanges
of messages on bulletin boards are oﬂen directed to speciﬁc members. Members
understand that their seemingly private exchanges are to be shared by other members as
well as targeted ones. Communication in online communities is a mixture of
interpersonal and (large) group communication. As Rheingold said ( 1993), participants
form webs of “(inter)personal” relationships through “public” discussions.

Online public disclosure is deﬁned as the willingness to share the core self with
other members. Although the disclosing member seeks social support as a reward, the

member is also aware of risks associated with the disclosure. They are concerned about

 

2 Rheingold (1993) deﬁned virtual communities as “social aggregations that emerge from the Net
when enough people carry on those public discussions long enough, with sufﬁcient human feeling,
to form webs of personal relationships in cyberspace” (p.5).

3 “Private” and “personal” is often confused as an antonym of “public.” In fact, the two words are
used interchangeably. Precisely, private is that which “is intended only for a certain person”
(DeCew, 1997, 56-58p), and personal means “pertaining to a particular person’s own affairs”
(Ben-Ze’ev, 2003, 452p). Not everything that is personal is also private. Messages exchanged in
an online community originate from individual members (personal), but are shared with the whole
community (public). Members may exchange emails (or other alternative mechanisms for
directing private messages to speciﬁc members such as a memo function) once they build
interpersonal relationships through public exchanges of messages. This interaction is personal and
“private” communication as opposed to “public" communication, and beyond the boundaries of
the community itself.

37

negative impressions that the disclosure would make on other members. The deﬁnition
includes only the depth dimension of self-disclosure. This refers not to a willingness to
make disclosures to speciﬁc members through private interactions like personal emails or
memo functions, but rather to the willingness to reveal such content to the whole
community as an audience by publicly posting messages. The discloser knows that his or
her message is to be posted on the community’s bulletin board where all members have

access.

Perceived Anonymity Affecting Online Public Disclosure

Deindividuation Model

 

The ﬁrst model explaining a relationship between perceived anonymity and online
public disclosure, the deindividuation model, is based on ﬁndings in previous studies on
self-disclosure in CMC, web-based surveys, and GDSS studies. The model proposes that
a third variable, evaluation concern (deﬁned as being worried about others’ opinion on
what the self does or says), mediates the effect of perceived anonymity on self-disclosure,
using the public self-awareness explanation of online self-disclosure.

A heightened tendency to self-disclosure online has been regarded as a
manifestation of “disinhibition.” Researchers examined different types of self-concepts to
explain how anonymity in CMC affects self-disclosure. McKenna and her colleagues
(Bargh, McKenna, & F itzsimons, 2002; McKenna, Green, & Gleason, 2002)

differentiated the “true” from the “real” self. Anonymity as a safety net enables the “true”

38

inner self to be expressed easily, which would be hidden under the “real” self in face-to-
face contexts.

“Private” versus “public” self-awareness also has been examined in relation to
self-disclosure in CMC (Matheson & Zanna, 1988; Joinson, 2001). The private self-
awareness explanation concerns self-information readiness by self-focus. The tendency to
disclose oneself increases when private self-awareness is heightened by making self-
relevant information more readily available (Franzoi & Davis, 1985). CMC users’ over-
estimation of their own contributions to online discussion groups demonstrates this
heightened private self-awareness in online interaction (Weisband & Atwater, 1999).
Being alone in front of one’s own computer screen in one’s own room or in a separate
computer laboratory cubicle where none but the self is visible, produces an anonymous,
especially other-anonymous, setting (Wallace, 1999). This sense of anonymity prompts
self-focus, which is conducive to self-disclosure.

The public self-awareness explanation concerns self-presentational motivation
(Joinson, 2001). If people experience reduced public self-awareness, it lowers self-
presentational concerns. Anonymity of others to the self in CMC decreases public self-
awareness and, subsequently, concerns about others’ evaluation about what the self says.
As a result, people tend to reveal more about themselves.

The self-awareness explanations parallel an early theoretical viewpoint of C MC ,
the Cue-Filtered-Out (CFO) approach. The CFO approach explained online hostilities
such as ﬂaming, using the concept of “deindividuation” due to reduced social cues
(Kiesler, Siegal, & McGuire, 1984) or reduced social presence (Short, Williams, &
Christie, 1976). Before being reformulated as Social Identiﬁcation of Deindividuation

(SIDE) theory, deindividuation referred to a reduction in self-focus. Zimbardo (1969)

39

argued that factors such as anonymity, arousal, and sensory overload, usually experienced
in crowds, lead to deindividuation. Prentice-Dunn and Rogers (1982) suggested that
deindividuation is caused by two factors - a reduction in accountability cues (via lowered
public self-awareness) and a reduction in self-awareness - which lead to decreased self-
regulation and use of internal standards.

In similar vein, web-based surveys, as compared to paper-based ones, were found
as having reduced socially desirable responses (Frick, Bachtiger, & Reips, 2001; Joinson,
1999), and increased levels of self-disclosure (Weisband & Kiesler, 1996) and the
willingness to answer sensitive questions (Tourangeau, 2003). In applied areas, CMC-
based interviewees tended to admit more health-related problems (Epstein, Barker, &
Kroutil, 2001), more HIV risk behaviors (Des Jarlais, Paone, Milliken, Turner et al.,
1999), and more drug use (Lessler, Caspar, Penne, & Barker, 2000). Anonymity in group
decision support systems leads to increases in idea generation because people ideas are
not withheld because of fear of negative reactions from other participants (Cooper,
Gallupe, Pollard, & Cadsby, 1998; Dennis & Valacich, 1999).

Such ﬁndings also can be explained by reduced public self-awareness which frees
individuals from concerns about others’ opinions in CMC. The deindividuation model of
perceived anonymity affecting online public disclosure derives from the same reasoning.
Anonymity perceptions experienced in an online social support community increase

online public disclosure, mediated by evaluation concern (see Figure 2-1).

Hla: Perceived anonymity increases online public disclosure.
H2a: Perceived anonymity decreases evaluation concern.

H3a: Evaluation concern decreases online public disclosure.

40

Figure 2-1. Deindividuation Model of Perceived Anonymity Affecting Online Public
Disclosure

  
  
  

   
 

     
 

Online Public
Disclosure

Perceived
Anonymity

 

 

Evaluation
Concern

SIP-Based Model

A competing model can be speciﬁed with the same variables differently from the
deindividuation model, based on the Social Information Processing (SIP) model. The
SIP-based model predicts that perceived anonymity decreases online public disclosure.
After revisiting the SIP model, a set of hypotheses are presented corresponding to each

hypothesis in the deindividuation model.
The Scml Information Processing (SIP) Model Revisited

The SIP model argues that differences in impression formation and relational
development between F tF communication and CMC lie in different rates of social
information exchange. Relational development takes more time in CMC than in FtF, but

eventually the same, sometimes higher, level of development is possible (Walther, 1996).

41

Three assumptions of the model made such prediction possible. First, people have the
afﬁliation motive. This basic human motivator leads people to relate to seek social
acceptance from each other (“relational motivators”). Second, in CMC, such motivations
are realized through verbalized exchanges of social information. The verbalization makes
social information processing take longer than face-to-face. CMC users adapt to the lack
of social context cues and learn how to encode and decode such information in texts.
Third, CMC users develop an interpersonal epistemology, which refers to a distinctive
representation of the communication partner. It is individuating knowledge gained
through ongoing interaction over time (Walther, 1992). Communicators in CMC, like
those in FtF interactions, are driven to develop social relationships. Previously unfamiliar
users become acquainted with others by forming impressions of others through textually
conveyed infonrration. They gradually reﬁne interpersonal knowledge by exchanging
information about each other. As such knowledge develops, they exchange more personal
messages.

Walther (1996) later developed the hyperpersonal perspective of CMC, an
extended version of the SIP model. Walther proposed the perspective after additional
analysis of the same longitudinal data that evidenced the SIP model. CMC groups were
rated signiﬁcantly more positive than their F tF counterparts on several relational
outcomes. The reduced social cues in CMC lead to optimized self-presentation and
idealized perception. The sender controls what information about the self is to be
communicated. The sender becomes more careful and selective in presenting himself.
Minimal, but reﬁned, information about the sender goes through an “overattribution”

process on the part of the receiver. The receiver builds stereotypical impressions of the

42

sender based on exaggerated information from the sender. As a result, more positive
relationships develop.

According to the SIP model, verbalized social information individuates CMC
users, proximating relational outcomes in CMC up to the FtF level (Tidwell & Walther,
2002; Ramirez, Walther, Burgoon, & Sunnafrank, 2002). The model was introduced as a
response to the Cue-Filtered-Out (CFO) approach that renders C MC impersonal.
According to SIDE theory, the lack of nonverbal cues in CMC emphasizes group salience,
and encourages individuals to perceive each other more similarly, and to form more
favorable impressions toward each other.

Although not explicitly stated in each theory, it can be inferred how much
individuating information each theory assumes is beneﬁcial to relationship development.
The CFO approach and SIP propose that social cues and information are necessary for the
development of positive relationship. Therefore, more information is better. On the
contrary, SIDE favors minimal information. Social information maximizes individual
differences, and reduces interpersonal attraction based on common group membership.
The hyperpersonal perspective suggests that increased “overrated” information contributes
to positive developments of relationships. Recalling that anonymity is deﬁned as a
perceived lack of identity information, perceived anonymity is favored in SIDE, but not in
CFO and SIP. The hyperpersonal perspective focuses on the “valence” of identity
information rather than its amount. That is, positively-loaded information contributes to
relationship development. Table 2-1 presents how the major CMC theories differ in

impression formation and relational communication.

43

Table 2-1. Relational Development in CFO, SIDE, SIP, and Hyperpersonal Perspective

 

 

 

 

 

 

 

 

Theory CFO SIDE SIP Hyperpersonal
“CMC is” Impersonal Impersonal or Interpersonal Hyperpersonal
Hypersonal
Bandwidth Inherent in Anonymity Can be Can be
restrictions CMC, cannot be rather than overcome overcome
overcome bandwidth through through
restriction is of verbalization of verbalization of
interest nonverbal cues nonverbal cues
Relational By individuating By social By individuating By exaggerated
closeness information categorization information self and other
images
Implication for More Minimal More Valence is more
relational information is information is information is important than
development better better better amount.

 

 

 

 

Note. CF 0, Cue-Filtered-Out approach; SIDE, Social Identity of Deindividuation theory; SIP,
Social Information Processing Model

Walther employed longitudinal experiments in order to demonstrate that C MC
reaches the same level of relational developments as FtF interactions, when sufﬁcient time
is allowed. The original focus was on changes within CMC, and the comparison with FtF
interactions were presented as providing baseline levels of relational development
(Walther & Burgoon, 1992). However, later studies by Walther and his colleagues,
especially after he found the hyperpersonal effects of CMC (1996), and other studies that
employed the SIP model, tended to move research foci from within-CMC changes to
between-CMC-and-FtF differences (e.g. Tidwell & Walther, 2002; Walther, Slovacek &
Tidwell, 2001; Weisgerber, 2000). Researchers, implicitly or explicitly, aimed at
evidencing that CMC relationships are more positive than FtF ones, and tried to ﬁnd
contingencies on which CMC effects diverge (e. g. anticipated future interaction, Walther,
1994). A theoretical implication is that channel effects of the medium (CMC versus F tF ),
which the SIP model originally criticized in opposing the Cue-Filtered-Out approach,
regained its central position, although the direction of the effects was reversed. The

present study, therefore, proposes returning research attention from comparative effects of

44

CMC to temporal changes inside CMC. Such perspective change also is consistent with
the conceptualization of anonymity as a perceptual continuous variable, as opposed to a

dichotomous technical variable.
SIP-Based Model, Respeciﬁcation of Deindividuation Model

Previous studies that compared CMC and F tF maintained that anonymity,
narrowly deﬁned as visual anonymity, positively affected (the amount of) self-disclosure
(Joinson, 2001). Visual anonymity, via reduced public self-awareness, decreases
evaluation concern. Reduced evaluation concern leads to increasing self-disclosure. We
call it the deindividuation model because its explanation parallels the traditional
deindividuation theory (Zimbardo, 1969; Prentice-Dunn & Rogers, 1982). Perceived
anonymity, based on the SIP model, does not equal visual anonymity. It also is
determined by other kinds of identity information available to communicators. This
section presents how the SIP model predicts the effect of perceived anonymity on online
public disclosure differently from the deindividuation model.

As shown above, SIP focuses on relational development in CMC. Cornerstone
theories on which Walther relied for developing SIP were Berger and his colleagues’
Uncertainty Reduction Theory (URT) (Berger & Calabrese, 1975; Berger, Gardner, Parks,
Schulman, & Miller, 1976; Berger, 1979; Berger & Bradac, 1982; Berger, 1987), and
Altman and Taylor’s Social Penetration Theory (SPT) (1973). URT predicts that people
will be motivated to share personal information the ﬁrst time they meet in order to reduce
uncertainty (or increase predictability) about the behaviors of both themselves and others.

According to SPT, relationships develop toward greater intimacy and afﬁliativeness as

45

people exchange more information about themselves at deeper levels on a broader range
of topics (Altman & Taylor, 1973). The ﬁrst assumption of SIP mentioned above —
people have the affiliation motive which increases communication with others — was
drawn from SPT, and the third assumption — CMC users develop interpersonal knowledge
about each other through communication -— from URT. The two theories concur with the
idea of Thibaut and Kelly’s Social Exchange Theory (1952) that people tend to regulate
relational closeness on the basis of rewards and costs (Berger, 1987; Littlejohn, 1992).
That is, people base the likelihood of developing a relationship with someone on the
perceived possible outcomes (rewards minus costs). Peoples’ afﬁliation motive prompts
them to seek intimacy in relationships. When people can anticipate desired relational
outcomes, they are willing to invest their resources to develop relationships. Applying
this perspective to the present study, social support from other online community
members is the anticipated reward in relational development. They are willing to share
their negative aspects of life because of expected social support from other members. But
when anonymity perceptions are high, a member feels like being in a community whose
members are indifferent and unresponsive to what his or her problems are because they
also do not know who is disclosing. Thus, the member will not bother to invest his or her
time and efforts by posting personal problems. It is not a fair exchange. When people
perceive Internet environments as warrrr, active, and sociable enough to provide
anticipated social rewards, their self-disclosing becomes more intimate, revealing negative
aspects of the self (Ma, 2003). Therefore, perceived anonymity tends to decrease online

public disclosure.

Hlb: Perceived anonymity decreases online public disclosure.

46

Online social support communities are characterized by highly caring messages
exchanged among members. They expect to be understood by other members who are in
similar situations. Therefore, they often reveal what they would not off-line. Sometimes,
they want to vent unﬁltered personal opinions or emotions (for example, ‘why I came to
have an extra marital relationship with a married man’). When disclosing such stories,
members will expect critical as well as sympathetic responses. Perceived anonymity
alleviates evaluation concern as in the deindividuation model. The path from evaluation
concern to online public disclosure in the deindividuation model is reversed in the SIP-
based model with online public disclosure positively affecting evaluation concern. This
reverse path implies that the public self-awareness explanation of the deindividuation
model does not hold true in the SIP-based model. Evaluation concern is not so important
in predicting online public disclosure in the SIP-based model, which underscores the role
of anticipated social support from others as a reward for self-disclosing in depth (see

Figure 2-2).

H2b: Perceived anonymity decreases evaluation concern.

H3b: Online public disclosure increases evaluation concern.

The major difference between the deindividuation model and the SIP-based model
lies in different communication situations. In the deindividuation model, no future
interaction is anticipated, as among strangers who will never meet again, and the primary
function of self-disclosure would be catharsis — releasing or venting negative emotions
and being freed from psychological discomfort. Disclosers do not expect their

relationships with the listeners to grow through reciprocal disclosures.

47

Figure 2-2. SIP-based Model of Perceived Anonymity Affecting Online Public Disclosure

  
  
  

   
 

     
 

Online Public
Disclosure

Perceived
Anonymity

 

Evaluation
Concern

In the SIP-based model, contrarily, people expect that their relationships continue

 

to grow. They use self-disclosure as a strategy to obtain information about others
(Tidwell & Walther, 2002). They weigh costs and rewards of self-disclosure in
relationship developments. Applying it to the context of online social support
communities, members participate in expectation of satisfying their need for social
support. Members with high social support need may perceive higher utilities of self-
disclosure as a means to induce desired responses. In fact, revealing intimate personal
facts is a popular strategy to receive emotional support from other members (Campbell,
2002; Preece, 1999). Dissatisfaction with their off-line social contacts causes them to
seek alternative online sources (Walther & Boyd, 2002). Perceived rewards of self-
disclosing would be higher than perceived risks for them. Higher need for social support
increases online public disclosure by increasing the predicted outcome value of
interactions (Sunnafrank, 1986). Need for social support is deﬁned as subjective
perception of the extent to which there are few people in one’s social circle who are

available when one is in need of social support.

48

H4: Need for social support increases online public disclosure.

It can be inferred that increased need for social support will lead online
community members to perceive more similarities with other members who participate in
the same community. According to URT, similarities reduce uncertainty (Berger &
Bradac, 1982). In reducing uncertainty, people create impressions about the self and
others. The less identity information they perceive exchanged about each other, the more

uncertain they are about the communication situation.

H5: Need for social support decreases perceived anonymity.

In the SIP model which originally dealt with interpersonal relationships in small
groups, community size was not included as affecting CMC users’ communication pattern.
Community size, however, has been found to affect the dynamics of community
interaction, conversational strategy, information overload, social loaﬁng, and member
sustainability (Butler, 2001; Markus, 1987; Morris & Ogan, 1996; Whittaker, Terveen,
Hill, & Cherny, 1998). In the context of online communities, which involve large group
communication, members have fewer opportunities to participate and less time to interact
(Butler, 2001). Humans’ limited information processing capacity and scarce time
availability cannot sustain remembering and interacting with all members. Perceived
community size, therefore, is added to the SIP-based model as a controlling variable.
Perceived community size is deﬁned as the perceived number of members who visit an

online community.

49

Variances in perceived community size are not always expected to be small within
the same community. Many online communities using bulletin board systems operate via
multiple discussion threads, which help members more easily to ﬁnd and engage in topics
of their interests. Therefore, even within the same community, the perceived number of
members would differ according to which discussion threads the member frequently
participates in. Figure 2-3 presents a reﬁned SIP-based model with need for social
support as an important predictor for perceived anonymity and online public disclosure
and perceived community size as a controlling variable.

Finally, a research question is advanced from SIDE theory. As mentioned earlier,
SIDE theory places more emphases on group salience than on anonymity. The effects of
anonymity in CMC are conditional on which identity salience is prevalent, personal or
group. Those who share a common identity or fate (common identity group, for example,
same ethnicity or suffering the same illness) show more normative behaviors than those
who casually gather based on attraction to each other (common bond group, for example,
friends or other social gathering) (Postrnes & Spears, 2000; Prentice, Miller, & Lightdate,
1994; Sassenberg, 2000, 2002). Online communities are characterized by common
interests. Especially, members of OSSCs gather because of the common fate they face.
Group salience tends be higher than casual dating communities or hobby communities.
Relationships in the above model may differ according to members’ identiﬁcation with

the community.

RQl: How will members’ identiﬁcation with the community affect the relationship

between perceived anonymity and online public disclosure?

50

Figure 2-3. Reﬁned SIP-based Model
Need for
Social Support

Perceived H lb: -

H4: -

Anonymity ' Disclosure

I
I
: H2b: ’ , . . ’
, x .
. l ' I
1 ’ r
. ' ’ l
I """" L ''''' , a '
: Perceived . z 11’ _________________________
{Community Size:
I ...............
' ------------ '
i .‘ “““ > controlling variable

51

H5: +

    
 

Online Public

H3b: +

 

l

Evaluation
Concern

CHAPTER 3
METHODS
Research Site

The present study employed two data collection methods, a qualitative pretest for
initial item generation, and an online survey for scale validation and nomological network
test. Subjects for the qualitative pretest and the online survey were recruited in an online
social support community, MissyUSA (www.missyusa.com).

MissyUSA is an online community for married Korean women who currently live
in the United States, or will marry and live in the United States within six months. They
are wives of students, students themselves, career women, married to Americans or not.
The community started in November 2000 at a commercial community portal site and
now has its own independent site. It is an asynchronous, text-based community using
computer-based bulletin board systems. The community serves more than 57,000
members, and 50 to 60 new members join each day, as of July 2005. MissyUSA has seven
major sections: Talk Lounge, Healthy Beauty, Home & Food, Motherhood, US Info,
Town Zone, and Missy School. As shown in the section titles, members exchange
informational and emotional social supports. Members of MissyUSA share several
demographics — Korean, married, female, and resident in the United States. They moved
to a foreign country where everyday activities cause them all kinds of unexpected
problems ranging from relationships with in-laws, homesickness, marital problems,

disputes with phone or insurance companies, how to cook traditional Korean food, and

52

parenting and childbirth. Advice and social approval from those who already went
through similar difﬁculties are extremely helpﬁil and alleviate emotional distress.

The nature of discourse in MissyUSA is primarily social support. Members often
confess their shameful experiences, hurt feelings, upset emotions, and wrongdoings as
well as share heartwarming stories and joyful news in their lives, and exchange useful
information, gossip about celebrities and humor. Those who reveal negative aspects of
their lives seek emotional catharsis and social validation from those who may understand
them better. Responses to such postings are usually positive — warm, emphatic,
supportive, encouraging, and caring. Other members reply how much they can
understand the confessor, or that they also had the same experience. Responses are
constructive as well. Replies which begin with emphatic tones sometimes include fair
judgments on what the confessor did wrong, and sincere advice about how to improve the
situation. If confessional postings contain controversial topics such as extramarital
relationships and religions, message tones become critical.4

By recruiting subjects from one community, the researcher can prevent the study
from being confounded with other community-related variables such as types of
communities. For example, online dating communities and fan club communities may
differ in terms of the nature of relationships that people expect to develop. In addition, the
fact that members are all women is also beneﬁcial to the present study. One of the
important predictors for self disclosure is gender. Since all subjects are women, there

would be no concern for a confounding effect by gender.

 

" This description is based on the researcher’s observation as a member. The owner of MissyUSA
did not allow analysis or citation of actual postings.

53

Initial Item Generation and Reﬁnement

The validity of a scale should start with initial item construction (Nunnally &
Bernstein, 1994). Nunnally and Bernstein’s “domain sampling” suggests that researchers
start with concept explication, specify domains that constitute the concept, and carefully
select potential items for each domain (dimension). Ideally, after domain sampling,
researchers select items reﬂecting each domain from previous studies, drop conceptually
overlapped items, or create new items representing domains missing in previous literature,
and reword items to be relevant to a study context (Bhattacherjee, 2002). However, no
prior scale for perceived anonymity exists. Therefore, the present study created items for
each perceived anonymity dimension based on the ﬁve categories of identity information
classiﬁed earlier (name or pseudonym, locatability, biographic information,
communication pattern and style, audio-visual information).

To ensure the content validity of items, a qualitative pretest was performed with
eight5 MissyUSA members between May 12 and 28, 2005 who were recruited off-line in
Greater Lansing, Michigan. They had been membesr of MissyUSA for a minimum of one
year and visited it at least twice a week. Given index cards of perceived anonymity items
along with items for other concepts in this study (online public disclosure, evaluation
concern, need for social support, group identiﬁcation), the pretest subjects were asked to
sort them according to concepts and deﬁnitions, and to evaluate how well each item

represents the relevant concept on a seven-point Likert scale (Appendices A and B). The

 

5 Usually, ﬁve to eight is said to be enough for homogenous qualitative samples.

54

sorting was conducted in places of subjects’ choice to ensure subjects’ convenience and
psychological comfort (for example, the subject’s residence).

Candidate items for online public disclosure, need for social support and group
identiﬁcation were modiﬁed from existing scales: online public disclosure from
Wheeless’ Revised Self-Disclosure Scale (RSDS) (Rubin, Palmgreen, & Sypher, 1994),
need for social support from the Interpersonal Support Evaluation List (ISEL) (Cohen, et
al., 1985), and group identiﬁcation from Arrow-Carini Group Identiﬁcation Scale (Henry,
Arrow, & Carini, 1999) and Luhtanen and Crocker’s Collective Self-Esteem Scale (1992).
The control of depth and the positive-negative dimensions of RSDS were used to create
items for online public disclosure. Among the 16 items of the short version of ISEL,
items that emphasize emotional social support were included. The ISEL scale stresses
tangible forms of support which is the least relevant to online social support (e. g.
borrowing a car, quick emergency loan, or help in moving). Additional items that are
speciﬁc to MissyUSA (e.g. I wish I had someone who listens to me when I struggle with
my life in the US.) were included. Candidate items for evaluation concern were newly
created to make them more appropriate for the context of MissyUSA. Candidate items are
listed in Table 3-1.

All subjects sorted item cards correctly. Based on subjects’ rating and feedback,
items were rephrased or dropped for clarity. If more than half the subjects rated an item
the lowest, it was dropped after comments from the subjects were reviewed. Table 3-1
lists the ﬁnal sets of items to be included for scale validation and nomological network

test.

55

Table 3-1. Candidate and Final Items by Concept

 

 

 

 

 

 

 

 

Concept # of # of Item Final items
candidate ﬁnal No. (excluded through pretest)
items items
Self- l7 l3 sal Some members can recognize my name. (R)
Anonymity sa2 Some members can recognize my usemame.
(R)
sa3 Some members may ﬁnd out my email address
or homepage address. (R)
(# Some members may ﬁnd out my mail
address or telephone number.)
sa4 Some members can recognize my IP address.
(R)
sa5 Some members can guess how old I am. (R)
sa6 Some members can tell my marital status. (R)
sa7 Some members can tell my profession. (R)
sa8 Some members can tell how much education I
have had. (R)
539 Some members can tell our household income
level. (R)
(# Some members can tell how many children
I have and their age.)
salO Some members can tell my hobbies or
interests. (R)
sall Some members can recognize me from my
writing style. (R)
sa12 Some members can recognize me from
expressions or words I use frequently. (R)
(# Some members can recognize me from the
way I approach the topic covered.)
(# Some members may imagine my
appearance.)
sal3 Some members may match me with pictures I
' posted. (R)
Other— l7 13 cal I can recognize the names of some members.
Anonymity (R)
oa2 I can recognize usemames of some members.
(R)
oa3 I may ﬁnd out email addresses or homepage
addresses of some members. (R)
(# I may ﬁnd out mail addresses or telephone
numbers of some members.)
oa4 I can recognize some members via their IP
addresses. (R)
oa5 Sometimes, I can guess how old other
members are. (R)
oa6 Sometimes, I can tell the marital status of
other members. (R)
oa7 Sometimes, I can tell the profession of other

members. (R)

 

56

 

Table 3-l. (continued)

 

 

 

 

 

 

 

 

 

 

Concept # of # of Item Final items
candidate ﬁnal No. (excluded through pretest)
items items
Other- l7 l3 oa8 Sometimes, I can tell how much education
Anonymity other members have had. (R)
oa9 Sometimes, I can tell the household income
level of other members. (R)
(# Sometimes, I can tell how many children
other members have and how old they are.)
oalO Sometimes, I can tell hobbies or interest of
other members. (R)
oall I can recognize some members from their
writing styles. (R)
oa 12 I can recognize some members from
expressions or words they use frequently. (R)
(# I can recognize some members from the
way they approach the topic covered.)
(# Sometimes, I can imagine their
appearance.)
oal3 Sometimes, I can match other members with
pictures they posted. (R)
Online Public 8 8 opdl I am willing to reveal negative things about
Disclosure myself.
opd2 I am willing to express my most intimate
feelings.
opd3 I am willing to share what I did wrong.
opd4 I am willing to share what I would not do with
my family, my off-line friends and colleagues
at work.
opd5 I am willing to talk about my shameful
experiences.
opd6 I am willing to talk about my hurt feelings.
opd7 I am willing to talk about my failures.
opd8 1 am willing to share my family history or
secrets.
Evaluation 9 5 ecl Other members will criticize what I posted.
Concern ec2 Other members will misunderstand me.
ec3 Other members will dislike what I posted.
ec4 Other members will disagree with me.
(# Other members will laugh at me.)
(# Other members will disapprove of what I
posted.)
ec5 Other members will oppose what I posted.
(# I will be rejected for what I posted.)
(# I will be ridiculed for what I posted.)
Group 11 7 (# I feel I do not have much to offer this online
Identiﬁcation community.)
gil I feel I am one of the least contributing

members in this online community. (R)
(# I regret I joined this online community.)

 

57

 

Table 3-1. (continued)

 

Concept

# of
candidate
items

# of
ﬁnal
items

Item
No.

F inal items
(excluded through pretest)

 

Group
Identiﬁcation

11

gi2

(# I do not tell anyone that I am a member of
this online community.)

I feel that this online community is
worthwhile.

(# I am ashamed to be a member of this online
community.)

My membership in this online community has
little to do with how I feel about myself. (R)

I think of this online community as part of
who I am.

I see myself as different from other members
of this online community. (R)

I often cite this online community when I talk
to others off-line.

I enjoy interacting with the members of this
online community.

 

Need for
Social
Support

 

 

12

 

 

nssl

nssZ

nss3

nss4

nssS

nss6

nss7

nss8

nss9

 

I wish I had someone who listens to me when I
struggle with my life in the US.

I wish I had someone who listens to me when I
have a marital problem.

I wish I had someone who listens to my
complaints about in-laws.

(# I wish I had someone who can provide
some advice when I have relational problems
with my friends or colleagues.)

I wish I had someone whom I can ask for
advice when things go wrong.

I wish I had someone with whom I can talk
about my problems.

I wish I had I someone who helps me decide
things.

I wish I had someone whose advice I really
trust

I wish I had someone who can provide
objective feedback about how I am handling
my problems.

(# I wish people had conﬁdence in me.)

I wish I had someone with whom I can share
my most private worries and fears.

(# I wish I were as close to my friends as are
many other people.)

 

Note. (R) indicates “reverse-coded in the analysis.”

58

 

Scale Validation and Nomological Network Test
Measures

Subjects were asked to answer how much they agree to each item in the ﬁnal set of
measures for self-anonymity (SA), other-anonymity (OA), online public disclosure (OPD),
evaluation concern (EC), group identiﬁcation (GI) and need for social support (N SS)
(Table 3-1). The items were anchored with 7-point Likert type scales from strongly
disagree, 1 to strongly agree, 7. Items for OPD, EC, GI, and N88 were analyzed for alpha
reliabilities (OPD, alpha = .957; EC, alpha = .955; NSS, alpha = .971). Only three of
seven GI items remained (alpha =.687).6 Only these items were included in further
analyses.

Perceived community size was measured by asking how many people they think
visit MissyUSA per day. Subjects were asked to choose from a set of response choices
that increased by 1,000 people incrementally. This controlling variable was included as
an observed variable. All measurements were translated into Korean, and the
questionnaire was revised based on comments from three doctoral students who are

bilingual.

 

6 Cronbach alpha values greater than .6 are considered adequate for exploratory work (Nunnally,
1994).

59

Data Collection

A total of 301 MissyUSA members participated in the survey between June 22 and July 3,
2005. Subjects were recruited through a banner advertisement put on the main page of
MissyUSA.7 Subjects who clicked the banner were led to a consent form page of an
online survey site (www.cmcresearch.net/survey_t.html) (Appendix C).

The recruitment banner advertisement consisted of four main pages. The ﬁrst two
contained information about the title of the study (abbreviated from the original title with
simple words in order to help subjects understand and to ﬁt the limited space of the
banner), the purpose of the study, the eligibility criterion for participation,8 and the name
and contact information (email address) of the researcher. The third page informed
potential subjects of compensation for participation. The last page included a link for
detailed information about the survey (Appendix D).9

The online survey consisted of four sections: (1) subjects’ general experiences in
MissyUSA; (2) how subjects feel about themselves when participating in MissyUSA; (3)
how subjects feel about themselves in their daily lives; and (4) demographic information
(Appendix E). ’0 A “conﬁrmation” page was added at the end of the survey (Appendix F).

In this page, subjects were provided the numbers of questions to which they did not

 

7 The owner of MissyUSA requested not to use message boards to recruit subjects. Posting
messages of subject recruitment was considered obtrusive and irrelevant in topical message boards.

8 The owner of MissyUSA requested not to use the name of the community in the banner (as well
as in the online survey itself) because the use of the community name would be dispiriting.
Therefore, the eligibility information was given as “members of an online community.”

9 The banner was approved by the owner of the MissyUSA in advance.

’0 The survey items and wording was approved also by the owner of MissyUSA in advance.

60

answer, and an opportunity to go back to unanswered items and respond to them, or
change their answers if they so chose.

Subjects were compensated with an Amazon.com email gift certiﬁcates of U885
each if they completed the survey. 1’ Participants who voluntarily provided email
addresses in the survey received gift certiﬁcates by email a week after the survey
completion.

The mean age of subjects was 31.73 years (SD=3.56), and they had stayed in the
United States for 59.26 months on average (SD=44.16). The average membership was
26.39 months (SD=I4.51). Most of the subjects had college or higher degrees (97.3%),

and had annual income less than US$60,000 (62.5%) (Table 3-2).

Table 3-2. Characteristics of Subjects

 

 

 

 

 

 

 

 

Age (years) Mean = 31.73 Range = 30 (18, 48)
Median = 32
Mode = 32
SD = 3.56
US stay (months) Mean = 59.26 Range = 233 (0, 233)
Median = 48
Mode = 48
SD = 44.16
MissyUSA membership Mean = 26.39 Range = 83 (1, 84)
(months) Median = 24
Mode = 24
SD = 14.51
Education High School graduates 7, 2.3
(members, %) College degree 184, 61.1
Master’s or Doctoral degree 109, 36.2
None of the above 1, 0.3
Annual Income Less than US$30,000 86, 28.6
(members, %) US$30,000 to 59,999 102, 33.9
US$60,000 to 99,999 72, 23.9
More than US$100,000 41, 13.6

 

 

” US$5 is the minimum dollar amount for an email giﬁ certiﬁcate available at Amazoncom.

61

Data Analysis

Data analysis for scale validation and nomological network test was conducted
with structural equation modeling (SEM) using Amos 4. First, an exploratory factor
analysis (EFA) was performed on perceived anonymity (PA) items (that is, self-
anonymity and other-anonymity items) as a preliminary step before a conﬁrmatory factor
analysis (CFA) using SPSS 11.5. Second, a second-order CFA was performed on the
remaining PA items. Convergent and discriminant validities were evaluated. Third, two
competing models — deindividuation and SIP-based models, were tested. Fourth, a
multiple group analysis was performed on the reﬁned SIP-based model. A composite
score was used to categorize low and high group identiﬁcation groups in order to
investigate group differences in path coefﬁcients.

Mean substitution was used to replace missing data since the number of missing
data was less than ﬁve percent for all variables. Outliers were replaced with a value of 3
SD from each mean. The correlations, means, and standard deviations of scales are

presented in Table 3-3, and those of all observed variables in Appendix G.

62

Table 3-3. Correlations, Means, and Standard Deviations for Scales

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SA OA DA opo EC NSS GI
SA
CA .392**
DA .561** .487**
OPD -.162** —.132* -.204**
EC -.295** -.138* -.284** .156*
NSS -.127* -.159** -.103 .259" -.009
GI -.158** -.191** -.164** .325" -.019 .328**
Mean 35.500 24.093 18.017 31.913 13.702 47.705 14.846
Median 36.000 24.000 18.000 33.000 13.000 49.000 15.000
Mode 38.000 20.0003 18.000 40.000 10.000 63.000 16.000
so 10.238 7.875 5.850 12.090 6.010 12.111 3.782
Range 48 36 24 48 30 54 18
(min, max) (8, 56) (6, 42) (4, 28) (8, 56) (5, 35) (9, 63) (2, 21)
Alpha reliability .888 .906 .885 .957 .955 .971 .687

 

 

Note. a. Multiple modes exist. The smallest value is shown.

All statistics for composite scores.
SA, Self-Anonymity; OA, Other-Anonymity; DA, Discursive Anonymity;

OPD, Online Public Disclosure; EC, Evaluation Concern;
NSS, Need for Social Support; GI, Group Identiﬁcation

63

CHAPTER 4

RESULTS

Exploratory Factor Analysis on Perceived Anonymity

A total of 13 self-anonymity (SA) and 13 other-anonymity (OA) items that were
generated through a qualitative pretest were entered into an exploratory factor analysis
(EFA) using an Oblimin rotation method (assuming an interrelationship between factors).
The principal component analysis identiﬁed ﬁve factors having eigen values larger than
1.0. Table 4-1 presents the factor loadings of the ﬁve factors, which accounted for 66.3%
of the variance. To facilitate factor interpretation, only those factor loadings with values
greater than 0.4 were reported. Items sa6, salO, 5313, and oa13 had low loadings
from .371 to .473, and 033 was cross-loaded on Factors 4 and 5. Excluding these ﬁve
items, another EF A identiﬁed three factors with eigen values larger than 1.0. The scree
plot of the ﬁrst EFA (with ﬁve factors identiﬁed) (see Figure 4-1), also indicated that the
three-factor solution was more appropriate. The factor loadings of the three-factor
solution are presented in Table 4-2. The three factors accounted for 61.2% of the variance.
Only loadings with factor scores greater than 0.4 are reported to facilitate factor

interpretation.

Table 4-1. Factor Loadings -- Five-Factor Solution

 

F1

F2

F3

F4

F5

 

In this online community,
sa : I feel that some members can
oa : I feel that I can

sa7, tell my profession.

sa3, ﬁnd out my email address or homepage address.
saS, guess how old I am.

sa4, recognize my IP address.

sa9, tell our household income level.
sa8, tell how much education I had.
sa2, recognize my usemame.

sal, recognize my name.

sa6, tell my marital status.

sal3, match me with pictures I posted.
sa10, tell my hobbies or interests.

oa6, tell the marital status of other members.

oa7, tell the profession of other members.

038, tell how much education other members had.

oa5, guess how old other members are.

oa9, tell the household income level of other members.
oa10, tell the hobbies or interests of other members.

sa12, recognize me from expressions or words I use
frequently.

oal I, recognize some members from their writing
style.

oa12, recognize some members from expressions or
words they use frequently.

sal I, recognize me from my writing style.

oal , recognize names of some members.

oa2, recognize usemames of some members.

oal3, match other members withpictures they posted.
oa4, recognize some members via their IP addresses.
oa3, ﬁnd out email addresses or homepage addresses
of some members.

 

 

.736
.705
.703
.701
.669
.662
.660
.610
.473
.432
.405

 

.858
.826
.800
.782
.670
.666

 

.769

.739

.738
.708

 

.909
.755

.421

 

 

65

 

Table 4-2. Factor Loadings -- Three-Factor Solution

 

Factor 1 Factor 2 Factor 3

 

In this online community,
sa : I feel that some members can
oa : I feel that I can

sa2, recognize my usemame. .770
sal, recognize my name. .748
sa7, tell my profession. .732
sa3, ﬁnd out my email address or homepage address. .726
sa5, guess how old I am. .700
sa9, tell our household income level. .658
sa8, tell how much education I had. .650
sa4, recognize my IP address. .637
oa6, tell the marital status of other members. -.858
oa7, tell the profession of other members. -.854
oa8, tell how much education other members had. -.815
oa5, guess how old other members are. -.781
oa9, tell the household income level of other members. -.702
oaIO, tell the hobbies or interests of other members. -.644

oa12, recognize some members from expression or

words they use ﬁequently. .879
oal I, recognize some members from expressions or

words they use frequently. .856
sa12, can recognize me from expressions or words I use

frequently. .681
sal I, recognize me from my writing style. .634
oa2, recognize the names of some members. .603
oa4, recognize some members via IP addresses. .525

 

 

 

 

oal, recognize the names of some members.

 

66

 

Figure 4-1. Scree Plot of First EFA

 

 

 

Eigenvalue

 

 

Component Number

Items loaded on Factor 1 were all SA items, especially, items for name (ID, name),
locatability (email address, IP address), and biographic information (job, age, income,
education) among the ﬁve types of identity information. Factor 2 includes items for OA.
Unlike in Factor 1, items only for biographic information (marriage status, job, education,
age, income, hobby) consisted of Factor 2. Factor 3 was a combination of items for
communication pattern and style in both SA and CA (writing style, frequently used words
and expressions), an item for OA name (ID), and an item for OA locatability (IP address).
Item oal was not independently loaded to any of the three factors.

The results of the second EF A pointed to a three-factor model that differs from the
originally hypothesized two-factor model. The major point of divergence is on the
separation of communication pattern and style from SA and OA. The ﬁndings suggest

that items for SA and OA communication pattern and style might represent a third factor.

67

To conﬁrm these EF A analyses, a series of conﬁrmatory factor analyses (CFAs) were

performed on this three-factor solution.

Conﬁrmatory Factor Analysis on Perceived Anonymity

A series of second-order CF As were conducted on the three factors generated from
the second EF A. To set scales for the second-order factors, the variance of PA was ﬁxed
to 1.0. Equality constraints were put on residual variances for factors 1 and 3, using the
critical ratio difference method in order to solve the just-identiﬁcation problem at the
upper level of the CFA model (Byme, 2001).12 The ﬁndings are presented in Table 4-3
and Figure 4-2. In the present study, parameters were estimated using Maximum
Likelihood Method. Overall goodness-of-ﬁt was evaluated using multiple indices of
goodness-of-ﬁt rather than the goodness—of-ﬁt chi-square, which is considered over-
restrictive as an evaluation of good-ﬁt, due to its sensitivity to sample size (Kline, 2004).
The indices adopted in the present study were the normed chi-square (Xz/df; Carmines &
McIver, 1981), the comparative ﬁt index (CFI; Bentler, 1990), the normed ﬁt index (NF 1;
Bentler & Bonnet, 1980), the nonnormed ﬁt index (NNFI; Marsh, Balla, & McDonald,
1988), and the root mean square error of approximation (RMSEA: Hu & Bentler, 1999).
Values greater than .9 for CFI, NP], and NNFI (Bentler, 1990), and less than 3 for XZ/df

(Kline, 2004) are considered to be a good ﬁt, whereas values less than .08 for RMSEA

 

'2 It is critical that the identiﬁcation status of the higher order portion should be checked ﬁrst when
hierarchical models are tested. With only three ﬁrst-order factors, the higher order structure of the
present C F A model is just-identiﬁed unless a constraint is put on at least one parameter in the
upper level of the model. To address this identiﬁcation issue, the differences between residual
variances were examined, and residual variances for factors 1 and 3 were found equal in the
population. Equality constraints were placed on the two residual variances.

68

indicate that there is adequate ﬁt (Hu & Bentler, 1999), although values approaching .95
for the ﬁrst three indices and .05 for RMSEA are preferred.

For the ﬁrst second-order CFA (CFA PA-l),l3 all indices except for RMSEA
indicated a good model ﬁt (XZ/df =.5.481, CF I=.954, NFI=.945, NNFI=.942,
RMSEA=.122) (see Table 4-3). However, ﬁrst-order factor loadings from Factor 3 to
items oa2 and oa4 were very low (.436 and .539 respectively). Factor loadings should be
greater than .70 for convergent validity (F omell & Larcker, 1981). Accordingly, another
CFA was performed without the two low factor loading items (CF A PA-2). The lowest
factor loading was .65 (Factor 1 to sa9). Excluding items oa2 (OA, ID) and 034 (OA, IP
address), Factor 3 now includes items for communication pattern and style only. Fit
indices also showed similar results to the ﬁrst CFA’s (Xz/df = 6.370, CFI=.952, NFI=.943,
NNFI=.937, RMSEA=.134) (see Table 4-3).

The values of XZ/df and RMSEA recommended modiﬁcation of the model CF A
PA-2 (6.370 and .134, respectively). To modify CFA PA-2, a covariance between e16
and e17 was added to CFA PA-2-1, following a modiﬁcation index (see Table 4-4).
Substantively, this covariance makes sense because items oa12 and oall measure the
same type of identify information (communication pattern and style, OA). Another
covariance between e4 and e8 was added to CFA PA-2-2. The covariance was accepted
since items sal (name) and sa2 (ID) are the same type of SA — name. A third error
covariance between e2 and e6 was added to CF A PA-2-3. Items sa7 (job) and sa8
(education) are items for SA biographic information. As demographic items, these two

items are closely related. The last error covariance was added to CFA PA-2-4. Items sa9

 

’3 To set scales for the ﬁrst-order factors, the highest loading in each factor (sa2 for factor 1, 036
for factor 2, and sa12 for factor 3) was ﬁxed to 1.0.

69

(SA) and oa9 (OA) are for education. Table 4-4 presents how much each modiﬁcation
improved a previous model. The ﬁt indices ﬁnally reached the cutoffs in CF A PA-2-4
(Xz/df =2.926, C FI=.983, NFI=.975, NNFI=.978, RMSEA=.080), therefore, any further

modiﬁcation st0pped. Seven of 18 ﬁrst-order factor loadings were lower than .70. The

lowest ﬁrst-order factor loading is .60 from Factor 3 to oa12 (see Figure 4-2).

Table 4-3. Fit Indices by Model

 

 

 

 

 

 

 

 

 

Model X2 df Xz/df X?diff Jim CFI NFI NNFI RMSEA
CFA PA-l 915.408* 147 5.481 .954 .945 .942 .122
CFAPA-Z 840.833* 132 6.370 .952 .943 .937 .134
CFA 551.400* 131 4.209 289.433* 1 .971 .963 .963 .103
PA-2-l

CFA 456.492* 130 3.511 94.908* 1 .978 .969 .971 .091
PA-2-2

CFA 406098“ 129 3.148 50.394* 1 .981 .973 .975 .085
PA-2-3

CFA 374.555* 128 2.926 31.543* 1 .983 .975 .978 .080
PA-2-4

 

 

 

 

 

 

 

 

 

 

Note. C FA PA-2-l to 2-4 compared to its previous model, respectively.

*p<.05

Table 44. Modiﬁcation Indices by Model

 

 

 

 

 

 

 

Model Description of parameter Modiﬁcation index
X2 (1)

CFA PA-2 b/w e16 and e17 203.426 *

CFA PA-2-l b/w e4 and e8 83.988 *

CFA PA-2-2 b/w e2 and e6 44.488 *

CFA PA-2-3 b/w e5 and e13 29.357 *

 

 

 

 

Note. Only the highest index is reported in each model.
* p < .05

7O

Figure 4-2. Final CFA Model (Model CFA PA-2-4)

'79 .69 sa7

        

sa3

9969

 
  

 
 
  

é

  

sa4

.84
.35

.80
.88

.50 ® .82 038
0 a5

.81

39 896

O

\l
O
U)
D)
—l

' be
3 o

.9]

.93

O
in
—i
N

@$@ @3699

3:

Signiﬁcance: p < .001

Fit indices: ,
X2 (129) = 375.365 NFI =.975
X2/df= 2.910 NNFI =.978
CFI = .983 RMSEA =.080

71

Convergent validity was evaluated for the three factors based on using three
criteria recommended by Fomell and Larcker (1981): (1) all measurement factor loadings
must be signiﬁcant and exceed .70, (2) construct reliabilities must exceed 0.80, and (3)
average variance extracted (AVE) by each construct must exceed the variance due to
measurement error for that construct (that is, AVE should exceed 0.50). All indicator
factor loadings were signiﬁcant at the 0.05 signiﬁcance level, and the lowest value
was .60 (from Factor 3 to oa12) (see Figure 4-2 and Table 4-5). Construct reliability
ranged from .856 to .906. Cronbach alphas ranged between .886 and .906. AVE ranged
from .501 o .618 (Table 4-6). The results satisﬁed all criteria except for the size of factor
loadings. Even though seven of 18 factor loadings were smaller than .70, they were not
smaller than .60 and satisﬁed Anderson and Gerbing’s (1988) criteria that an individual
item’s standardized coefﬁcient is signiﬁcant, namely greater than twice its standard error
(i.e., t-value >2). Table 4-5 shows that coefﬁcients for all items greatly exceed twice
their standard error.

For discriminant validity, Fomell and Larcker (1981) suggested that AVE for
- each construct should exceed the square correlation between any pair of constructs. The
highest squared correlation was. .465 between Factors 1 and 3, which was smaller than the

lowest AVE (.501 for Factor 1). Therefore, the test for discriminant validity was met.

72

Table 4-5. CFA Results

 

 

 

Factor Item Mean Standard Standardized t-Statistic
deviation factor load'gg for FL
Factor 1 sa5 4.096 1.637 .789
sa7 4.286 1.700 .689 12.141
sa3 4.472 1.779 .770 13.839
sa2 4.412 1.806 .667 11.696
sa9 5.096 1.463 .658 1 1.882
sa8 4.286 1.544 .638 11.094
sa4 4.663 1.821 .711 12.625
sal 4.243 1.902 .628 10.914
Factor 2 oa6 3.580 1.678 .802
oa7 4.116 1.578 .883 17.569
oa8 4.123 1.588 .820 15.926
oa5 3.993 1.545 .796 15.304
oa9 4.468 1.597 .702 13.401
oa10 3.794 1.551 .700 12.962
Factor 3 sa12 4.385 1.669 .905
oa12 4.781 1.704 .600 11.591
oall 4.631 1.714 .632 12.471
sall 4.219 1.691 .929 21.679

 

 

 

 

 

 

 

Table 4-6. Convggent and Discriminant Validity

 

Factor correlations (squared)

 

 

 

 

 

 

 

 

 

 

# of Cronbach Construct
Factor items Alpha Reliability AVE Factor 1 Factor 2 Factor 3
Factor 1 8 .888 .889 .501
Factor 2 6 .906 .906 .618 .424 (.180)
Factor 3 4 .885 .856 .610 .682 (.465) .405 (.164)

 

 

 

Note. Construct Reliability (CR) = (2102/ [(2102 + 81-23)]

Average Variance Extracted (AVE) = 2112/ [2A2 + 2(1-A’)1

73

 

Contrary to the hypothesized two-factor structure, the EF A and the second-order
CFA analyses generated a three-factor solution. Examining items revealed themes for
each factor. Factor 1 included SA items for name (ID, name), locatability (email address,
IP address), and biographic information (job, age, income, education) among the ﬁve
types of identity information. Factor 2 included items for OA. Unlike in Factor 1, items
only for biographic information (marriage status, job, education, age, income, hobby)
consisted of Factor 2. Factor 3 was a combination of items for communication pattern
and style in both SA and DA (writing style, frequently used words and expressions).
Therefore, Factor 1 was named as self-anonymity (SA), Factor 2 as other-anonymity (OA),
and Factor 3 as discursive anonymity (DA) (see Table 4-7). Of interest points are: (1) DA
was identiﬁed as a separate factor from SA and CA; (2) among the ﬁve types of identify
information, audio-visual information was not included in any of the three factors,
suggesting that this type was not a contributing factor in PA, or was a distinct factor unto
itself; and (3) only items of biographic information were included in OA, implying that
OA was not decided by name and locatability information. The discrepancy between SA
and OA indicates unbalanced anonymity perception between self and others. It reﬂects a
tendency that people are more sensitive to their than to others’online privacy.

Among the three sub-dimensions of PA, OA seemed the least contributing
dimension. Its second-order path from PA was the lowest (Beta = .50) while the other
two paths were of the same size (PA to SA, Beta =.84; PA to DA, Beta =.81). This result
reﬂects that online community members tended to be self-focused.

The separation of DA from SA and DA demonstrated the importance of the
communication factor in determining anonymity perception. It concurred with previous

studies that deal with discursive anonymity separately (Anonymous, 1998; Scott, 1999).

74

Table 4-7. Final Items by CFA

 

 

 

 

 

 

 

 

 

Factor Item Item Type of
No. Identity
Info.
SA sal Some members can recognize my name. N
sa2 Some members can recognize my usemame. N
sa3 Some members may ﬁnd out my email address or homepage
address. L
sa4 Some members can recognize my IP address. L
sa5 Some members can guess how old I am. BIO
sa7 Some members can tell my profession. BIO
sa8 Some members can tell how much education I have had. BIO
sa9 Some members can tell our household income level. BIO
OA oaS Sometimes, I can guess how old other members are. BIO
oa6 Sometimes, I can tell the marital status of other members. BIO
oa7 Sometimes, I can tell the profession of other members. BIO
oa8 Sometimes, I can tell how much education other members have
had. BIO
oa9 Sometimes, I can tell the household income level of other
members. BIO
0310 Sometimes, I can tell hobbies or interests of other members. BIO
DA sall Some members can recognize me from my writing style. CPS
sa12 Some members can recognize me from expressions or words I
use frequently. CPS
oall I can recognize some members from their writing styles. CPS
0312 I can recognize some members from expressions or words they
use frequently. CPS
Note. All items were reverse-coded.

SA, Self-Anonymity; OA, Other-Anonymity; DA, Discursive Anonymity

N, Name or Pseudonym; L, Locatability; BIO, Biographic information;
CPS, Communication Pattern and Style

The deindividuation model was tested, and the results are presented in Figure 4-3.

Perceived Anonymity Affecting Online Public Disclosure

The model exhibited good ﬁt with observed data (XZ/df =2.487, CF I=.976, NF I=.960,

NNFI=.972, RMSEA=.070). Of greater interest are the path estimates and variance

explained in each dependent variable. Two structural paths, perceived anonymity to

online public disclosure and evaluation concern were signiﬁcant at the .05 signiﬁcance

level (Hla, Beta = -.22 and H2a, Beta = -.35). As hypothesized, perceived anonymity

75

 

decreased evaluation concern (H23). Contrary to hypothesis, however, perceived
anonymity negatively affected online public disclosure (H l a). Evaluation concern to
online public disclosure (H3a) was not signiﬁcant.

The SIP-based model is an equivalent model of the deindividuation model (see
Figure 4-4). It is recommended that equivalent models be considered in SEM analysis
(Stelzl, 1986). Equivalent models yield the same predicted correlations or covariances,
but they do so with a different conﬁguration of paths among the same variables. For a
given path model, there may be many equivalent variations. A choice among equivalent
models should be based on theoretical rather than mathematical grounds. The SIP-based
model’s ﬁt indices showed the same results as the deindividuation model’s (XZ/df = 2.487,
CFI=.976, NFI=.960, NNFI=.972, RMSEA=.070). Two structural paths, perceived
anonymity to online public disclosure (Hlb, Beta = -.24) and perceived anonymity to
evaluation concern (H2b, Beta = -.33) were signiﬁcant in the hypothesized direction at
the .05 level. The path from online public disclosure to evaluation concern was in the
hypothesized (positive) direction, although not signiﬁcant (H3b).

The examination of model ﬁt indices and path signiﬁcance showed that the SIP-
based model represented the data better than did the deindividuation model. The subjects
of online community members were more willing to disclose negative aspects about
themselves through publicly posted messages as they perceived that they could recognize
each other more. Previous studies on self-disclosure in CMC maintained the mediating
role of evaluation concern. That is, anonymity decreases evaluation concern, which in
turn, decreases self-disclosure. The present study invalidated such an explanation by

supporting the SIP-based model over the deindividuation model. The large proportion of

76

variance left unexplained in each dependent variable suggests that other predictors may be

missing from the current model.

Figure 4-3. Deindividuation Model

q
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

 

-----------------------.
-------—— ---b-----------—--

Signiﬁcance: p < .001
Fit indices:

X2 (425) = 1056.918 NFI =.96O
X2/df= 2.487 NNFI =.972
CFI = .976 RMSEA =.070
* p < .05
PA, Perceived Anonymity OPD, Online Public Disclosure
SA, Self Anonymity EC, Evaluation Concern
OA, Other Anonymity
DA, Discursive Anonymity

77

Figure 4-4. SIP-Based Model

q
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
l
I
I
I
I
I
I
I
I
I
I
I
-

 

 

u.----—--—---—-----———-—-----
—-——----—- -u-—p-----—-----——

Signiﬁcance: p < .001

Fit indices:
x2 (425) = 1056.918 NFI =.960
XZ/df= 2.487 NNFI =.972
CFI = .976 RMSEA =.070
* p < .05

PA, Perceived Anonymity OPD, Online Public Disclosure
SA, Self Anonymity EC. Evaluation Concern

OA, Other Anonymity

DA, Discursive Anonymity

Finally, the reﬁned SIP-based model with need for social support as a predictor for
perceived anonymity and online public disclosure and perceived community size as a
controlling variable, was tested. Separate paths to SA, OA and DA were directed from
need for social support and perceived community size instead of single paths to PA. No
path could be directed to PA because the variance of PA was ﬁxed to 1.0 in order to set
scales.

Results for the three paths, PA to OPD, PA to EC, and OPD to EC, were

comparable to the SIP-based model. Hypotheses 4 and 5 were supported (H4, Beta = -.16

78

for SA, -. l 9 for OA, and -.13 for DA; H5, Beta = .26). Need for social support decreased
each sub-dimension of PA, and increased OPD as hypothesized (see Figure 4-5 and Table

4-8).

Figure 4-5. Reﬁned SIP-Based Model

 

 

.26 *
-.20* OPD
“all
-.34 ‘ .09
’ 1
EC
' R2=.14

 

Signiﬁcance: p < .001
Fit indices:

x2 (761) = 1717.651 NFI =.955
Xz/df= 2.257 NNFI =.971
CFI = .974 RMSEA =.065
* p < .05
PA, Perceived Anonymity OPD, Online Public Disclosure
SA, Self Anonymity EC, Evaluation Concern
OA, Other Anonymity NSS, Need for Social Support

DA, Discursive Anonymity PCS, Perceived Community Size

79

Table 4-8. Results b Hypothesis

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Hypothesis Standard. Unstandard. S.E. C.R.
Coefﬁcients Coefﬁcients

Hla PA (+7) OPD -.216 -.341 .112 -3.056 p<.05

H2a PA (-) EC -.347 -.410 .077 -5.346 <.05

H3a EC (-) OPD .079 .106 .085 1.248 ns

Hlb PA (-) OPD -.243 -.384 .104 -3.711 p<.05

H2b PA (-) EC -.329 -.389 .079 -4.911 p<.05

H3b OPD (+) EC .074 .056 .045 1.248 ns

I-Ilb‘I PA (-) OPD -.204 -.322 .101 -3.173 p<.05

H2ba PA (-) EC -.338 -.400 .079 -5.100 p<.05

H3ba OPD (+) EC .088 .066 .044 1.501 ns

H4a NSS (-) SA -.159 -.153 0.057 -2.676 p<.05
NSS (-) 0A -.191 -.190 0.059 -3.206 p<.05
NSSAQ DA -.126 -.141 0.066 -2.150 p<.05

HSa NSS (j) OPD .022 .303 .068 4.424 <.05
PCS (c) SA .124 .026 .013 2.047 p<.05
PCS (c) GA .161 .035 .013 2.725 p<.05
PCS (c) DA .103 .026 .015 1.726 ns
PCS (c) OPD .022 .006 .015 .380 ns
PCS (c) EC -.069 -.013 .011 -1.l91 ns

Note. a. Results from the reﬁned SIP-based model
(+) positive relationship hypothesized; (-) negative relationship hypothesized;
(c) controlled, not hypothesized

PA, Perceived Anonymity OPD, Online Public Disclosure

SA, Self-Anonymity EC, Evaluation Concern

OA, Other-Anonymity NSS, Need for Social Support

DA, Discursive Anonymity PCS, Perceived Community Size

Multiple Group Analysis

To assess the reﬁned SIP-based model in different GI (group identiﬁcation) groups
(RQI), measurement invariance was ﬁrst tested. A composite score for G1 was computed,
and categorized into low and high GI groups using median split. Items gi2, gi4, and gi7

consisted of GI (alpha = .687). ’4 The results are presented in Table 4-9.

 

'4 Reliability analysis showed that four items including three reverse-coding items (gil , gi3, giS,
and gi6) were not reliable. Excluding the four items, alpha reliability was .687.

80

It is customary to consider a baseline model that is estimated for each group
separately without equality constraints (single group analyses). The overall ﬁt of the
measurement model explained the data somewhat better for high GI group than for low GI
group (low or, Xz/df= 2.010, CFI=.962; high GI, Xz/df= 1.797, CFI=.966). The
X2(1556) statistic for the model with equality-constraints on all ﬁrst-order factor loadings
is 2928.103 (Model 1). '5 The change in the overall chi-square (X2 m, (34) = 31.367) was
not statistically signiﬁcant. This result implies that the ﬁrst—order factor loadings as a set
did not differ signiﬁcantly across low and high GI groups. In model 2, all second-order
factor loadings were additionally ﬁxed to be invariant across the groups. The overall chi-
square change was not statistically signiﬁcant (X2 din: (37) = 44.664). Finally, all structural
paths were constrained as equal. The chi-square change was not signiﬁcant again (X2 din:
(49) = 60.464). The results suggest that the measurement and the structural paths were

comparable in high and low GI groups (see Figures 4-6).

Table 4-9. Multiple Group Analysis — Low vs. High Group Identiﬁcation (GI)

 

Model jX2 lDf [Xi/dflfm |dfm |CFI

Single group analyses

 

 

 

 

 

 

 

 

 

Low GI I 1529.399 * 761 2.010 .962
High Gl | 1367.328 * 761 1.797 .966
Multiple group analyses 7

Baseline: Unconstrained 2896.736 * 1522 1.903 .964

 

Model 1:All l"-order factor loadings 2928103" 1556 1.882 31.367 34 .964
invariant

 

 

 

Model 2: Model 1 plus 2941.400 * 1559 1.887 44.664 37 .963
all 2"d-order factor loadims invariant
Model 3: Model 2 plus 2957200“ 1571 1.882 60.464 49 .963

 

 

 

 

 

 

 

all structural mths invariant

 

Note. All models compared with Baseline model.

 

'5 Although it is theoretically possible, cross-group equality constraints are usually not imposed on
estimates of variances or covariances. This is because groups may be expected to differ in their
variabilities on either the latent factors or unique factors (MacCallum & Tucker, 1991).

81

Figure 4-6. Multiple Group Analysis by Group Identiﬁcation

 
    
 

-.13* (-.11)
-.09 (-.08)

 
  

   

1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

24* (.20)

   
  
 

-.20“‘ (-.l9)

 

-.36* (-35) . ,

-----—---— --p--——-—-———-—--

 

.11 (.13)
. .16*‘ 1‘ ,r .
‘~ (.20) .I’-00(.00)

---------

Signiﬁcance: p < .001

Fit indices:
X2 (1571) = 2957.200 NFI =.925
X2/df = 1.882 NNFI =.960
CFI = .963 RMSEA =.055

Path coefﬁcients, Low GI (High GI); * p < .05

PA, Perceived Anonymity OPD, Online Public Disclosure
SA, Self Anonymity EC, Evaluation Concern

OA, Other Anonymity NSS, Need for Social Support
DA, Discursive Anonymity PCS, Perceived Community Size

82

Secondary Analyses

Three secondary analyses were performed. The ﬁrst analysis compared perceived
anonymity and technical anonymity in affecting online public disclosure and evaluation
concern. The second analysis examined separate paths from self-anonymity, other-
anonymity, and discursive anonymity to online public disclosure and evaluation concern.
Lastly, reverse paths from online public disclosure to the three dimensions of perceived

anonymity were tested.

Comparison between Perceived Anonymity and Technical Anormnity

Technical anonymity was operationalized as nominal anonymity. '6 Independent
samples t-tests were conducted on perceived anonymity (PA), online public disclosure,
and evaluation concern by technical anonymity (TA) (see Table 4-10). No statistical
difference was found in perceived anonymity between identiﬁed and anonymous groups.
The mean scores of online public disclosure and of evaluation concern were higher when
members did not have to reveal their names in their favorite message board (anonymous
message board) than when they did have to (identiﬁed message board).

It should be noted that nominal anonymity and perceived anonymity affected
online public disclosure and evaluation concern in the opposite direction (i.e. nominal

anonymity increased both). The relationships appear to be spurious. That is, the purpose

 

'6 Subjects were asked whether they have to or do not have to reveal their names in their favorite
message board. Sixty-ﬁve of 301 subjects answered that they have to reveal their names, and the
others that they do not have to. Composite score was used for perceived anonymity, online public
disclosure, and evaluation concern.

83

of the most popular anonymous board in MissyUSA is catharsis. Members reveal their
most intimate feelings and thoughts, often too strongly to be endorsed by even other
MissyUSA members. Therefore, rather than nominal anonymity itself, the purpose of the
anonymous board seemed to affect online public disclosure and evaluation concern.
Members who want to vent potentially self-disparaging emotions use the catharsis board,
and they expect critical Opinions about their venting from other members. The level of
perceived anonymity did not differ between identiﬁed and anonymous boards. This

ﬁnding supports the SIP model’s argument against technological determinism.

Table 4-10. Mean Differences by Technical Anonymity — Perceived Anonymity, Online
Public Disclosure and Evaluation Concern

 

 

 

 

 

 

 

 

 

 

 

 

Technical Anonymity
Identiﬁed Anonymous
(N=65) (N=236)
Perceived Anonymity
Mean 78.000 77.485
SD 21.535 18.867
Observed Mean Difference .515
t-value .188
(if 91.021
Signiﬁcance (two-tailed) .862
Online Public Disclosure
Mean 28.692 32.800
SD 12.911 11.728
Observed Mean Difference -4. 108
t-value -2.3 l 5
df 95.061
Signiﬁcance (two-tailed) .023
Evaluation Concern
Mean 12.195 14.117
SD 5.723 6.032
Observed Mean Difference -1.922
t-value -2.299
df 106.453
Signiﬁcance (two-tailed) .020

 

 

Note. Equal variance not assumed. Composite scores were used.

84

 

Separate PathsL from Self-, Other-, and Discursive Anonymity

to Online Public Disclosure and Evaluation Concern

The results show more detailed relationships among perceived anonymity, online
public disclosure, and evaluation concern. First, other-anonymity (OA) did not affect
either online public disclosure or evaluation concern. It is consistent with the ﬁnding that
0A was the least important among the three sub-dimensions. Second, self-anonymity
decreased evaluation concern, but not online public disclosure, and discursive anonymity

decreased online public disclosure, but not evaluation concern (see Figure 4-7).

Reverse Paths from Online Public Disclosure to Perceived Anonymity

The SIP-based model predicted that perceived anonymity causes reduced online
public disclosure, and was supported. However, it cannot be ruled out that online public
disclosure decreases perceived anonymity. Therefore, reverse paths from online public
disclosure to self-, other-, and discursive anonymity were tested. Online public disclosure

decreased SA and DA, but not DA (see Figure 4-8).

85

Figure 4-7. Separate Paths from Self-Anonymity, Other-Anonymity, and Discursive

Anonymity to Online Public Disclosure and Evaluation Concern

 

Signiﬁcance: p < .001

Fit indices:
X2 (758) = 1686.860

X2/df= 2.225
CFI = .975
* p < .05

PA, Perceived Anonymity
SA, Self-Anonymity

OA, Other-Anonymity
DA, Discursive Anonymity

 

 

NFI =.956
NNFI =.972
RMSEA =.064

OPD, Online Public Disclosure
EC, Evaluation Concern

NSS, Need for Social Support
PCS, Perceived Community Size

86

Figure 4-8. Reverse Paths from Online Public Disclosure to Perceived Anonymity

 

 

 

 

Signiﬁcance: p < .001

Fit indices:

X2 (758) = 1686.856 NFI =.956

X2/df = 2.225 NNFI =.972

CFI = .975 RMSEA =.064
* p < .05
PA, Perceived Anonymity OPD, Online Public Disclosure
SA, Self-Anonymity EC, Evaluation Concern
OA, Other-Anonymity NSS, Need for Social Support
DA, Discursive Anonymity PCS, Perceived Community Size

87

CHAPTER 5

DISCUSSION

Summary of Results

One of the most distinctive features in CMC compared with F tF communication is
anonymity, resulting from restricted bandwidth. In online communities, especially in
online social support communities, anonymity is believed to play a role in promoting
hyperpersonal communication among participants (Walther & Boyd, 2002). Anonymity
in CMC encourages people to reveal negative facts about themselves which they usually
hide from others off-line. Previous studies have underscored this point but one major
limitation was that anonymity had been deﬁned as a dichotomous variable. This study
found that anonymity is a multi-dimensional construct that can be measured on a
continuum and showed how the redeﬁned construct differs in affecting online self-
disclosure. The SIP-based model was found to better explain relationships among
perceived anonymity, online public disclosure and evaluation concern, than the
deindividuation model that previous online self-disclosure studies featured.

First, perceived anonymity consists of three subvdimensions, self-anonymity,
other-anonymity and discursive anonymity. Second, perceived anonymity (PA) decreased
online public disclosure (OPD), supporting the SIP model. Evaluation concern (EC) did
not mediate between perceived anonymity and online public disclosure. Need for social
support (N SS) decreased perceived anonymity and increased online public disclosure.

Third, the measurement and the path coefﬁcients were comparable across low and high

88

identiﬁcation group. Fourth, the effects of technical anonymity (deﬁned as nominal
anonymity) and perceived anonymity on online public disclosure were in opposite
directions. Nominal anonymity increased online public disclosure. Fifth, among the three
sub-dimensions, discursive anonymity was the contributing dimension that decreased
online public disclosure. Only self-anonymity decreased evaluation concern. Finally, the
data also supported reverse causations between online public disclosure and SA and DA,

but not OA.

Contributions of the Study

Theoretical Contributions

The ﬁndings contribute to our theoretical understanding of perceived anonymity in
online social support communities. First, the present study supported the notion that
perceived anonymity is different from technical anonymity (often deﬁned as nominal
anonymity or visual anonymity), as employed by SIDE theory and GDSS research. It
found no support for the deterministic view of CMC, demonstrating that technically
imposed anonymity does not necessarily deﬁne the mental state of communicator (see
Table 4-10).

The two competing models — deindividuation model and SIP-based model -
represent two cases in which perceived anonymity and technical anonymity are differently
related. First, CMC systems induce corresponding levels of perceived anonymity. This is
the case in which SIDE and GDSS experiments and previous studies about online self-

disclosure were conducted. CMC users had no prior history of interaction, less interaction

89

time was allowed, and/or no future interaction was anticipated. They were strangers to
each other. Technically imposed anonymity perceptions thus were left intact. If
participants revealed negative aspects of the self, it was because they felt deindividuated,
lowering public self-awareness, and decreasing accountability. New members in an
online community who pour out unﬁltered emotions or hurt feelings perhaps can be
explained through this deindividuation model. Second, perceived anonymity does not
match technical anonymity when technical conditions no longer bind CMC users’
anonymity perceptions through continuous interactions, as in real online communities.
CMC users overcome physical constraints. Some users perceive identiﬁability in
technically anonymous conditions. Online community members who regularly
participated sensed reduced perceived anonymity, and felt more conﬁdent that they would
receive warm and caring responses from other members even when they revealed negative
personal stories. The monitoring (or “lurking”) period which members usually observe
before posting messages supports the SIP-based model. Community members seem to
wait until they build sufﬁcient rapport and can anticipate how other members respond to
their self-disclosing. The present study corresponded to this second case. These two
relationships can be described graphically as in Figure 5-1. As shown, technical

anonymity is a necessary, but not sufﬁcient, condition of perceived anonymity.

9O

Figure 5-1. Perceived Anonymity and Technical Anonymity
TA

 

TA, Technical Anonymity; PA, Perceived Anonymity
PA corresponding to TA
PA not corresponding to TA

Second, the study ascertained the separate dimension of discursive anonymity and
its importance in CMC. As the SIP model maintained, CMC users develop ways to
verbalize non-verbal cues, and gradually adapt to such verbalizations. Just as speaking
and listening are the basic communication skills in FtF interactions, so are the abilities to
verbalize social information and to detect others’ unique communication styles and
patterns valuable in the textualized world. In this regard, the separation of discursive
anonymity (DA) from self- (SA) and other-anonymity (OA) appears a natural result.
Consistent with the result is another ﬁnding that among the three sub-dimensions of
perceived anonymity, only DA, but not SA or OA, affected online public disclosure.
Communication pattern and style are the unique parts of online identity which send out
subtle cues about the person’s personality

The deﬁnition of discursive anonymity in this study — a perceived lack of
individuating communication pattern and style in a message -- was more pertinent to the

SIP model’s arguments on the verbalization of social information than Anonymous’

91

(1998). Anonymous deﬁned discursive anonymity as the inability to attribute a speciﬁc
message to a message source, compared with visual anonymity. According to this
deﬁnition, all other types of identity information except for audio-visual information (that
is, name, locatability, and social categorization) included in a message may also determine
discursive anonymity. In another study, the same researcher (Scott, 1999) operationalized
discursive anonymity as whether GDSS participants placed their names before each
comment or not — in other words, nominal anonymity.

Third, the theoretical applicability of the SIP model was supported in a large group
communication context. Communication in online communities is a mixture of
interpersonal and (large) group communication. Messages exchanged in an online
community originate from individual members, directed to a speciﬁc other(s), but shared
with the whole community. Studies about relational development in online communities,
however, have focused on interpersonal relationships, which sometimes migrate offline
and go beyond the realm of the community (Parks & Floyd, 1996; Utz, 2000). The other
important aspect, that participants also publicly communicate with the community as a
whole, has been largely neglected. Not only the SIP model but also other CMC theories
such as SIDE theory and hyperpersonal communication were usually tested in small group
conditions. The largest group was Walther’s 54 international interactants using CMC for
a class project (1997). The size of successful online communities such as MissyUSA can
grow limitlessly if technically supported. It appears that only the SIP model is a valid
theoretical framework for a large group communication context of online communities
among the current CMC theories, at least when explaining online public disclosure.

Fourth, the results illuminated a social exchange approach to CMC, previously

asserted but unveriﬁed within SIP. Focusing on relational deveIOpments in CMC, SIP

92

adopted two major theories on how relationships develop through communication in F tF
interactions — Uncertainty Reduction Theory (URT) and Social Penetration Theory (SPT).
Social exchange, a background theory common in the two theories, explained why
perceived anonymity reduces members’ willingness to talk about themselves in depth
publicly in online social support communities. In situations of high perceived anonymity,
people expect more costs than rewards as a result of self-disclosure.

The two theories - and also SIP — emphasize the role of information exchange in
relational development (Littlejohn, 1992). Perceived anonymity in the present study also
reﬂects such idea by deﬁning it as a perceived lack of identity information exchanged
among CMC users. Uncertainty levels increase with perceived anonymity. Previously,
URT has been employed to predict higher levels of self-disclosure in CMC than in FtF
(Tidwell & Walther, 2002), or when perceived anonymity is high (Snyder, 2004). That is,
uncertainty in CMC or in Communication situations where anonymity perceptions are
higher, motivates people toseek more information in order to increase predictability.
Such efforts, however, were not successful. There was no signiﬁcant difference in the
proportion of intimate self-disclosure between CMC and FtF conditions (Tidwell &
Walther, 2002), or the amount of self-disclosure was higher when perceived anonymity
was lower (Snyder, 2004). Such studies failed to predict hypothesized relationships
because they focused only on the third axiom of URT — high levels of uncertainty cause
increases in information seeking behavior. As uncertainty levels decline, information
seeking behavior decreases — which has been criticized for its validity (Kellerrnan &
Reynolds, 1990). More directly, in axiom 4, URT predicts that high levels of uncertainty
in a relationship decrease the intimacy level of communication content and that low levels

of uncertainty produce high levels of intimacy (Berger & Calabrese, 1975). A perceived

93

lack of identity information exchanged among community members increases their
uncertainty levels which, in turn, lowers intimate self-disclosure.

Fifth, the failure of evaluation concern to mediate between perceived anonymity
and online public disclosure suggests that evaluation concern is not so important in the
SIP-based model as in the deindividuation model. The result contradicted the public self-
awareness explanation for increased self-disclosure in CMC. The reasoning of the public
self-awareness explanation is that decreased public self-awareness in CMC reduces
concerns about others’ evaluations, and, freed from evaluation apprehension, CMC users
tend to disclose themselves more (Prentice-Dunn & Rogers, 1982). The present results
showed that perceived anonymity decreased evaluation concern (EC), but the decreased
level of EC did not lead to more online public disclosure. The reason why the
deindividuation model failed appears to be that the population of this study was different
from that of previous online self-disclosure studies. Members in online social support
communities get together based on the same life circumstances they face. They are
willing to side with each other. Social support, rather than judgmental comments, is what
they expect ﬁrst. Expected rewards (i.e. social support) were higher than expected costs
(i.e. negative evaluation). Researchers who explored the deindividuation model had
subjects without any prior contact in their experiments (Joinson, 2001). They did not
share any commonality. In SIDE terms, individual salience was high. They had no
reason to expect positive responses from others ﬁrst. The experimental results were
applied to how web-based surveys increase willingness to answer sensitive questions

(Joinson, 1999; 2005).

94

meticgl Implication

The present study suggests a practical implication that concerns anecdotal
evidence that supports the existence of discursive anonymity as a separate dimension. In
one of MissyUSA ’s anonymous message boards, for example, members actively identiﬁed
those who continuously post messages that dispirited the entire community based on
message tone and writing style, and collectively sanctioned such members by notifying
other members that the messages contained undesired content. In another anonymous
message board, discussion among members often became inﬂammatory because members
mistakenly attributed some messages to each other. Such cases illustrate that discursive
anonymity is difﬁcult to overcome as well as achieve. In the ﬁrst case, those who spread
dispiriting messages may disguise themselves, relying on nominal anonymity. However,
they overlooked that additional caution should be taken in order to achieve discursive
anonymity. The second case exempliﬁed that nominal anonymity does not deﬁne the
subjective anonymity perception. Participants in anonymous boards believed that they
could correctly link message content with message sources without knowing names.
Their subjective perception overcame nominal anonymity. However, such assertion can
be erroneous in many cases. Attribution accuracy is another important issue in managing

online communities especially in text-based bulletin board systems.

95

Limitations and Future Research

Perceived Anonymity Affecting or Affected by Online Public Disclosure ?

The reconceptualization of anonymity as a measured variable introduced the
ambiguity of the direction of causation in this cross-sectional survey study, because the
time order could not be controlled as in experiments. It is entirely plausible that online
public disclosure causes perceived anonymity. When a member self-discloses, he or she
is providing identity information about the self, and becomes a contributing member of
the group. Therefore, anonymity perceptions decrease. This rival prediction was tested
by the third secondary analysis, and the result showed that this possibility cannot be ruled
out (see Figure 4-8). A longitudinal survey design is necessary in order to assess if the

SIP-based causation, its reverse causation, or both, are true.

External Validity and Reliability

The current study endeavored to validate a second-order structure of perceived
anonymity. Such attempts usually begin with existing measurements for the construct of
interest. Existing measurements are examined based on a relevant theory. In this way, the
construct is likelier to cover the breadth of measurement exhaustively. However, the
current study started from a theory without any existing measurement. Therefore, it is
possible that other dimensions of perceived anonymity (PA) exist but were not
conceptualized in the present study. Along with the possible incompleteness, the

conceptualization that perceived anonymity depends on the amount of identity

96

information exchanged among members might overlook the relative importance of
identity information. Certain types of anonymity may be more important than others
according to types of online communities. For example, in an Internet Relay Chat (IRC)
community where the use of an avatar is an integral part of community participation,
visual anonymity would be more important than discursive anonymity (Kang & Yang,
2004). The hierarchical multi-dimensional structure of PA should be tested also with
other types of samples.

Individual items, especially items for biographic information, would change
according to types of online communities. For example, thanks to the homogeneity of
subjects in nationality and gender, these biographic items were not included. However, if
an online dating community were to be surveyed, such information should play an
important role. Perhaps, more items for biographic information may be recommended.

The sample was somewhat homogeneous in terms of culture and gender. Other
demographic characteristics such as age, marital status, income and education displayed
homogeneity (see Table 3-2). Further, the sample was unique in a sense that MissyUSA
members are foreigners who still are under the cultural inﬂuence of their home country
rather than being assimilated into the local culture. These features of the sample might
reduce the external validity of the results. Therefore, the construct should be tested for
external validity with other types of samples.

The representativeness of the sample was also questionable in terms of community
participation level. Nearly 80% of the subjects answered that they visit MissyUSA ﬁve to
seven days per week. Considering that the data was collected for ten days through an
online survey, members who visit the community less than ﬁve days per week might have

been under-sampled. Further, due to the monetary compensation to those who completed

97

the survey, members familiar with such. Internet environments as e-commerce could have
been over-sampled.

The lack of external validity is also expected in terms that a particular type of
CMC technology — electronic bulletin board system — was the focus of the study. Besides
such text-based asynchronous CMC, instant messaging (IM) such as MSN messenger is
popular, especially among teens (Lenhart, Madden, & Hitlin, 2005). IM tools
dramatically have increased Internet use as a social medium, and fostered a sense of
online community among users more than any other CMC application (Alvestrand, 2002).
Synchronicity in IM seems to increase levels of perceived intimacy (Hu, Wood, Smith, &
Westbrook, 2004), which might affect anonymity perceptions. The scale of PA should be
tested across different CMC media.

The grouping variable used in multiple group analysis - low and high
identiﬁcation groups — poses a low reliability problem. The variable was converted from
a continuous variable which was originally measured with three items on a Likert-type
scale. The original continuous scores were categorized into low and high identiﬁcation
groups using median split. The ﬁrst source of the problem came from the low alpha
reliability of the original continuous measure. That is, the alpha reliability for three group
identiﬁcation items was merely .687, lower than the usual cutoff of .80, but high enough
for exploratory work (Nunnally, 1994). The second source lies in the fact that the
converted variable is not so appropriate to multiple group analysis as such originally
discrete variables as gender. That is, subjects whose continuous group identiﬁcation
scores ranged around the median might have been differently categorized if more reliable

measures were employed.

98

Different Functions of Self-Disclosure

Self-disclosure serves various functions such as catharsis, social validation,
relationship development, and response solicitation (Derlega & Grzelak, 1979). Different
functions of self-disclosure should be noted in relation to perceived anonymity. For
example, among several anonymous boards in MissyUSA, two are the most popular -
“private talk lounge” and “motherhood.” In “motherhood,” which can be characterized by
solid group identity, members self-disclose in order to solicit or express emotional support.
Increased levels of online public disclosure attract more social interactions, which
reinforces relational closeness among members. On the contrary, in “private talk lounge”
where members share all kinds of distressing experiences such as marital disputes,
conﬂicts with in-laws/neighborhood, problems at work/school, and difﬁculties living in
the US. as a foreigner, emotional catharsis is the primary function of self-disclosure.
Because of low group identity compared to “motherhood,” self-disclosing messages
seemed to spawn negatively toned responses rather than supportive ones (announcements
are regularly made that the messages are being monitored, and removed if containing
potentially harmful content). Future studies may improve understanding of perceived
anonymity affecting online public disclosure by examining different purposes of

anonymous boards.

Explaining the Unexplained

The large proportion of variance remained unexplained in each dependent variable.

Need for social support and perceived community size explained 2 to 7 percent of the

99

variance in self-anonymity, other-other anonymity and discursive anonymity (excluding
the portion of the variances explained by perceived anonymity). For online public
disclosure, 1 1 percent of the variances were explained. These results suggest that there
would be other variables that can better explain the unexplained variance. The present
study proposes four possibly missing variables in the nomological network model.

First, Internet self-efﬁcacy (LaRose, Eastin,& Gregg, 2001), Internet social support
efﬁcacy (Eastin & LaRose, 2004), or familiarity with or experiences in online
communities (Jaffe, Lee, Huang, & Oshagan, 1999) may reduce anonymity perceptions
by increasing members’ ability to predict others’ responses.

Second, trust can explain more variance in online public disclosure. Trust has
been studied as an important predictor for Internet-related behaviors such as information
sharing (Ridings, Gefen, & Arinze, 2002), online purchase (McKnight, Choudhury, &
Kacmar, 2002), and social support exchange (Blanchard & Markus, 2004). Members’
trust in their online community or the owner of the community in terms of privacy
protection will positively affect online public disclosure (Joinson & Paine, 2005;
Henderson & Gilding, 2004; Ryan, 2003).

Third, the uniqueness of the population — Korean married women living in the
US. — suggests that loneliness or depression, a psychological variable that has been
extensively studied in relation to the amount of Internet use (for discussion, LaRose,
Eastin, & Gregg, 2001), plays a more important role in explaining online public disclosure.
That is, Korean wives in the US. feel lonely or depressed in a foreign country generally,
so they are more willing to talk about themselves with others in the same situation whom

they can meet online.

100

Finally, the predictive validity of perceived anonymity should be tested also with
another anonymity-related behavior, ﬂaming (Taylor & MacDonald, 2002). The current
study focused on online public disclosure, a pro-social communication outcome in a
speciﬁc type of online community - online social support community. The explanatory
power might be higher for such anti-social communication outcomes than for pro-social

ones.
Conclusion

The goals of this research were, ﬁrst, to create and validate the scale for perceived
anonymity, and second, to test its nomological network validity in an online social support
community. The results demonstrated that the perceived anonymity construct has a three-
dimensional hierarchical structure. In addition, perceived anonymity was not bound by
nominal anonymity which is technically deﬁned. Perceived anonymity affected online
public disclosure negatively, supporting the SIP model.

Not all CMC develops to the hyperpersonal level. Then, how can we explain
anecdotal or empirical evidence supporting CMC as a hyperpersonal medium, repeatedly
reported from the practice as well as from academia? Perhaps, anecdotal evidence
represents extreme cases, which were skimmed from the whole spectrum of CMC from
impersonal to hyperpersonal. Empirical evidences might result from weaknesses of
experimental research that employed technical, dichotomous deﬁnitions of anonymity
using subjects with zero history.

Looking inside the CMC phenomenon more closely, the SIP model found that

CMC is not inherently different from FtF interaction. When CMC is compared to FtF

101

interactions, it is often said that CMC has the potential for hyperpersonal communication.
CMC researchers proposed higher levels of self-disclosure in CMC as an avenue for
hyperpersonal communication, and the role of anonymity was emphasized. Such cross-
media comparisons lead researchers to overlook whether the same prediction holds within
CMC. The within-media investigation of this study revealed that relational development
in CMC was analogous to that in F tF communication. This research unveiled a theoretical
loophole in cross-media comparison studies by creating a valid scale for perceived

anonymity and testing it with a sample from a real online social support community.

102

APPENDIX A-l

Pretest Protocol and Sorting Instruction -- Korean

103

71

1 r1
rIu:
21111
_>_4.

)1:
21:

0116172191 717.1711 71117110078154: 71.1.1 75.1 a 144-701;; 74—41:) 311371
719—1711777117 (7177113171 711: 77717.1 18- 7.7.1:- 871)4;1 771(1):.) 7111,1617 ., 91%
«113171er 7117171177: 71-1 3171715717171 71171 4713167471. 171417150111;
71"” 7151;" 3177117172 017.11:- 771° :77 7114.9. 77014—1—4171747121-3—471. 74
51—31—3437,: tux/s}. E—stLOixpggal @1240) L45],

31317147].qu1- "IEIE °I 23’5“"?! 7I‘CEII 7(I7I XIMOII EIIZSII L —”I7E “Iii
MissyUSA §I‘%_'§ °I MissyUSA 01] 21011710114 fiI—‘a 1° __§*‘I XI-7I 7~Irﬂ°ll7il EH31) WI
67IXI °6I°3 o; 7‘II"IEI1C’I° .1—. 51‘3“" 7IE°II7C 167I’2Io 0.51,] :rL‘IIz—IQI O“73: Hoff-E
100 711 7’ 7771215701 471° $1451. 71714 71.1217; 71-37—51; 11.7.1711 711-1—53 5.77.317)
51111414. 7I'E— Err/‘I ZIPS—3' °I7§73IAIE 3101-1137718 339111414.

U
I.”

1. 717167814 717174 712—3- 7101217118 171E717) 717111471. 7171171917.. 7- 7 171711 711711
7177171711 71—5:— 71147—11719. 71714 71.1.2711 471 7111.171 :1 7.171717771017171 7'7.
7141— 717417 1471714771712

3. 1007891 117.3341 715% 51.141. 731701 it‘ll 9151741361737 93173111011711? ‘" 7.7311 717411613.

4 1:44 715% 717.174 5171711111. 71717181014144. 7.7174 71.1101 747.1 E87401
0111.1 7174(7171714 71E)91 11711.1. 77 11°71 47—371 787471711371 71171715- 74174 711.7; 017101
811717193711171 1:17.174 71.13 73%— E—‘i— {1% 781111.93. E77711 21711713. 77 71711714011
71117171: 1:474 71.12.91 7:71: 917871 0.1 91710115 17.17:— T94 177.1711 715— 7’7 717.174
7127115717371 71151: 813L151)

5. 47147177171 717147717151 .7147 4117717171717. 74712017117: (317471
91—17 171. 171711 71.1.7711 711-.- 8—7814171— 711181 711.1 01711 71717117401 4171 717111. 7.11171.

6. 71—8-0 77.1 747-7.- 7187171 717421 447-71 17171—71 7117.1 71111—8- 71711171
774-717-7371 :1 71471714701 71717114771712.

1: 7114 117147121 1011-7171:} 2: 4717171711 7.47.771
3: 77. 77147-7) 9411351 4: 1177171

5: 714. 7.717.771. 6: 1177171 1171-31-71
7

: 714 117871711 9.71771
11191 74 77.1 211711 71171711. as 74714—41...

7_g3§ﬁﬂ44ﬂ§ﬁ§§gwmeE@a%ﬂQE%aﬁ%4&

104

Other- ' Group
Anonymity ' Identiﬁcation

32! ‘53}
§

 

105

7H €38 91437-51) E12}

Self-Anonymity

XV] 91‘5”?! 789—]: 1V] ZP‘JOI the: 49:01] 71] 22‘341121 ﬁbikﬂur °}-‘?-E Z}él% ‘20}EXI
3% QHEH‘Z- 21/54. GTE-41%] 7V131 9.1—1:- #7174410“ EM} 751E7P" Tiﬁﬁ}: °J’-‘J. $1
4%}94 4158153 2.194%}:7311— EMUW ER] 2.1 Urdu}. ’3‘81-1-3-7F8 (é‘ﬁ,
*}%2}°}OIQ,. ); (URN/‘3 (TE-Ti, °IQT¢JIPTQ 13.3361"); 143131-5-
(EJ-‘u‘ggl‘HTI—T T’é,éi‘;€ 91113401,...3 91A}_T_% “116%“ ﬁnk/1‘4; ° Tr’ﬁx ‘1‘ All} 781W?“ J,
8 73°

wo1%%%®%dQﬂ-.
}%2}°}°lcl—§- a-Mg ad ”or.
olﬁTiQ§WﬂﬂT¢§§ﬂ€kTﬂ§ﬁéﬂk
JT iTﬁﬂwié a0
°}N3Ti§°J°}% a
JMH£T§T°éa
#91 ET *Jai—S: °Jo+f£ T° a .

7} 2”“ tM—czli _. 8. 840,130] T9112] ‘34—:- ai Qt}.

La
La mesE-s— 034% 31 {$5}
LH

19':
(43
win win

51%.

~

ﬂﬂ
mw
I11 WI”

32.
C:

331.
(+0
N!

B.
‘40
N
10 rlo F10 F10 F10 rlo F10 rlor rlo F10 F10

1 Lu
J— Lu
-1 I“
J Lu
Pun

.__._
—
——

WNQVPP‘NT‘
191.
[11

51-3-3
311%..3-
51531-3— 330
51%.aL. 4PM TL}°}°I%91 1401-; °
51%{32517 LH T40143J§*}§ 9.34%? £5;
ERR—E- ° LH7} T?°JZI 14°19. 3 é’it 123%] ‘
31%. -‘:é:°LH7}T?°J11H17}2}-’F3ﬁt§£dolb}% 51%}
ﬁl-‘EJ. 417} -“r-TWJZI 137} 3BR} $111011 EH51} 7833}?

_\o

—?— 7}ng T°‘7§E% Morg-

I—I‘F—‘H
E92“?
0

—‘
DJ

.3
222322222232223

r5. r5. r5. r5. r5. r13. r35. r33. r5. r5. r5. r5. r35. r5. r3.

N'lﬁm
rlo 71”

.—a
LII

7151-5}.

5“

2.

FE.
L :1:
42
5:
:1.
MO
:21
r51
35‘“
mln
MO .1
r8. ”

13

.9.

E.
E 19*:
dB r
min min

106

Other-Anonymig

E191 91°32 2£191: 51% A112101 -‘1’--'1‘°J11 ”01% -1— $31111— °JTJ E1°J°11t11§iMJ¥117§i71%
21221212 2121. 2 21221 211212222122 21227121 2121 ° 2 21221. 2.12—2-
71‘8 (2‘5.*1°8-21°1°151,...); £11131°J*a1(T%1.-’1‘-.T. 01221.42,” Ti, 75131213,...) )1
42131-3:- %(°J-?%211t131” §)*3, {1:222 E1211,T1U1,...); 914.45% 1111121 5'11 °c1’-‘1; 131-"£1 ‘21
A121 ‘22 (111751, 311731: ° —-°—1,

1242222222222222T 2.

I9.

19. 212 212 2121—221 11211010122. T22 T 2121

20. 212 2121 212.221 21222212210121 2.21.2 2212 T 212— 21 2121

21. 21221211 2121—22 2T2 2.121222 2212 T 112— 21 221

22. .12. 2121212222 212 212.22 222.121 2121- :1.

23.112 21212 1:12 212. 221 21212 222T 2121.

24.222121222212222 a 222112222T1121

25. 212 2112112 212 212122121 21—2— 21112 122 212 -1212 21211-1. 11221 T221

T2121

26.2122112112212- 21222 21.2121 22-221mm.
27.12131: 11111111in §1ﬁ§€5€171711$mzo$§$31$ﬂﬂk
28.212121212212212 122121L1T21o1o122121212T—22T2121
29. 22121221221° 22121212122212 TTT2T1121.

30.1212“: 21251:- *51943013 311-?!23-01-1-511317‘1 011

r 111°
11
11
2°.
.11

31. 2.21T-222o1212212c1221° 221TT21212212T2121.
32.112 21121 T211211u11211212212 2121 ° .2 212 2121221TT21212212 T 2121
332122122122 o1o122121112212122T21-2—21221.

34. 21221212212221:21222122222121 2.1212T11—221221

107

Online Public Disclosure
29:29:91 11151? {31711431 72191: {-2121 71111216101121 111% 312-13231 21211191 111911731 11°12:— 71PM
{3111312.— 721. 311221.322:— 01211‘21 561711-3— %-211 l{17112—1 1129-2- 22 31.9.3 71t11§131 l113/1121]
11171-217112- %-111 911112141111211/‘1Ea’1 2131 31131." =7, X171 55171171512 312-3417111112 ~1-
31—527-‘1 0113121011 1:111:11 5121112114. %21°J 71111215101121 °1”J°1L1:XI71%‘-% $111331
711°J—2’z1t113212171 €171171°112121,€-21°J 71111115191 E— 51211221114121-5‘2‘5‘
711411812§11 71171 €171. 1171-2- {517112121- A112$ 71141-91 52-01 23217; 71121111011 EE—
31%1‘201’SE- 7115-3151— 7‘12 ° 5'. 31% 1111.

35. 7177101111 21,301] 2112.1 111711
36. 71711011194 712121—11 J7 71
37. 7177101 L117121E 211111. ‘21
38. 7171101 111 71&,2.1?22
39. 71771012191 721111111131.
40. 71721011191 ”2112-8-
41. 7111011191 212121211
42. 7172101 111 7151111 111

31:01:},-

Hﬁ 1113—12114“ °1—°a- 71%EO1O171‘213F9JE1.
{13% 410:11:11“ 010171213?- 0°1C1.

1711011 E1121111 010171 {2114511151.

1010171‘21’“°1151

3%11“—?°1C1.

:45}.
11131—2.
:12.
O
'11"
~11);
1111?!

Group Identiﬁcation
’11?! $2121 7:191: 71171 2111.12- ° Z1*"J °1 27-11} 7111:2131 $32111 3115- 721

43211710122191 2111212121 7101121 11171122. ‘31—2 E— 2 21.

44.112 21 221°J 7111112121 211217111 7121571 121—8.— 221 ~2- 21219217121.
45.212 21 2219.1 91111121211 71212.1 312— 2212.121.

46.21%:— 01 22112.1 21211121 2121212} 212° 219—21215: 2.12121 922

47.01 22121 71131-11212:— °1T 11912.1 1212110121.

48.21 2219.1 21131—2121 2121.012} 212.101 2112121.

49.21171 01 22191 711111421 212% 3121111171111 2121-9..- 0112711 22711:: ‘2’
112.101 3121.

50.21 22121 211—13114212— 11171 27121712- 20112321.

51.11—2— °1 9—21°J 21T1-1T12191 213 21212—21 21221,

52.212 212 2121-2 21 21017121 1111 2% 2122M
53.21%:— 01 221°J 212112191 219112— 21 3112212

33

108

Evaluation Concern
15713-311121 72191: 171%— 711‘111—2-01 7171194 $1111 33%-% 0171171937121 721°l71°11 E1131
21 7231315 7211

54. C1E §1°J E01 L1011711 %916171 911% 7101111.

55.1:1-E— 33196 1%01 L1171 €- LHE—E 1117121 72101111.

56. C1-E— §1£E°1L1E 2611821 7,10151.

57. E1E §1° E01 L1171ﬁ—C L11%—% 41.101731 7)101111.
58.1313 §1°1a-°1 I1E- 111° ° 7101121.

59. qéigq 3319.1”01 1.117} 94;} 3011,} A1210 311-113 7101:}
60. D1E ﬁEE—E L1171-53-E EO1L17117<J°11121.1t1111231 7101141.
61. L1—‘= L1171E%1 EON *17JE: 7 oJ611325115Fl 7101111.
62. L1: L1171.°j—°J 30147171; 0.13112: 761%” 71.01121

Needs for Social Suggort
51-9111E-‘Ei-7—E191 71111011 E11111} €31 7219—1: 111% A1%‘E—E—‘7-E191 KNEE 93$ {1 1111
211-9101 E—S—% E 2.11:} 7on 3114+:- 233.7,; 214.

63. L117} 11151721221 01:21a- 0;. 6': 01 2.: 1:1, L1 0171—:- 23:01; 12.2: 712.101 21-9401
ﬂﬂgEEQ.

64. L1171 @31353011 E71171131'1-E 1111, L11 01171-2} 1501? JJ 1113101 31591011 312319.151.
$1111

65. AC1‘1 1%? °11 1111§_11~115§% 5301-3 13.15517119301‘11011 “9.12131: $1131.

66. 0111. °a101a11 % 519?}??? “1,3531% 515:1 3F— 2113 71136101 T971011°A121215 127.1131

67. L11 §71%% "01912.1 *1%11-§“$1°1121219.15 2’SJC1.

68.01J?J°11t11611€7‘c1—°- 11181011211111,E%%%‘¥_1§ 1*1%1°12$1°11° t11.11.31.151.

69. 78239.3 J§1§T " ‘21—‘5— Eﬁ‘d—S- 311% 711230 21591011 9.1?13111 {1111.Sd

70. L11 3771155-3- 231175751“: 116151011 t115511 311117511 1115—1111-3- %*1%1°1¥-$1°1121?19.‘5 {£51.
71. A136115-01 11°11 E117511‘1’101 783191312331; @111.

72. 2151551 71121 41°J 74123011151:- 81-81% 1713* E101 3159141 9,1912%! 125.1131.

73. 111%711361—37‘33 73117—331 71121321171113- 1111§EEEE ZE1131

109

APPENDIX A-2

Pretest Protocol and Sorting Instruction — English

110

Sorting Instruction

You are given six blue cards and 73 yellow cards. Each blue card has a concept

(a term that represents a particular phenomenon of interest under study) and its deﬁnition
(statement/s that explain/s what the term means). Each yellow card has a statement that is
an example of one of the phenomena represented by concepts on blue cards. The
statements are to be used as questionnaire items in a survey.

The six blue cards list six ways in which people feel about themselves in their

daily lives and how MissyUSA members feel about themselves when they participate in
MissyUSA. The 73 yellow cards contain statements that exemplify the seven ways. Now,
let us start to sort yellow cards according to blue cards. Sitting at a desk may be more
convenient to you.

1.

DJ

Carefully read the six blue cards. Ask the researcher about what you do not
understand. Only after you completely grasp the idea of each concept and accept
its deﬁnition on each blue card, proceed to the next step.

Spread the blue cards in a row on a desk.

Read through the yellow cards and ask the researcher if you have any question.
Read each of the yellow cards again, carefully this time. Place each yellow card
under a blue card on which you think the yellow card statement best exempliﬁes
the term. Sort all yellow cards in the same way. The number of yellow cards is
not the same for each blue card (you do not have to place an equal number of
yellow cards under each blue card).

Write down your sorting result in Table l. The ﬁrst row of the table contains
concepts on blue cards. Enter yellow card statement numbers in blanks under
each concept in the table.

Evaluate each item according to how well it reflects the concept based on the
given deﬁnition using the following seven-point scale.

1: Very clearly unrepresentative 2: Clearly unrepresentative
3: Unrepresentative to some extent 4: Neutral
5: Representative to some extent 6: Clearly representative

7: Very clearly representative

Please write the number next to each item in Table 1, using the response choices
given above.

7.

Provide comments on unclear and inappropriate items.

111

Table l. Sortin Result
Self- Other-
Anonymity

Anonymity

Stmnt Rating Stmnt Rating
No. No.

Online Public
Disclosure

Stmnt Rating
No.

Group

Identiﬁcation

Stmnt
No.

112

Raung

Evaluation

Concern

Stmnt
No.

Rating

Need for
Social

Stmnt
No.

Ratmg

 

Concepts, Deﬁnitions, and (Candidate) Items

Self-Anonymity
Self-anonymity is deﬁned as the degree of feeling that the self is unknown or not identiﬁable

to others. It is a perceived lack of identity information about the self given to others. There
are ﬁve types of identity information signifying one’s identity: Name or pseudonym
(nickname, username, ...); Locatability (mail address, email address, IP address, phone
number, ...); Social categorization (demographics, organizational memberships, hobby, ...);
Communication pattern and style; Audio-visual information (pictures, background music,

...)

PWNQP‘PP’NT‘

t—ou—ot—tt—nu—Iu—tu—nu—A
NOMAWN—‘o

Some members can recognize my name.

Some members can recognize my usemame.

Some members may ﬁnd out my email address or homepage address.
Some members may ﬁnd out my mail address or telephone number.
Some members can recognize my IP address.

Some members can guess how old I am.

Some members can tell my marital status.

Some members can tell my profession.

Some members can tell how much education I have had.

. Some members can tell our household income level.

. Some members can tell how many children I have and their age.

. Some members can tell my hobbies or interests.

. Some members can recognize me from my writing style.

. Some members can recognize me from expressions or words I use frequently.
. Some members can recognize me from the way I approach the topic covered.
. Some members may imagine how I look (my appearance).

. Some members may match me with pictures I posted.

113

Other-Anonymity

Other-anonymity is deﬁned as the degree of feelings that individuals do not know or cannot
identify others. That the self lacks identity information about others, is a subjective
perception. There are ﬁve types of identity information signifying one’s identity: Name or
pseudonym (nickname, username, ...); Locatability (mail address, email address, IP address,
phone number, ...); Social categorization (demographics, organizational memberships,
hobby, ...); Communication pattern and style; Audio-visual information (pictures,
background music, ...)

18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.

I can recognize the names of some members.

I can recognize usemames of some members.

I may ﬁnd out email addresses or homepage addresses of some members.

I may ﬁnd out mail addresses or telephone numbers of some members.

I can recognize some members via their IP addresses.

Sometimes, I can guess how old other members are.

Sometimes, I can tell the marital status of other members.

Sometimes, I can tell the profession of other members.

Sometimes, I can tell how much education other members have had.
Sometimes, I can tell the household income level of other members.
Sometimes, I can tell how many children other members have and how old they are.
Sometimes, I can tell hobbies or interest of other members.

I can recognize some members from their writing styles.

I can recognize some members from expressions or words they use frequently.
1 can recognize some members from the way they approach the topic covered.
Sometimes, I can imagine how other members look (their appearance).
Sometimes, I can match other members with pictures they posted.

114

Online Public Disclosure

Online public disclosure is deﬁned as the willingness to share the core self with other
members. Although the disclosing member seeks social support as a reward, the member is
also aware of risks associated with the disclosure. They are concerned about negative
impressions that the disclosure would make on other members. This refers not to a
willingness toward speciﬁc members through private interaction like personal emails or
built-in memo functions, but to a willingness to reveal such content to the whole community
as an audience by publicly posting messages. The discloser knows that his or her message is
to be posted on the community’s bulletin board where all members have access.

35.
36.
37.
38.

39.
40.
41.
42.

I am willing to reveal negative things about myself.

I am willing to express my most intimate feelings.

I am willing to share what I did wrong.

I am willing to share what I would not do with my family, my off-line friends and
colleagues at work.

I am willing to talk about my shameful experiences.

I am willing to talk about my hurt feelings.

I am willing to talk about my failures.

I am willing to share my family history or secrets.

Group Identiﬁcation
Group identiﬁcation is deﬁned as the degree to which one identiﬁes oneself to one’s social

group.

43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.

I feel I do not have much to offer this online community.

I feel I am one of the least contributing members in this online community.
I regret I joined this online community.

I do not tell anyone that I am a member of this online community.

I feel that this online community is worthwhile.

I am ashamed to be a member of this online community.

My membership in this online community has little to do with how I feel about myself.
I think of this online community as part of who I am.

I see myself as different from other members of this online community.

I often cite this online community when I talk to others off-line.

I enjoy interacting with the members of this online community.

115

Evaluation Concern
Evaluation Concern is deﬁned as the tendency of people to be concerned about how others
evaluate what they say or do.

54.
55.
56.
57.
58.
59.
60.
61.
62.

Other members will disagree with me.

Other members will criticize what I posted.
Other members will misunderstand me.

Other members will dislike what I posted.

Other members will laugh at me.

Other members will disapprove of what I posted.
Other members will oppose what I posted.

I will be rejected for what I posted.

I will be ridiculed for what I posted.

Need for Social Support
Needs for social support refers to the subjective perceptions of the extent to which there are

few people in one’s social circle who are available when one is in need of social support.

63.
64.
65.
66.
67.
68.
69.
70.

71.
72.
73.

I wish I had someone who listens to me when I struggle with my life in the US.

I wish I had someone who listens to me when I have a marital problem.

I wish I had someone who listens to my complaints about in-laws.

I wish I had someone whom I can ask for advice when things go wrong.

I wish I had someone with whom I can talk about my problems.

I wish I had I someone who helps me decide things.

I wish I had someone whose advice I really trust.

I wish I had someone who can provide objective feedback about how I am handling my
problems.

I wish people had conﬁdence in me.

I wish I had someone with whom I can share my most private worries and fears.

I wish I were as close to my friends as are many other people.

116

APPENDIX B-l

Pretest Consent Form — Korean

117

2.2 7.1712271
227112. 2210.1 7121127212 217171 2127.1 2:712 7112101122 22

°J¥X1z Robert LaRose 311"}: Email. 1211‘11seﬁifn'1s11£1111 Tel.353-6336
2711 71.1, 217121173, Email. wage-431311211 Tel. 485-3852
Department of Telecommunication, Information Studies, and Media
Michigan State University

2212227122712222227322 2.27131. 7.17.17127122 1222 721 22
27171271201712.2222. 22101712212 720010112171.1 01711-0121 2022
22.22 2.1712 227.1011 2722 2212121121 2 0.17171 011717101 71212-01 20.121

7.1217171717171212711010171 71:71201101-{2‘217:1~7101 220012211121

19:31°J 712112121 (".1 On1121 1.21 '1‘ 9.11“ MissyUSA91 3"11"11—‘a301MissyUSAg 0178-31131“

X1"J‘:°11 E113110111517113277115713 3.37112115781—341311 :15'5‘23— 1’1’T0‘1 5% 718317151"?é
MissyUSA91 31° 19171171011 1 (142101191 712104 ~3- l11132113111711 519.31%.1’1131 1217711171ch—
”13.13/11 011 111111 715%% "CW3“ T711133. 51-1— ‘1 ,0‘113111: X11"J«1 MissyUSA 3J6 751%011
511311’11“ 04731123 °1° 718317151

71222011222:01172717226171.0201221. 222220121201 271.120.122.271
21011:012_ 22201222712222 222212.12 227112.201, 212122
2210101712771 2101212712 21201171 210117.122. 222 7121121. 012122 712.01
227171 2122 2222.:- 22 212 71. 211121. 727122 271.22 712. 22 7271212
22222222.

0222010122:- 012122 7122 2710122 71.22., 7.1 7.2.7101 71712 2271712 2.7121
271271011 212 72712.12 7127171212 22201 72712 22717120122 27.1721
22 2011712 2171171 71717171 22221.2. 12.1012. 21212 7171222 7.12712 7
2222.

0120112 201712212711 717.1221 01227171212 2171:: 27.117 2.1.221 7112 201
271271220171: 01211101 2120171122 7171712:— 2122201. 21 7121221.
2271012 21711 222.. 222.212 7.1.2 212—221.

2 05 7011 E11 2.1 2.12121 7.1. 2 Robert LaRose 111.2011 711 2'1: 211112111 1:1 (address: CAS 413, Dept.
of Telecommunication, Information Studies and Media, Michigan State University, East Lansing,
Michigan, 48824; email. 1.11m; d‘mstl edu, tel. 353-6336). E- 911?” {1338 C111 121:11.191

4131 41151011 1111 {191-31 171 {51315.1 7E , 3:- E’. E11410 751- ° University Committee on Research
Involving Human Subjects(UCRIHS) $1 1?} "J'°J Peter Vasilenko111'7'11011711 $91 1112311.] 31
(address: 202 Olds Hall, East Lansing, MI 48824; email. 131111 11..r_11s_11_.tg111; tel. 432-4503). 91 ”($1 2%
$1312 0 51-, 91° “1 SEE 71%‘1‘114 131.

2 2120/1 217/2 2221172.

 

 

 

118

APPENDIX B-2

Pretest Consent Form — English

119

Consent Form

Title:
The Creation and Validation of a Perceived Anonymity Scale
Based on the Social Information Processing Model and
its N omological Network Test in an Online Social Support Community

Researchers: Prof. Robert LaRose iari'gjgsgggﬂi1139,ch USA 517 353-6336
Haejin Yun, Doctoral Candidate gunmen:msuuit; USA 517 485-3852
Department of Telecommunication, Information Studies, and Media -
Michigan State University

More and more people enter the Internet to meet people willing to provide information and
emotional support. Among many factors related to participation in online social support
communities, this study focuses on people’s perceived anonymity in terms how this factor affects
the way people talk about themselves online. More speciﬁcally, as a preliminary step, the purpose
of the present study is to generate questionnaire items to be used in a survey.

You will be given index cards of statements describing how people might feel about themselves
while participating in MissyUSA, an online social support community. You are asked to
participate in this study because you are a member of MissyUSA. During the study, you will not
be asked about yourself, but only will be asked to sort the statement cards according to the
instruction.

The sorting should take about 60 minutes to complete. Your sorting result and comments will be
kept confidential, and your privacy will be protected to the maximum extent allowable by law.
Your identity will not be associated with answers you provide. This consent form will be
collected separately from your sorting.

Your participation in this study is completely voluntary, and you may choose not to participate at
all, you may refuse to participate in certain procedure, or you may discontinue your participation
at any time without penalty. You also may withdraw your consent to participate at any time,
without penalty.

While this study is not expected to yield any immediate beneﬁt to individual participants, it will
add to the general store of knowledge about people and the Internet. There are no anticipated risks
associated with participation.

If you have any question on this study, please contact the investigator (Professor Robert LaRose),
at CAS 413, Department of Telecommunication, Information Studies and Media, Michigan State
University, East Lansing, Michigan, 48824; larosegggirisked;1 and/or phone USA 517 353-6336. If
you have any question or concerns regarding your rights as a study participant or at any time are
dissatisfied with any aspect of this study, you may contact — anonymously if you wish - Peter
Vasilenko, Ph.D., Chair of the University Committee on Research Involving Human Subjects
(UCRIHS); ucrihsm.msucdu and/or phone USA 517 432-4503 and/0r regular mail at 202 Olds
Hall, East Lansing, MI, 48824 USA.

Your signature below indicates your voluntary agreement to participate in this study.

 

 

 

Name Signature Date

120

APPENDIX C-l

Online Survey Consent Form - Korean

121

a? 5&7} seem
nae. eat?! arrqaonaet 91‘8*33+ :47] 5371M?

 

C{if—r93}: Robert LaRose 511-”?- Email. ﬁrgsiggggggugig Tel. 517-353-6336
ﬁ-ﬁﬂzﬂ, Ell/43423 Email. gunhucz’rt‘msumlu Tel. 517—485-
3852

Department of Telecommunication, Information Studies, and Media

sate; arm ase- esn are Mae-2+ am} 2351—:— maan 2W4 mere- oer-a
an]: {MEX} %fz} Ate-act me an we: sas- 58M?— an} $813111]?
Ate—2:24 63:47} 2.2%} e017}: glean}. saw awe] 5304011°33J~% n1 211—:— «21:21
391—:— e—a— 03%.:— 9.111751 ween ass—s— ate—m. alarm. ‘35, 212131 avenge]
4337130] am; glen/x1 xm won Luau °l°t7lﬁlr‘:— How on 0129 oats-.2.— v1 21:57th
ea ‘33.??4‘4‘4. e— a—aaaonae— as as} 7}—.%ti1*1,lalil 0431-301 was:
autism 9.1%:- %al?l ate-146mm, octet-e- Mon man cream/~11: 7t°11ttléﬂ°4§
aaqq.

"-33.4011 $535+:— oM} Ame— 30 'E‘ZSE (3141:}. alt-35A} 53°41:— a‘iZ-igi eta—5.2]
Natl—:1 94AM] €21 31.9.01, F304;..— azgaaqa 761/82$ WILE‘EHCL mas-H 31w
LH%°H BMW-E 3134—27 FMS—a atom, $301 $2615— @ELHOM 24:14:32.: 043151—91
EatolBM-E— Heist Satan}. sate! Halo] 14015:”.— ”9%8- 9181 L11 40115.7} x}??—
°l%6}~.—‘:— EH] (cookie)9i ee @aieel ciao] 31%;! $347} 9.1%: 7Iae— $4.61 Arse-5m
‘Bé—EH 1:}. Octal—3:94 earns—e- tt-E— %L7}2}%-°4 ream an" %741i1v1151°l 131%
aqua/371 EMEJ. 99.?3-‘5-9. *é—E—ZEA} til 0] 151% ﬁrst? —T— ale—Ht}. 1323 2.3—5301]
tHszH eta/85% smell $9.49. 5104, ﬁat—EX] gel-32A} 730E13— east/2 3?
sale—Ht}.

”SE-33"} 131-219101] ”+844 °i°iti9i 01032155173: 7l‘135l‘t‘: 33 Ext—7?} $1312: 0431—51-94
Xi‘é‘z—‘l 944°" “lg-LIB}. DH}, 4‘84} 4°qu 0]“351551-‘5 Q-E-Zﬁl‘lw ‘3‘?“ zit—4‘1 9341i
é-El‘ﬂ ”El-gag: EL“ 531531 {QB-23"] ‘3}. §%§*l% 3217411 ”WV/it‘ll Eiﬁ’l 3?— °l§
Liii ”iii 5 €— tl‘éigi Amazon.com "o‘%‘?l% °i "QT-ii ELLHEFQ 01123 ‘il‘riul. Iii-8%}
°i°lti9i °l".—1l—’F‘i-‘E 61%?!23: ELIE/t] f7- O‘iEi-E‘Sl ﬂﬁixJ—é‘ 945% #1113 Zi‘izileiti.

e 993101121 maﬁa-ea] ataxia 019.1% mm sa12]:- aaaaaa ease“
aaaeaaaneaaaaaewaeeaaaeaamﬂaaaaqq
‘23-?étali «2.1611 eta—Era] east—e stems ea seat}.

a- 990“ EH at area 1 3- Robert LaRose fit-011 7%] aavieq Ct (Address: CAS 413, Dept.
of Telecommunication, Information Studies and Media, Michigan State University, East Lansing,
Michigan, 48824; Email. lamsez’ri;msuedu, Tel. 517-353-6336). 3- ‘E ‘7" 9i 33%;?! “i 3151—9]

'21. 3i *l' 3c} °li CH 7?} 9'34 ‘4’ '33“? X3 , 37% 543*} iii-9: University Committee on Research
Involving Human Subjects(UCRIHS) 94 ii ’99.] Peter Vasilenko EH} 01] 7" E—Ql ‘5}?an El
(Address: 202 Olds Hall, East Lansing, MI 48824; Email. gigrihsjg‘fmschg; Tel. 517-432-4503).
9.} egg- saevai 733-, 94 @235 ﬁrs-EM 12}.

a salon act are sane, owns] tea-a- east #2471 area at.

. 122

APPENDIX C-Z

Online Survey Consent Form — English

123

Consent Form

Title:
Perceived Anonymity and Online Self-Disclosure
in an Online Social Support Community

Researchers: Prof. Robert LaRose lamsc'agnsu.orig USA 517 353-6336
Haej in Yun, Doctoral Candidate you {13] 5151112311,, grin USA 517 485-3 852
Department of Telecommunication, Information Studies, and Media -
Michigan State University

 

More and more people enter the Internet to meet people willing to provide information and
emotional support, and tend to disclose themselves to people whom they never have met before in
person. Among many factors related to the participation in online social support communities,
this study focuses on people’s perceived anonymity in terms of how this factor affects the way
people talk about themselves online. For this purpose, this survey asks how you feel about
yourself in your daily life and in an online community where you have membership.

This online survey will take 30 minutes or less to complete. Participation is completely voluntary
but, of course, greatly appreciated. Your responses to this survey will remain strictly conﬁdential.
Your privacy will be protected to the maximum extent allowable by law. The online survey site
does not utilize any device such as cookie for tracking your identity as commercial sites usually do.
Your answer will be aggregated in a summary report. Only the researchers listed above will be
allowed access to the data. You may decline to answer certain questions or at any point may
discontinue your participation.

Providing your user ID and email address at the end of the survey is completely voluntary, but
those details will be used to send you an electronic gift certiﬁcate in appreciation of your
participation and completion of the survey. Within a couple of days after it is conﬁrmed that you
completed the survey, an Amazon.com electronic gift certiﬁcate in the amount of U835 will be
delivered to your email account. After you receive a certiﬁcate, your user ID and email addresses
will be deleted from the data set, to protect your identity.

While this study is not expected to yield any immediate beneﬁt to individual participants, it will
add to the general store of knowledge about people and the Internet. There are no anticipated risks
associated with participation.

If you have any question on this study, please contact the investigator (Professor Robert LaRose),
at CAS 413, Department of Telecommunication, Information Studies and Media, Michigan State
University, East Lansing, Michigan, 48824; lugggsggfvulsﬁu,cgg and/or phone USA 517 353-6336. If
you have any question or concerns regarding your rights as a study participant or at any time are
dissatisﬁed with any aspect of this study, you may contact - anonymously if you wish - Peter
Vasilenko, Ph.D., Chair of the University Conunittee on Research Involving Human Subjects
(UCRIHS); gcrihs-(a‘insuedu and/or phone USA 517 432-4503 and/or regular mail at 202 Olds
Hall, East Lansing, MI, 48824 USA.

 

If you wish to participate, please indicate by clicking on the link below.

124

APPENDIX-D

Banner Advertisement for Subject Recruitment

125

>h_w.u:..._>_zs

1 111131.. 1.5.1

“.5th Z.<.U_:U_.—.2.
:umjwpéwmcci
“Lam“ Lin—<5
”@52va =mMr
ﬁHMfMWWv

1i. _..

1:51
......m

 

>h_w¢u>_za

II Iii... tilt!

ubhmz<ufu=¢

.EOQ.=°.NnEm
..m.lﬂ.\\\.\i\‘t..
.. ...Jﬂt‘ma_u.....$
1.1 11... . .. .4114
...;Le. 1......i:
\wu. ...,H. 111:... :1:
, \E:
. . t ..I
.. . 21....
\
lilatrlil
.. . .1;
..llubf 1"
, 1...!
. .. 7th
. 1.. 1’.
Ujmu

WI. NM H..< ﬂommw
7%.? Wm ._M_D
=m =50 38m... JWJ I

8.3%
Em gun.

m owmm

>.h._.m¢u..».._z....

“2.5 z<o:.._u_2
sum .sz®m2:§
HE mm ﬁn _< E
ﬁmﬁafﬂvmﬁw
Aﬁmmwmv
mm
EIMKQMw
Amzanmv

 

3mg?
gage

m own;

I _ ... .... s _, .3. a
uh<pm Z<U_:U_Z

wmjzzimm
3.5M W3?
WES MES
133:. :31...“

...mjﬁmms
amzms my. 2
z _ ,___ w .... 2... a

wmjmmmm

RWR 6.420”.

WELD M335
2.13m 53....

_ omen

126

Page 1.

Page 2.

“No one recognizes who you are in an online community?”
“Do you recognize anyone in an online community?”
“How much do you disclose yourself in an online community?”

“Survey that all participants win prizes”

“Anonymity and Self-Disclosure in Online Communities”

“Eligibility — Online community member”

“Researcher — Haejin Yun (Doctoral Candidate), Michigan State University,

311?»:11‘s,4§.’££;.!1_1.31.._1.£id._l;”

Page 3.

Page 4.

“Participate and win an Amazon gift certiﬁcate!”
“The ﬁrst 300 participants will receive a $5 online gift certiﬁcate.”

“For details and participating in the survey, click here.”
“Researcher — Haejin Yun (Doctoral Candidate), Michigan State University,

<JL.I_B§_!L§<_-l_ll”

!
'\ [:1 iiii'is):

127

APPENDIX E-l

Online Survey Questionnaire — Korean

128

 

 

PART 1/4
carer-ewe: crate—o1 was: areas. are ea}?! ate-war ea ism} m1 alibi
sea-e- e211 ~2- a—E-aw AME—e- Baez-swat" mean}. .2- est—e a}; :1 {Lara
aeqaowa area-ea east asxieow area-e.- xwon man hvlt— are eaten}
we.

01

Q-Fcﬁi arm 511%.. 1~5 $119}, @253- 7217111373 Bill-‘1 35.11% ’ﬂﬁ‘i’d 9:11191 7'1-‘r'i—1-1E1011
EH11} 7151-1121 €51” Lit}.

 

311% %Ei?l7l$’r1-1E191§l%l9.§ 71%131tlzl‘Q‘31L15-Iﬁb13?
k1 __ 711%]

 

sue aw arena:— a mg 2.9224 fit—tie] 21cm Met-saw

5.000 "5' ”1‘11
5.000~9.999 %’
10.000~14.999 ”3
15,000~19.999 ‘8‘
20.000~24.999 ”3
25.000~29.999 ”3
30.000~34.999 ”3'

35.000~39.999 ‘3
40.000~44.999 ”5'
45.000~49.999 “I?
50.000 ”3 01001313! 4.111510% 31114179.?! $11-$- ﬂog—HIE.

ar‘r-r-gz-ocraquocrca

09.

l

311% 0%”5‘91 3F“?— ‘ﬁﬁ‘ "J—E—Zt—r— a User. saw macaw}?
1.000 “5‘ Ulla}
moo-1,999 ta
2.000~2.999 “3
3.000~3.999 ”3
4.000~4,999 ‘73
5.000~5,999 ”3
6,000~6,999 ‘8
7.000~7,999 ‘8'
8.000~8,999 rg
9.000~9,999 ”5'
10.000~10.999 ‘S
11.000~11,999 a
12,000~12.999 $1
13.000~13.999 ‘3
14,000~14.999 ‘35
15,000~15.999 ”é!
16.000~16.999 ‘5
17.000~17,999 ”3
18.000~18.999 ‘5‘
19.000~19.999 ”3
20,000 03 oleolatsz A3215! A121 $241112)

1'

cr'wT‘DEPPS’TF'T'E'P'QQTFDP-WVP

aeaaeag. a

a
. are aeqaow crate—or we create Awe-e— are—2:1 2.21113. C}% 2:;- 01: 739—01]
614% ewe?

129

 

a 3011-1 A17.;1-3- %% [H H 0130111 010133 13.15613? 01 6161.
30M 11711—5:- 33 "H H 013010: 01011313361119.3105 5.11:1.

5. 33339.33. 3’ ”.191 331-131010181-3—01717:.1 0136 3 16133013816. 364-3- 313—6131
”H.121“ 1110711" (711 A131 3391 3.6133 613.3 3191:6H 31119..) ”3

# 711361161 6~12 £11.91 ”33:01] 3113613 “115,33 :17111331 31313- at} “13:13. 3E1“

7131-11314] tH6101 331611 36171 11131-1111

6. 3816.1 3 (39- A161) 3:3 8171?} 61*: 751391 213° 39313111 E13? £61) 0111 11751-3- °
361—:— 733-3— 13—3— 336101 3.1013 113 6H3} 3 1‘3. 913L133: 131-361*1L17/1’ 01213-91
7.13011 71101 7101—3— $113113- 113611 31119..

a. 7191 01%} (3133.101 5~7 3.1)

1). 313931011 311%!

c. 3331011613 013

(1. 3.1310” 011.1%}

e. 31%011315‘?‘

f. 3*1L1‘3(2~4)7H%1°H 613-

g. 5~7 7113301] €513—

h. s~10 71131011613

i. 11~12 3931011613

j. 11.1011 613 51.61 :1 E 3111"

7/8. 38111 313% A1711):..3-3-0171111 613 75151-91 111.91 31(31311 313 £31) 01L1 A111

361—:- 733—3- 51—3— 33161013113323. 3 9.1 61% 311513- “.-.1-3.—61111-1771O
7 . 0 L1 8 ._ 3 CH 3361*. 3.0.; 31613113..

7. 213—220“ 1°.)

8. 131011 ( )3 (101613233 2121532151E .3340] 710 310111. 0113— 3:01
613°- °‘ 73° .0.53 7131613610.)

9. 6H%1 3819.1 313L133 ”013—61-3— 30113.3 61 3.113%?= A16.1) >33.— 3171‘2. 6 ~3— 73391
211.1 °1=3<°1 3311 '33 £31) °1L1 A161% 3613 73° 3‘3 3.3— 136101 615?— Edit 01
13591617..” :1 30111131111771? ( )6111( )3—

10. 3136.1 7.1%. 3714. 336.1 7331-3 L1—‘7-7‘11-1 01813—91 3230111 1121—3 {5711-1 E13
E1713. 1161-3- 3613 3.310111 Z1—r 6H%1 71111-151011 =" 38 11:3 3 3EWL1771?
0181—3-91 33°11 717:1 71771:?- 331.2 ° {1361 T1113.

3. 7191 I113.1 (3.133011 5~7 Ci=1)
b. 339-1011110131.

c. 313—24011 2.13131.

d. 6.1%} 13111111.

e. @213131‘91

f. 31013 (2~4)7H§°H 6131

g. 5~7 71131011 3.131

h. 8~107H%°11 611211.

i. 11~127H§°H 31131.

1'. 11301161131. £131 E15371]

130

 

&

 

 

11. 333 213 321111.333 23.1 3 L1-. 21L1 01213 °3301L1 3213 3.213. 333
3213. 2133 333 3. 33011 117.33 12' 71121 3- <33 212.113 22113 913312101
3a13112212 1 17111 1 01 31.21.33 21a1 32133 3301 2133143. 0113 301.
33011 3 21131 2.13. 0.53 713 611 32110.1

12.313 3391 2131431011 3 («.1321 1:13 3310101 2121. 13 33 011. 3 .33 213 31.1 3
332123 01.: 2.1301 213013a13L121°
1 12131 13

PART2/4

01211 013301 31213 7131131011 30121131 31 01213 213011 111811 013711 3.21213 21
01312131101. 33 3712133 111111 3:113 3213 3 33.121311510113301 11.13311
32121 0131401 21213 32.101 013301 .1221 213 013 01.: 33 33161 1112131321 7 3
2133 33132171 013113.

 

1 361 13121 931:1; 2 2321 93:1; 3 ‘33 3.321 931:1:
4 111301111;
5 33:33:13 23121: 7 °13 131:1

1 2 3 4 5 6 7
6132131111 1351-)
13. °11£1 513.133 L11°133 31°13 31 73151
14. 01111. 319.133 111 *1 21010114331: <3°13 31 7.11:1.
15- 31113131133 0133.343 1012135333011 " 913 a1 31:1
16. 0113 313133- 111 010131133 031013 31 3.11:1.
17. 0111.1 31 °133 111 L1°13 3334‘— 213— 7A1 351.
18.0115 31 °133 L191 33 3151133334“- ‘213— 7A1 3C1
19. 0113 31313-3— 111713721-3— 1:1L1311 33— 2191101 331161 3°13 a1 3C1
20. 0111i 32133331413333301321 1 .
21. 01111 31° 33— °111 71 .121 332.5153 301331 731131.
22. 0113:6133- 5111113313 13 3.313321 73151.
23. 0131. 5133 L1171T3°1 61 1191 3°13 3:619:13 3°13 31 351
24. 01111 191.33- L1171-r3°111 11171113 23:3..— 33°IL113 33°13 31 31121
25. 01111 3113133— L191 L1171313113 6175.133— 3361011 L13 3°13 7;.1 3C1
26. 315—0115313130 031 3—3333 011:1.
27. L13 01 .1 31131391 A13210101143 333T * 9.1111
28. 1413 01 .1 51331391 01313311131014 3333 3.101 3 3F 213— 313:1.
29. L13 0101:113— 5330111131133 3332111301351.
30. L11: 11111113513 £19.33 91301333333311.

31. L13 111111113513— 51313 9133315113 33 3"“14.

32. L13 “1111113 I113 31313 01 317113- 14113—3133- 013 21337111112132] 3
33. L13111111113143 $112,394 1132133333 213

34.113 111111113 13133313391 717111—311 z353 333-1- * 2151

35. L1.—‘:- 111111; £13313. 4 41011131312113 333-”? 21131.

36.113 325-1:— 3.131313 31331313.: 011 39.121 o31°13 * 01:1

37. 3213153 3610114 31°13 1:133:13 01 T312121 °1°1 69.114.

1111

3333

131

 

38.11:: 1111711; 61°11- 112101111. 1171112—13; (117:1. 1 111011191131 1.1L1.

39. 01 131—61°J 7111L1L1°1|A111 717101 L11 711.1011 L111} 1113 412.1 A111.11 131121 1 91:1.
40.1911110151111L161011111: 717101 L191 71111131121 {:1 (135811-21 1’1 21L1.

41. 01 1E1°J 71114131011111:— 7171101111712; ,1. 11:3. 712-011 1611010171111 1’1“— 21L1.

42. 01 1E1°J 7111116101111? 717101 LH 711;— 1111 :12-11 211751 111—1.17111: #1111911- 7,1311.

010171111?- 31L1.
43. 011a1°J7111L1E1°111117171°1 L191 1131§1IJ 73111-2 C>11L11is1111°1°17l1—1—1 " o1L1
44.0111a10J711-1L1E101111t— 71710111191 @1711 J1: 1:31LH6HA1010171 111-191
45. 01 9—E1°J 911111qu101111: 7177101L1°1*1711°11L116 *1 0101:7121-1— * °1L1.
46. 01 131—61°J 7111-1151011111: 717101 L11 7111101 ﬁle 11L L111? °1L1.
47. L11:- 01 1E1?! 71111L1E1-91 0101 7121 7101:: 71 1519— §1°1 1‘1 61L1°171L1.
48. 01 1E1°J 71111115111 011- 11911.1 L1i11°1L1.
491117101 11112.! 9‘111L1E1 11119.1 311'1L1171L11 74&1.- 01113-711 51.1711: 11 1:11.10] 31L1.
50. 01 1B1°J 711111111311: L11711‘1' 11°J711— 51.01111L1
51. L11:— 0] 18131°J 71111145101 L11: 111—.1271L1EL1.
52. L11.- L11: 1111131101017111 u111a- 1- 1} 01161°J7111L161 a @1111
53. 0113101711r111€14§1111211m1r6h 11% 11171L1.
54.#1111110]17111111181111111113101L1.
55. L111 5111011111- 26111; 71°1L1.
56. L11_'-— 51111101 L11711f- H1111- 111011‘11 71°1L1.
57. L111 6111101401111191617181—3- 710114.
58. L11:— 511121-13 L1171 13111 1.101L1/117lo111a11L11111 31°1L1.

11

6161.

m

 

 

HWUBM
011213111: 01315191 24%‘32‘711 1111811 %311} 12111111.

 

 

-’1‘—%°ﬂ (§~%)

 

 

59 61111.61 11:21 7111. 161 11 1247:1131—2- 21111 1.51211: 111—g- 21121121 61712112161 711—611.:
611- 211-1216111231» 211114617 1111 1

712.1611 (E a)

60. 6121 61 1111.1. 7171.161 1- ‘11 01.111213111211- 112.11- 1112111 6171112161 1216111-
61130161.:212L1si1111'1w1_A1z_1 1':

 

# 2111811111 6161191 18161 13617-2 11°15, 2172111171131 61211 2111011 611311 :_—.7;1.:_= 111% 2131
211-1171 6111161. 114% 12.161 61111:— 21731 c1111 1:611- B11 111» 211616121 116111131.
21121. 7 a 215611 6161 111.1311 11171.2.

1 79.31] 1%21 941131: 2 3.121 ‘31:}:3 $3. 1117‘] 941131:

4 51.101131;
5 211- :11L1:6 :1L1; 7 01111L1

132

1 2 3 4 5 6 7
<- JEM earn} :ﬁt} ->

6:. ulvrvlww 0164% 2a. él-g—oi Faun Ln °H71§ 2?: gerw}%o 1315-9101] 0121.01: '
62- LH7}€%*§%°11+1117}:33;1% “H LH °H71§3 2221-01? {PEG/W30 I
63. AIwN—Tl-Vgon um LH :3 3.3. £01 an: 5;»?an 2530" 0104393
64. 0H1 °a‘°la 7“ % 5191% “H 5.?1_ —-*}%‘°l T9401] 21212.6 {EEC}.
65. LR 5:11: '-—% “91%?4'7é01 ¥~ °1°ﬂ 21% .

66. ‘3 2:301] EHSH E7373: LHBiOk <2} “41.12%9; {£51711 4% I T1401] “91.0. 0 ﬁt}.

 

 

67. $1232: M12! * °':iﬂ%6H%T-APE3°1%$01 21222.21; 2w

68. Lug—211%%snes}—Ewon unawaammwa-g-w of] “1011 m 95 23C}
69. 2123-] 7u<212421 7‘1401L}$B4%%L}-‘1§* ‘1}0] BIL-$4011 «2121:». M4

PART4/4

1": ’25:“1'3‘3“ 59'1”] 73"}ant}. “Write-i. "Siﬁli‘lag $172} $71” 7132519 E'E'g‘
—"‘4317~}‘: @145} 915*]. NIH]?! 515.41% ‘éﬁ-rl‘va, 3311i 9:3}?! ’&%?i (“131’ 5% ’2}?
Amaxon. com ’9%‘ i)% ELH— €151qu °]“"‘§ ‘?"% "371-13 7,3 ° 01615—4 74%3 °1
914€§°ﬂ 331 2184—15}. clung 2:55: 123:3}?! ’8%i% ‘32.” 7;3°] ERIE]: "tﬂi
@3529] 31°]EMIH4‘1’HIE 723%}‘44.

 

70. Q3 93 DJ *1]

71. 0151101] NEIL—$112] (20%} Elﬂiqv}? _ki ﬁlﬁ

 

72.°4€1%7H§9931$%€%10}L}El~:—Zlg?

$30,000 “19}
$30,000~$59.999
$60.000~$99.999
$100,000 01%

0.9699

73. 04a ’39} 31%‘233—3 OHS-71151403.?
21. 3135-7351
b. Uliﬂm
c raga/514% an»
d 71E}

74. 0101192 $31273“ 340151503.

 

133

 

APPENDIX E-Z

Online Survey Questionnaire — English

134

 

 

PART 1/4

You are led to this online survey site by clicking a banner ad for recruiting survey
participants in an online community where you have membership. This survey asks you
about your experiences in the online community and how you feel about yourself in
daily life.

Questions 1 to 5 ask you about the online community where you were exposed to the
recruitment banner ad.

 

How long have you been a member of the online community?
year(s) month(s)

How many members do you estimate the online community has?
Under 5,000

5,000 to 9,999

10,000 to 14,999

15,000 to 19,999

20,000 to 24,999

25,000 to 29,999

30,000 to 34,999

35,000 to 39,999

40,000 to 44.999

45,000 to 49,999

50,000 and over Please specify

r‘r'r'vqom 9.09"!»

How many members do you estimate visit the online community a day?
Under 1,000
1,000 to 1,999
2,000 to 2,999
3,000 to 3,999
4,000 to 4.999
5,000 to 5,999
6,000 to 6,999
7,000 to 7,999
8,000 to 8,999
9,000 to 9,999
10,000 to 10,999
11,000 to 11,999
12,000 to 12,999
13,000 to 13,999
14,000 to 14,999
15,000 to 15,999
16,000 to 16,999
17,000 to 17,999
18,000 to 18,999
19,000 to 19,999
20,000 and over Please specify

Frau-evsvpar‘rr‘reqorme-csre

Think about your favorite message board in the online community. To which case does the
message board belong?

135

 

a. The message board requires me to reveal my name and user ID when posting a message or
a picture.
b. I don’t have to reveal my name or user ID when posting a message or a picture.

5. How many members do you estimate read messages posted on your favorite message board on
average? (You may want to look at the hit numbers for each posted message on the message
board for estimation) members

# For questions 6 to 12, please continue to answer regarding the online community where
you were exposed to the recruitment banner ad in mind.

6. How often do you visit the online community, including reading posted messages (or pictures)
and posting your own messages (or pictures)? Please choose the choice that is the closest to
your case.

Almost everyday (5~7 days per week)

Several days per week

One or two days per week

Two or three days per month

One day per month

One day per two to four months

One day per five to seven months

One day per eight to ten months

One day per eleven to twelve months

Less than one day per year

T‘"'F'Q°?">f°P-P.°‘P

7/8. How many days on average do you visit the online community, including reading posted
messages (or pictures) and posting your own messages (or pictures)? Please answer in a more
appropriate way between #7 and #8.

7. ( )days per week

8. ( ) days per month (You may answer in decimals. For example, if you visit the online
community once per two months, please enter 0.5)

9. If you visit the online community, how much time do you spend, including reading posted
messages (or pictures) and posting your own messages (or pictures)? ( ) hours ( )
minutes

10. How often do you post messages in the online community, including asking questions, sharing
information, writing your feelings and thoughts, answering to others’ messages, or posting
pictures. Please choose the choice that is the closest to your case.

Almost everyday (5~7 days per week)

Several times per week

One or two times per week

Two or three times per month

Once per month

Once per two to four months

Once per ﬁve to seven months

Once per eight to ten months

Once per eleven to twelve months

Less than once per year

‘r'r'P‘qomaensre

136

1 1. How often do you post messages in the online community, including asking questions, sharing
information, writing your feelings and thoughts, answering to others’ messages, or posting
pictures. ( ) times (You may answer in decimals. For example, if you visit the online
community once per two months, please enter 0.5)

12. How much time do you spend on average when you post messages or pictures in the online
community? ( ) hours ( ) minutes

 

 

PART 2/4

Now, we would like to ask you how you feel about yourself when you participate in the
online community in which you were exposed to the recruitment banner ad. Please
indicate how much you agree with each of the following statements ranging from 1,
Strongly Disagree, to 7, Strongly Agree.

 

 

l Strongly Disagree; 2 Disagree ; 3 Disagree to some extent;
4 Neutral;

5 Agree to some extent; 6 Agree; 7 Strongly Agree

(I!
O5
\l

l 2 3 4
(- Disagree Agree 9

13. Some members can recognize my name.

14. Some members can recognize my usemame.

15. Some members may find out my email address or homepage address.

16. Some members can recognize my IP address.

17. Some members can guess how old I am.

18. Some members can tell my marital status.

19. Some members can tell my profession.

20. Some members can tell how much education I have had.

21. Some members can tell our household income level.

22. Some members can tell my hobbies or interests.

23. Some members can recognize me from my writing style.

24. Some members can recognize me from expressions or words I use frequently.
25. Some members may match me with pictures I posted.

26. I can recognize the names of some members.

27. I can recognize usemames of some members.

28. I may ﬁnd out email addresses or homepage addresses of some members.
29. I can recognize some members via their IP addresses.

30. Sometimes, I can guess how old other members are.

31. Sometimes, I can tell the marital status of other members.

32. Sometimes, I can tell the profession of other members.

33. Sometimes, I can tell how much education other members have had.

34. Sometimes, I can tell the household income level of other members.

35. Sometimes, I can tell hobbies or interest of other members.

36. I can recognize some members from their writing styles.

37. I can recognize some members from expressions or words they use frequently.
38. Sometimes, I can match other members with pictures they posted.

137

39. I am willing to reveal negative things about myself in this online community.

40. I am willing to express my most intimate feelings in this online community.

41. 1 am willing to share what I did wrong in this online community.

42. I am willing to share what 1 would not do with my family, my off-line friends and colleagues
at work in this online community.

43. I am willing to talk about my shameful experiences in this online community.

44. I am willing to talk about my hurt feelings in this online community.

45. I am willing to talk about my failures in this online community.

46. I am willing to share my family history or secrets in this online community.

47. I feel I am one of the least contributing members in this online community.

48. I feel that this online community is worthwhile.

49. My membership in this online community has little to do with how I feel about myself.

50. I think of this online community as part of who 1 am.

51. I see myself as different from other members of this online community.

52. I often cite this online community when I talk to others off-line.

53. I enjoy interacting with the members of this online community.

54. Other members will criticize what I posted.

55. Other members will misunderstand me.

56. Other members will dislike what I posted.

57. Other members will disagree what I posted.

58. Other members will oppose what I posted.

 

 

PART 3/4
We also would like to ask you about your daily life.

 

 

0n Weekdays (Mon to Fri)
59. How much free time - excluding housekeeping. child rearing, study, work and so on — do you
have on a typical weekday on average? hour(s) minutes

0» Weekend (Sat to Sun)
60. How much free time - excluding housekeeping, child rearing, study, work and so on — do you
have on a typical weekend day on average? hour(s) minutes

# Please continue to answer the following questions regarding your daily life. Indicate how
much you agree with each of the following statements ranging from 1, Strongly Disagree to 7,
Strongly Agree.

1 Strongly Disagree; 2 Disagree ; 3 Disagree to some extent;

4 Neutral;
5 Agree to some extent; 6 Agree; 7 Strongly Agree

1 2 3 4 5 6 7
(- Disagree Agree 9

61. I wish I had someone who listens to me when I struggle with my life in the US
62. 1 wish I had someone who listens to me when I have a marital problem.

63. I wish I had someone who listens to my complaints about in-laws.

64. I wish I had someone whom I can ask for advice when things go wrong.

138

65. I wish I had someone with whom I can talk about my problems.

66. I wish I had I someone who helps me decide things.

67. I wish I had someone whose advice I really trust.

68. I wish I had someone who can provide objective feedback about how I am handling my
problems.

69. I wish I had someone with whom I can share my most private worries and fears.

 

 

PART 4/4

Thank you for responding to this survey. As the last set of questions, we would like you to
answer several demographic questions. Also, if you provide your email address, we will
send you an email gift certificate ($5 Amazon.com certiﬁcate) in appreciation of your
participation. They will be deleted from your data after we conﬁrm that you receive an
email gift certiﬁcate.

 

70. Your age? years

71. How long have you lived in the United States?
year(s) month(s)

72. What is the annual income level of your household?
a. Under $30,000
b. $30,000 ~ $59,999
c. $60,000 ~ $99,999
(1. $100,000 and over

73. What is your education level?
a. High school graduates
b. College graduates
c. Master’s or Doctorate
(1. None of the above

74. Your email address

 

139

 

APPENDIX F

Correlations for All Observed Variables

140

 

Need for Social Support

 

 

 

 

 

 

 

 

 

 

 

PCS nssl nssZ nss3 nssS nss6 nss7 nssS nss9 nssl 1
PCS
nssl 0.019
nss2 0.063 0.906
nss3 0.034 0.754 0.806
nssS -0.002 0.810 0.835 0.768
nss6 0.046 0.820 0.858 0.787 0.892
nss7 0.038 0.802 0.834 0.762 0.864 0.894
nssS 0.080 0.742 0.791 0.701 0.816 0.826 0.875
nss9 0.066 0.764 0.784 0.711 0.808 0.826 0.870 0.858
nssll 0.031 0.770 0.810 0.765 0.767 0.823 0.806 0.748 0.796
ec2 -0.091 0.066 0.053 0.053 0.053 0.054 0.050 0.003 -0.006 0.027
ec3 -0.100 0.030 0.016 0.017 0.011 0.033 0.030 -0.038 -0.042 -0.001
ec4 -0.061 -0.021 -0.038 -0.006 -0.058 -0.012 -0.042 -0.108 -0.118 -0.041
ec5 -0.028 -0.018 -0.021 0.008 -0.042 0.025 -0.004 -0.065 -0.071 -0.025
ec6 -0.062 -0.001 -0.013 0.022 0032 0.013 -0.013 —0.075 -0.070 -0.014
opdl 0.004 0.203 0.193 0.152 0.136 0.139 0.157 0.078 0.104 0.185
opd2 -0.024 0.292 0.291 0.263 0.219 0.209 0.208 0.152 0.168 0.260
opd3 -0.024 0.221 0.219 0.200 0.164 0.150 0.171 0.115 0.150 0.195
opd4 0.090 0.280 0.273 0.287 0.230 0.214 0.234 0.184 0.210 0.271
opd5 0.080 0.248 0.244 0.244 0.185 0.175 0.196 0.154 0.183 0.233
opd6 0.060 0.273 0.273 0.284 0.225 0.194 0.218 0.199 0.208 0.247
opd7 -0.018 0.264 0.268 0.273 0.232 0.208 0.229 0.180 0.193 0.261
opd8 -0.031 0.223 0.213 0.241 0.151 0.150 0.142 0.082 0.099 0.223
sa12 0.078 -0.121 -0.112 -0.103 -0.063 -0.095 -0.107 -0.083 -0.082 -0.095
oa12 0.103 0100 -0.100 -0.088 -0.061 -0.102 -0.094 -0.062 -0.062 -0.101
oall 0.082 -0.127 -0.105 -0.126 0.073 -0.113 -0.086 -0.048 -0.056 -0.091
sal 1 0.027 -0.079 -0.069 0103 -0.017 -0.083 -0.039 0.002 -0.010 -0.039
oa5 0.154 -0.076 -0.081 -0.098 -0.113 «0.089 -0.124 -0.075 -0.128 -0.067
oa6 0.195 -0.065 -0.066 -0.028 -0.078 -0.062 -0.099 -0.066 -0.072 0.026
oa7 0.120 -0.140 -0.107 -0.087 -0.119 -0.122 -0. 179 -0.106 -0.122 -0.087
oa8 0.097 -0.208 -0.186 -0. 160 -0.180 -0. 190 -0.243 -0.160 -0.160 -0.131
oa9 0.015 -0.1 19 -0.069 —0.081 -0.123 -0.109 -0.175 -0.140 -0.124 -0.078
oa10 0.147 -0.178 -0.169 -0.127 -0.154 -O.19l -0.216 -0. 149 -0.176 -0.128
sal 0.007 -0.157 -0.136 -0.060 -0.059 -0.090 -0.135 -0.102 -0.073 -0.101
sa2 0.027 -0. 140 -0.130 -0.069 -0.051 -0.082 -0.123 -0.1 17 -0.081 -0.140
533 0.017 -0. 160 -0.119 -0.118 -0.061 -0.107 -0. 160 -0.1 17 -0.120 -0132
sa4 0.096 -0.073 -0.055 -0.046 -0.024 -0.033 -0.084 -0.045 -0.034 -0.045
535 0.122 -0.141 -0.089 -0.075 -0.066 -0.074 -0.109 -0.078 -0.081 -0.051
537 0.131 -0.116 -0.051 -0.058 -0.032 -0.056 -0.112 -0.063 -0.090 -0.01 1
sa8 0.080 -0. 197 -0. 150 -0.1 13 -0.121 -0. 146 -0.209 -0.125 -0. 144 -0.090
539 0.058 -0.100 -0.052 -0.098 -0.034 -0.081 -0.1 17 -0.064 -0.056 -0.066
Total
M 9.256 5.252 5.213 5.039 5.219 5.306 5.367 5.548 5.449 5.186
SD 6.095 1.466 1.510 1.624 1.474 1.483 1.442 1.390 1.384 1.540

 

141

 

(continued)

 

Eavaluation Concern

Online Public Disclosure

 

 

 

 

 

 

 

 

 

 

 

ec2 ec3 ec4 ec5 ec6 opdl opd2 opd3 opd4 opd5
PCS
nssl
nssZ
nss3
nssS
nss6
nss7
nssS
nss9
nssll
cc2
ec3 0.855
cc4 0.746 0.814
ec5 0.737 0.809 0.856
ec6 0.734 0.785 0.871 0.896
opdl 0.106 0.154 0.138 0.151 0.174
opd2 0.150 0.178 0.142 0.149 0.163 0.749
opd3 0.139 0.165 0.131 0.152 0.194 0.789 0.864
opd4 0.090 0.104 0.079 0.1 10 0.121 0.658 0.783 0.818
opd5 0.075 0.099 0.086 0.104 0.123 0.679 0.756 0.815 0.851
opd6 0.072 0.096 0.097 0.127 0.130 0.646 0.754 0.804 0.819 0.892
opd7 0.080 0.079 0.068 0.083 0.094 0.612 0.714 0.729 0.737 0.784
opd8 0.129 0.177 0.213 0.136 0.189 0.580 0.639 0.651 0.612 0.666
5312 -0.214 -0.254 -0.203 -0.232 -0.238 -0.l7l -0.189 —0.212 -0.191 -0.141
oa12 -0.204 -0.259 -0.l75 -0.220 —0.225 -0.176 -0.231 -0.241 -0.230 -0.158
oall -0. 157 -0.217 -0.246 -0.215 -0.252 -0. 143 -0.l66 -0.157 -0. l 35 -0.109
sall -0.188 -0.227 -0.278 0239 -0.266 -0.157 -0.l64 -0.l85 -0. 144 -0.11 l
035 0.077 -0.080 -0.082 —0.052 -0.076 -0.065 -0.074 -0.051 -0.010 -0.041
oa6 -0.059 -0.032 -0.046 -0.039 -0.058 -0.105 -0.077 -0.066 -0.009 -0.036
oa7 -0.127 -0.137 -0.134 -0. 131 -0.156 -0.l40 -0.125 -0.1 19 -0.105 -0.096
oa8 -0.086 -0.087 -0.129 -0.108 -0.l47 -0.l97 -0.162 -0.156 -0.153 -0.l4l
oa9 0178 -0.175 -0.218' -0.179 -0.219 ~0.114 -0.123 -0.142 -0.134 -0.l32
oa10 -0.066 -0.062 -0.082 -0.080 -0.079 -0.150 -0.112 -0.107 -0. 100 -0.098
sal -0.1 12 -0.217 -0.198 -0.233 -0.235 -0.097 -0.097 -0.1 16 -0.134 -0.107
sa2 -0.046 -0.183 -0.138 -0.177 -0. 154 -0.1 14 -0.121 -0.127 -0. 154 -0. 141
sa3 -0.106 -0.209 -0. 193 -0.201 -0.207 -0. 137 -0.060 -0.078 -0.064 -0.080
sa4 -0.176 -0.256 -0. l 89 -0.219 -0.220 -0. 130 -0.097 -0.1 18 -0.106 -0.102
sa5 -0.216 -0.290 -0.194 -0.211 -0.214 -0.150 -0.155 -0.150 -0.1 19 -0.121
sa7 -0. 191 -0.242 -0.180 —0. l 79 -0.188 -0.110 -0.112 -0.104 -0.110 -0.112
538 -0.l99 -0.256 -0.188 -0.207 -0.l97 -0.237 -0.142 -0.l71 -0.161 -0.189
sa9 -O.266 -0.360 -0.296 -0.252 -0.298 -0.125 -0.122 -0.098 -0.070 -0.1 13
Total
M 2.927 2.814 2.677 2.708 2.577 3.734 4.070 3.097 4.163 4.143
SD 1.354 1.378 1.285 1.271 1.240 1.715 1.697 1.700 1.743 1.714

 

142

 

(continued)

 

Online Public Disclosure

Discursive Anonymity

Other-Anonymity

 

 

 

 

 

 

 

 

 

 

 

 

opd6 opd7 opd8 sa12 0312 oall sall oall oa6 oa7
PCS
nssl
nssZ
nss3
nssS
nss6
nss7
nssS
nss9
nssll
ec2
ec3
ec4
ec5
ec6
opdl
opd2
opd3
opd4
opd5
opd6
opd7 0.792
opd8 0.648 0.745'
sa12 -0.l42 -0.159 -0.126
oa12 -0.167 -0.157 -0.144 0.844
oall -0.062 -0.085 0142 0.570 0.576
sall -0.086 -0.113 -0.153 0.533 0.560 0.876
035 -0.030 -0.090 -0.044 0.304 0.275 0.414 0.405
oa6 -0.014 -0.061 -0.009 0.222 0.179 0.354 0.322 0.354
oa7 -0.076 -0.l33 -0.069 0.279 0.269 0.435 0.383 0.435 0.719
038 -0.106 -0.156 -0.130 0.299 0.295 0.451 0.393 0.451 0.618 0.733
oa9 -0.102 -0.l24 -0.075 0.299 0.297 0.435 0.393 0.435 0.516 0.633
oa10 -0.063 -0.118 -0.078 0.372 0.322 0.507 0.431 0.507 0.572 0.571
sal -0.115 -0.086 -0.057 0.406 0.362 0.237 0.229 0.237 0.100 0.165
532 -0.135 -0.075 -0.038 0.411 0.395 0.266 0.244 0.266 0.014 0.118
sa3 -0.067 -0.038 -0.084 0.503 0.457 0.350 0.328 0.350 0.094 0.233
sa4 -0.102 -0.083 -0.069 0.459 0.442 0.344 0.323 0.344 0.111 0.202
sa5 -O.138 -0.137 -0.147 0.522 0.487 0.381 0.372 0.381 0.260 0.310
537 -0.077 -0.071 -0.051 0.425 0.379 0.297 0.247 0.297 0.321 0.481
sa8 -0.167 -0.149 -0.092 0.428 0.390 0.292 0.281 0.292 0.379 0.443
539 -0.076 -0.039 -0.087 0.432 0.393 0.337 0.304 0.337 0.187 0.301
Total
M 4.199 4.093 3.603 4.219 4.385 4.631 4.781 4.631 3.580 4.116
SD 1.724 1.714 1.794 1.691 1.669 1.715 1.704 1.715 1.678 1.578

 

143

 

(Continued)

 

Other-Anonymity
oa8 oa9

oalO

Self-Anonymity

sal

$32

sa3

sa4

sa5

sa7

sa8

 

PCS

 

nssl
nssZ
nss3
nssS
nss6
nss7
nss8
nss9
nssll

 

ec2
ec3
ec4
ec5
ec6

 

opd 1
opd2
opd3
opd4
opd5
opd6
opd7
opd8

 

salZ
oa12
oall
sall

 

oa5
oa6
oa7
oa8
oa9
ca 1 0

0.648
0.618 0.516

 

sa 1
sa2
sa3
sa4
sa5
sa7
538
539

0.176 0.193
0.153 0.114
0.178 0.177
0.166 0.255
0.257 0.296
0.391 0.402
0.457 0.374
0.338 0.459

0.194
0.126
0.213
0.214
0.306
0.304
0.307
0.267

0.727
0.569
0.391
0.487
0.430
0.354
0.429

0.631
0.468
0.498
0.447
0.375
0.409

0.621
0.553
0.461
0.423
0.504

0.597
0.435
0.396
0.450

0.595
0.541
0.510

0.683
0.540

0.520

 

Total

 

SD

 

4.123 4.468
1.588 1.597

3.794
1.551

 

4.243
1.902

4.412
1.806

4.472
1.779

4.663
1.821

4.096
1.637

4.286
1.700

4.286
1.544

 

144

 

(continuery

 

 

 

 

 

 

 

 

 

 

 

 

 

Low 01 High GI
sa9 M SD M SD

PCS 8.556 5.641 10.050 6.512
nssl 4.956 1.438 5.589 1.430
nssZ 4.913 1.510 5.553 1.441
nss3 4.819 1.625 5.404 1.572
nssS 5.058 1.497 5.555 1.406
nss6 5.031 1.515 5.617 1.387
nss7 5.088 1.477 5.683 1.337
nss8 5.316 1.459 5.812 1.263
nss9 5.175 1.443 5.759 1.247
nssll 4.875 1.569 5.539 1.432
ec2 2.994 1.343 2.851 1.368
ec3 2.856 1.354 2.766 1.407
ec4 2.767 1.290 2.574 1.277
ec5 2.819 1.288 2.582 1.243
ec6 2.669 1.242 2.472 1.233
opdl 3.531 1.602 3.965 1.814
opd2 3.725 1.617 4.461 1.705
opd3 3.613 1.656 4.241 1.694
opd4 3.788 1.673 4.589 1.728
opd5 3.820 1.633 4.511 1.735
opd6 3.894 1.654 4.546 1.742
opd7 3.669 1.620 4.574 1.696
opd8 3.335 1.696 3.908 1.859
5312 4.450 1.577 3.957 4.163
oa12 4.581 1.568 4.163 1.755
oall 4.750 1.712 4.496 1.714
sall 4.888 1.656 4.660 1.756
oa5 4.119 1.472 3.851 1.617
oa6 3.669 1.628 3.479 1.734
037 4.188 1.493 4.035 1.671
oa8 4.225 1.483 4.007 1.697
oa9 4.581 1.523 4.340 1.673
0310 3.956 1.514 3.610 1.576
531 4.531 1.870 3.915 1.892
sa2 4.594 1.792 4.206 1.807
sa3 4.625 1.769 4.298 1.780
sa4 4.673 1.786 4.652 1.867
sa5 4.269 1.593 3.901 1.670
sa7 4.450 1.516 4.099 1.876
sa8 4.338 1.466 4.227 1.632
sa9 5.181 1.409 5.000 1.521
Total (N=l60) (N=141)

M 5.096

SD 1.463

 

 

I45

REFERENCES

Ahuja, M. K., & Galvin, J. E. (2003). Socialization in virtual groups. Journal of
Management, 29(2), 161-185.

Altman, I., & Taylor, D. (1973). Social Penetration: The Development of Interpersonal
Relationships. New York: Holt, Rinehart, and Winston.

Alvestrand, H. (2002). Instant Messaging and Presence on the Internet. Internet Society
(ISOC). Retrieved October 2, 2004, from http://www.isoc.org/brieﬁngs/009.

Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A
review and recommended two-step approach. Psychological Bulletin, 103, 41 1-
423.

Anonymous. (1998). To reveal or not to reveal: A theoretical model of anonymous
communication. Communication Theory. 8(4), 381-407.

Bargh, J. A., McKenna, K. Y. A., & Fitzsimons, G. M. (2002). Can you see the real me?
Activation and expression of the "true self" on thelntemet. Journal of Social
Issues, 58(1), 33—48.

Barreto, M., & Ellemers, N. (2000a). Motivating pro-group behavior: Degree of
identiﬁcation and type of anonymity as determinants of the choice to work on
behalf of a low status group. In T. Postmes, R. Spears, M. Lea & S. Reicher (Eds),
SIDE Issues Centre Stage: Recent Developments in Studies of De—individuation in
Groups (pp. 117-130). Amsterdam, The Netherlands: Royal Netherlands Academy
of Arts and Sciences.

Barreto, M., & Ellemers, N. (2000b). You can't always do what you want: Social identity
and self-presentational determinants of the choice to work for a low status group.
Personality and Social Psychology Bulletin, 26, 891-906.

Baxter, L. A., & Montgomery, B. M. (1996). Relating: Dialogues and Dialectics. New
York: Guilford.

146

Bentler, P. M. (1990). Comparative ﬁt indices in structural models. Psychological Bulletin.
107, 238-246.

Ben-Ze'ev, A. (2003). Privacy, emotional closeness, and openness in cyberspace.
Computers in Human Behavior, 19, 451-467.

Berger, C. R. (1979). Beyond initial interaction: Uncertainty, understanding, and the
development of interpersonal relationships. In H. Giles & R. S. Clair (Eds),
Language and Social Psychology (pp. 122-144). Oxford: Blackwell.

Berger, C. R. (1987). Communicating under uncertainty. In M. E. Roloff & G. R. Miller
(Eds), Interpersonal Processes: New Directions in Communication Research (pp.
39-62). Newbury Park, CA: Sage.

Berger, C. R., & Bradac, J. J. (1982). Language and Social Knowledge. London: Edward
Arnold Publisher Ltd.

Berger, C. R., & Calabrese, R. J. (1975). Some explorations in initial interaction and
beyond: Toward a developmental theory of interpersonal communication. Human
Communication Research. 1, 99-1 12.

Berger, C. R., Gardner, R. R., Parks, M. R., Schulman, L., & Miller, G. R. (1976).
Interpersonal epistemology and interpersonal communication. In G. R. Miller
(Ed.), Explorations in Interpersonal Communication (pp. 149-171). Beverly Hills,
CA: Sage.

Bhattacheijee, A. (2002). Individual trust in online ﬁrms: Scale development and initial
test. Journal of Management Information Systems. 19(1), 21 1-242.

Birchmeier, Z., Joinson, A., .& Dietz-Uhler, B. (2005). Storming and forming a normative
response to a deception revealed online. Social Science Computer Review, 23(1),
108-121.

Blanchard, A. L., & Markus, M. L. (2004). The experienced "sense" of a virtual
community: Characteristics and processes. The Data Base for Advances in
Information Systems. 35(1).

Burnett, G. B. (2000). Information exchange in virtual communities: A typology.
Information Research. 5(4).

147

Butler, B. (2001). Membership size, communication activity, and sustainability: A
Resource-based model of online social structure. Information Systems Research,
12(4), 346-362.

Butler, 8., Sproull, L., Kiesler, S., & Kraut, R. (2002). Community effort in online
groups: Who does the work and why? In S. Weisband & L. Atwater (Eds),
Leadership at a Distance.

Campbell, K. L. (2002). Soliciting and Expressing Social Support over the Internet: An
Investigation of Online Eating Disorder Support Groups. Paper presented at the
Association for Education in Journalism and Mass Communication, Miami Beach,
FL.

Carmines, G. C., & McIver, J. P. (1981). Analyzing models with unobserved variables. In
G. W. Bohmstedt & E. F. Borgatta (Eds), Social Measurement: Current Issues
(pp. 65-1 10). Beverly Hills, CA: Sage.

C helune, G. J. ( 1975). Self-disclosure: An elaboration of its basic dimensions.
Psychological Reports, 36, 79-85.

Cohen, S., Mennelstein, R., Kamarck, T., & Hberman, H. (1985). Measuring the
functional components of social support. In G. Sarason & B. R. Sarason (Eds),
Social Support: Theory, Research and Application (pp. 73-94). Hhe Hague,
Netherlands: martines Niighoff.

Collins, N. L., & Miller, L. C. (1994). Self-Disclosure and Liking - a Metaanalytic
Review. Psychological Bulletin, 1 16(3), 457-475.

Connolly, T., Jessup, L. M., & Valacichi, J. S. (1990). Effects of anonymity and
evaluative tone on idea generation in computer-mediated groups. Management
Science, 36, 689-703.

Cooper, W. H., Gallupe, R. B., Pollard, S., & Cadsby, J. (1998). Some liberating effects of
anonymous electronic brainstorming. Small Group Research. 29(2), 147-178.

Cozby, P. C. (1973). Self-disclosure: A literature review. Psychological Bulletin, 79(2),
73-91.

148

Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests.
Psychological Bulletin, 52, 281 -302.

Cummings, J. N., Butler, R., & Kraut, R. (2002). The quality of online social relationships.
Communications ofthe ACM 45(7), 103-108.

Daft, R. L., & Lengel, R. H. (1984). Information richness: A new approach to managerial
behavior and organizational design. Research in Organizational Behavior, 6, 191-
233.

Daft, R. L., & Lengel, R. H. (1986). Organizational information requirement, media
richness and structural determinants. Management Science, 32, 554-571.

DeCew, J. (1997). In Pursuit of Privacy: Law, Ethics, and the Rise of Technology, Ithaca:
Cornell University Press.

Dennis, A. R., & Valacich, J. S. (1999). Electronic brainstorming: Illusions and patterns
of productivity. Information Systems Research, 10.

Derlega, V. J ., & Grzelak, J. (1979). Appropriateness of self-disclosure. In G. J. C helune
(Ed.), Self-Disclosure: Origins, Patterns, and Implications of Openness in
Interpersonal Relationships. San Francisco: Jossey-Bass

Derlega, V. J., & Chaikin, A. L. (1977). Privacy and self-disclosure in social relationships.
Journal of Social Issues, 33(3), 102-115.

Des Jarlais, D. C., Paone, D., Milliken, J., Turner, C. F ., Miller, H., Gribble, J., et a1.
(1999). Audio-computer interviewing to measure risk behavior for HIV among
injecting drug users: A quasi-randomised trial. The Lancet, 353(9165), 1657-1661.

DeSanctis, G., & Roach, M. (2002). Structure and Interaction in Online Discussion
Forums. Retrieved September 20, 2004, from

ht_tp://web.cbsdk/departments/int/link2002/papers/DeSanctis.pdf

Detweiller, L. (1993). Identity, Privacy and Anonymity on the Internet. Retrieved
November 28, 2004, from http://www.eserver.org/Internet/Identity-Privacy-
Anonymitytxt

149

DeVito, J. (2000). Human Communication: The Basic Course (8th ed.). New York:
Addison Wesley Longman.

Donath, J. (1999). Identity and deception in the virtual community. In M. A. Smith & P.
Kollock (Eds), Communities in Cyberspace (pp.29-59). London: Routledge.

Douglas, K., & McGarthy, C. (2001). Identiﬁability and self-presentation: Computer-
mediated communication and intergroup interaction. British Journal of Social
Psychology. 40, 339-416.

Douglas, K., & McGarthy, C. (2002). Internet identiﬁability and beyond: A model of the
effects of identiﬁability on communicative behavior. Group Dynamics: T heory.
Research, and Practice, 6(1), 17-26.

Eastin, M., & LaRose, R. (2004). Alt.support: Modeling social support online. Computers
in Human Behavior, in press.

Epstien, J. E., Barker, P. R., & Kroutil, L. A. (2001). Mode effects in self-reported mental
health data. Public Opinion Quarterly. 65, 529-550.

Fisher, D. V. (1984). A conceptual analysis of self-disclosure. Journalfor the T heorv of
Social Behavior, 14(3), 277-296.

F oddy, W. H. (1984). A critical evaluation of Altman's deﬁnition of privacy as a
dialectical process. Journal for the theory of social behavior, 14, 297-307.

Fomell, C ., & Larcker, D. F. (1981). Evaluating structural equations with unobservable
variables and measurement error. Journal of Marketing Research, 18(2), 39-50.

Fox, S. (2000). Trust and Privacy Online: Why Americans Want to Rewrite the Rules.
PEW Internet & American Life Project. Retrieved November 11, 2004, from
http://www. pewintemet.org/reports/toc.asp?report=1 06.

Franzoi, S. L., & Davis, M. H. (1985). Adolescent self-disclosure and loneliness: Private
self-consciousness and parental inﬂuences Journal of Personality and Social
Psychology, 48, 768-780.

150

Frick, A., Bachtiger, M. T., & Reips, U.-D. (2001). Financial incentives, personal
information and drop-out in online studies. In U.-D. Reips & M. Bosnjak (Eds),
Dimensions of Internet Science (pp. 209-219). Lengerich: Pabst.

George, J. F., Easton, G. K., Nunamaker, J. F., & Northcraft, G. B. (1990). A study of
collaborative work with and without computer-based support. Information Systems
Research, I, 394-415.

Goddyn, B. (2001). Defining Anonymity and its Dimensions in the Electronic World.
Retrieved September 15, 2004, from
http://users.skvnetbe/bgoddyn/publpdf/deﬁningpdf

Goffman, E. (1959). The Presentation of Self in Everyday Life. New York: Doubleday.

Goodstein, L. D., & Reinecker, V. M. (1974). Factors affecting self-disclosure: A review
of the literature. Progress in the Experimental Personality Research, 7, 49-77.

Hatch, D., & Leighton, L. (1986). Comparison of men and women on self-disclosure.
Psychological Reports. 58(1), 175-178.

Hayne, S. C ., & Rice, R. E. (1997). Attribution accuracy when using anonymity in group
support systems. International Journal of Human Computer Studies. 47, 429-452.

Henderson, S., & Gilding, M. (2004). 'I've never clicked this much with anyone in my
life': Trust and hyperpersonal communication in online friendships, New Media &
Society, 6(4), 487-506.

Henry, K. B., Arrow, H., & Carini, B. (1999). A tripartite model of group identiﬁcation:
theory and measurement. Small Group Research, 30(5), 558-581.

Hiltz, S. R., Turoff, M., & Johnson, K. (1989). Experiments in group decision making, 3:
Disinhibition, deindividuation, and group process in pen name and real name
computer conferences. Decision Support Systems, 5, 217-232.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for ﬁt indexes in covariance structure

analysis: Conventional criteria versus new alternatives. Structural Equation
Modeling, 6, 1-55.

151

Hu, Y., Wood, J. F., Smith, V., & Westbrook, N. (2004). Friendships through IM:
Examining the relationship between instant messaging and intimacy. Journal of
C omputer-Mediated Communication, 10(1).

Jaffe, J. M., Lee, Y. E., Huang, L., & Oshaga, H. (1999). Gender identiﬁcation,
interdependence, and pseudonyms in CMC: Language patterns in an electronic
conference. The Information Society. 15(4), 221-234.

Jessup, L. M., Connolly, T., & Galegher, J. (1990). The effects of anonymity on GDSS
group processes with an idea-generating task. MIS Quarterly, 14, 313-321.

Jessup, L. M., & Tansik, D. A. (1991). Group decision making in an antomated'
environment: The effects of anonymity and proximity with a group decision
support system. Decision Sciences, 22, 266-279.

Jettmar, E. M., & Rapp, M. W. (1996). Computer mediated communication: A relational
perspective. Paper presented at the Western States Communication Association,
Pasadena, CA.

Joinson, A. N. (1999). Anonymity, disinhibition and social desirability on the Internet.
Behaviour Research Methods, Instruments and C omputers. 31, 443-438.

Joinson, A. N. (2001). Self-disclosure in computer-mediated communication: The role of
self-awareness and visual anonymity. European Journal of Social Psychology, 3 I ,
177-192.

Joinson, A. N. (2005). Internet behaviour and the design of virtual methods. In C. Hine
(Eds), Virtual Methods: Issues in Social Research on the Internet. Oxford: Berg.

Joinson, A. N., & Paine, C. B. (2005, in press). Self-disclosure, Privacy and the Internet:
Privacy and Self-Disclosure Online Project.

Jourard, S. M. (1971). Self-Disclosure: An Experimental Analysis of the Transparent Self
New York: Van Nostrand Reinhold.

Jung, C. G. (1953). Two Essays in Analytical Psychology. New York: Pantheon.

152

Kang, H. S., & Yang, H. D. (2004). The eﬂects of anonymity on the usage of avatar:
Comparison of Internet relay chat and instant messenger. Paper presented at the
Americas Conference on Information Systems, New York, NY.

Katz, E., Blumler, J. G., & Gurevitch, M. (1974). Utilization of mass communication by
the individual. In J. G. Blumler & E. Katz (Eds), The Uses of Mass
Communication (pp. 19-32). Beverly Hills, CA: Sage.

Kelly, A. E., & McKillop, K. J. (1996). Consequences of revealing personal secrets.
Psychological Bulletin, 120, 450-465.

Kiesler, S., Siegel, J ., & McGuire, T. W. (1984). Social psychological aspects of
computer-mediated communication. American Psychologist, 39, 1123-1134.

King, S. A., & Moreggi, D. (1998). Internet therapy and self help groups -- The pros and
cons. In J. Gackenbach (Ed.), Psychology and the Internet: Intrapersonal.
Interpersonal and T ranspersonal Implications (pp.77-109). San Diego, CA:
Academic Press.

Klaw, E., Huebsch, P. D., & Humphreys, K. (2000). Communication Patterns in an On-
line Mutual Help Group for Problem Drinkers. Journal of Community Psychology:
28(5), 535-546.

Klein, E. E., Clark, C. C., & Herskovitz, P. J. (2003). Philosophical dimensions of
anonymity in group support systems: Ethical implications of social psychological
consequences. Computers in Human Behavior, 19(3), 355-382.

Kline, R. B. (2004). Principles and Practices of Structural Equation Modeling (2nd ed.).
New York: Guilford.

LaRose, R., Eastin, M., & Gregg, J. (2001). Reformulating the Internet paradox: Social
cognitive explanations of Internet use and depression. Journal of Online Behavior,
1(2), Retrieved from the World Wide Web:
http://www.behgvior.net/JOB/vlnl/pgradoxhtml.

LaRose, R., & Rifon, N. (2003). Your privacy is assured -- of being invaded: Web sites
with and without privacy seals. Paper presented at the e2003 conference, Lisbon,
Portugal.

153

Lea, M., & Spears, R. (1991). Computer-mediated communication, de-individuation and
group decision-making. International Journal of Man-Machine Studies, 3 9(Special
issue on CSCW and groupware), 283-301.

Lea, M., Spears, R., & de Groot, D. (2001). Knowing me, knowing you: Anonymity
effects on social identity processes within groups. Personality and Social
Psychology Bulletin, 27(5), 526-537.

Lea, M., Spears, R., Watt, S. E., & Rogers, P. (2000). The InSIDE story: Social
psychological processes affecting on-line groups. In T. Postrnes, R. Spears, M. Lea
& S. Reicher (Eds), SIDE Issues Centre Stage: Recent Developments in Studies of
De—individuation in Groups (pp. 47-62). Amsterdam, The Netherlands: Royal
Netherlands Academy of Arts and Sciences.

Lee, W., & Kuo, E. C. Y. (2002). Internet and displacement effect: Children's media use
and activities in Singapore. Journal of C omputer-Mediated C ommunication. 7(2).

Lenhart, A., Madden, M., & Hitlin, P. (2005). Teens and Technology: Youth are Leading
the Transition to a Fully Wired and Mobile Nation. Washinton, D. C.: Pew
lntemet & American Life Project.

Lessler, J. T., Caspar, R. A., Penne, M. A., & Barker, P. R. (2000). Developing computer
assisted interviewing (CAI) for the National Household Survey on Drug Abuse.
Journal of Drug Issues, 30, 19-34.

Licker, P. S. (1992). You said what? Understanding the role and function of anonymity in
GDSS. Paper presented at the Conference of the Administrative Science
Association of Canada, Toronto, Canada.

Littlejohn, S. W. (1992). Theories of Human Communication. Belmont, CA: Wadsworth
Publishing.

Luhtanen, R., & Crocker, J. (1992). A collective self-esteem scale: Self-evaluation of
one's social identity. Personality and Social Psychology Bulletin. 18(3), 302-318.

Ma, M. L. (2003). Unwillingness-to—Communicate, Perceptions of the Internet and Self-
Disclosure in IC Q. Unpublished master's thesis. The Chinese University of Hong
Kong, Hong Kong.

154

MacCallum, R. C., & Tucker, L. R. (1991). Representing sources of error in common
factor analysis: Implications for theory and practice. Psychological Bulletin, 109,
501-511.

Markus, M. L. (1987). Toward a critical mass theory of interactive media: Universal
access, interdependence, and diffusion. Communication Research, 14(5), 491-51 1.

Marsh, H. W., Balla, J. R., & McDonald, R. P. (1988). Goodness-of-ﬁt indexes in
conﬁrmatory factor analysis: The effect of sample size. Psychological Bulletin,
I 03, 391-410.

Marx, G. T. (1999). What’s in a name? Some reﬂections on the sociology of anonymity.
The Information Society.

Matheson, K., & Zanna, M. P. (1988). The impact of computer-mediated communication
on self-awareness. Computers in Human Behavior, 4, 221-233.

McKnight, D. H., Chouddhury, V., & Kacmar, C. (2002). The impact of initial consumer
trust on intentions to transact with a web site: A trust building model. Journal of
Strategic Information Systems. I I (297-323).

McKenna, K. Y. A., & Bargh, J. A. (1998). Coming out in the age of the lntemet: Identity
'demarginalization' through virtual group participation. Journal of Personality and
Social Psychology, 681-694.

McKenna, K. Y. A., Green, A. S., & Gleason, M. E. J. (2002). Relationship formation on
the Internet: What's the big attraction? Journal of Social Issues, 58(1), 9-31.

McLeod, P. L. (1997). A Comprehensive Model of Anonymity in C omputer-Supported
Group Decision Making. Paper presented at the International Conference on
Information Systems, Atlanta, GA.

McLeod, P. L., Baron, R. S., Marti, M. W., & Yoon, K. (1997). The eyes have it:
Minority inﬂuence in face-to-face and computer-mediated group discussion.
Journal of Applied Psychology. 82, 706-718.

Morris, M., & Ogan, C. (1996). The lntemet as mass medium. Journal of C omputer-
Mediated Communication, 46(1).

155

Neuman, S. (1991). Literacy in the Television Age. Norwood: Ablex Publishing.

Nissenbaum, H. (1999). The meaning of anonymity in an information age. The
Information Society, 15, 141-144.

Nonnecke, B. (2000). Lurking in Email-Based Discussion Lists. Unpublished Ph.D., South
Bank University, London.

Nunnally, J ., & Bernstein. I. H. (1994). Psychometric Theory. New York: McGraw-Hill.

Omarzu, J. (2000). A disclosure decision model: Determining how and when individuals
disclose. Personality and Social Psychology Review, 4(2), 174-185.

O'Sullivan, P. B. (2000). What you don't know won't hurt me: Impression management
functions of communication channels in relationships. Human Communication
Research, 26(3), 403-431.

O'Sullivan, P. B., Rains, S., & Grabb, J. (2001). Anonymity, Deindividuation, and
Strategic Channel Use in Personal Relationships. Paper presented at the
International Communication Association, Washington, DC.

Papacharissi, Z., & Rubin, A. M. (2000). Predictors of lntemet use. Journal of
Broadcasting & Electronic Media, 44, 175-196.

Parks, M. R., & Floyd, K. (1996). Making friends in cyberspace. Journal of C omputer-
Mediated Communication, 1(4).

Petronio, S., Martin, J., & Littleﬁeld, R. (1984). Prerequisite conditions for self-
disclosingi A gender Issue. Communication Monographs, 51(3), 268-273.

Phillips, W. (1996). A Comparison of Online, E-Mail. and In-Person Self-Help Groups
Using Adult Children of Alcoholics as a Model, from http://psydok.sulb.uni-
sggrlandde/volltexte/2004/1 17/htm1/psycvber/afcoahtml

Pinsonneault, A., & Heppel, N. (1997-1998). Anonymity in group support systems
research: A new conceptualization, measure, and contingency framework. Journal
of Management Information Systems, 14, 89-108.

156

Postmes, T., & Spears, R. (2000). Reﬁning the cognitive redeﬁnition of the group:
Deindividuation effects in common bond vs. common identity groups. In T.
Postmes, R. Spears, M. Lea & S. Reicher (Eds), SIDE Issues Centre Stage:
Recent Developments in Studies of Deindividuation in Groups (pp. 63-78).
Amsterdam, The Netherlands: Royal Netherlands Academy of Arts and Sciences.

Postmes, T., Spears, R., & Lea, M. (2002). Intergroup differentiation in computer-
mediated communication: Effects of depersonalization. Group Dynamics: Theory,
Research, and Practice, 6(1), 3-16.

Postmes, T., Spears, R., Lea, M., & Reicher, S. (2000). SIDE Issues Centre Stage: Recent
Developments in Studies of De-individuation in Groups. Amsterdam, The
Netherlands: Royal Netherlands Academy of Arts and Sciences.

Postmes, T., & Lea, M. (2000). Social processes and group decision making: Anonymity
in group decision support systems. Ergonomics, 43(8), 1252-1274.

Preece, J. (1999). Empathy online. Virtual Reality, 4, 74-84.

Prentice, D. A., Miller, D. R., & Lightdale, J. R. (1994). Asymmetiies in attachments to
groups and to their members: Distinguishing between common-identity and
common-bond groups. Personality and Social Psychology Bulletin. 20, 484-493.

Prentice-Dunn, S., & Rogers, R. W. (1982). Effects of public and private self-awareness
on deindividuation and aggression. Journal of Personality and Social Psychology,
43, 503-513.

Radin, P. M. (2001). "Hello, My Friends ": Social Capital and Transformation of Trust in
a Grassroots Breast Cancer Website. Unpublished Ph.D., University of
Washington, Seattle, WA.

Rawlins, W. K. (1992). Friendship matters: Communication, dialectics. and the life
course. New York: Aldine de Gruyter.

Reicher, S. (1984). Social inﬂuence in the crowd: Attitudinal and behavioral effects of de-
individuation in conditions of high and low group salience: Intergroup processes.
British Journal of Social Psychology, 23(4), 341-3 50.

157

Reicher, S. (1987). Crowd behavior as social action. In J. C. Turner, M. A. Hogg, P. J.
Oakes, S. D. Reicher & M. S. Wetherell (Eds), Rediscovering the Social Group: A
Self-Categorization Theory (pp. 171-202). Oxford: Basil Blackwell.

Reicher, S., & Levine, M. (1994). Deindividuation, power relations between groups and
the expression of social identity: The effects of visibility to the out-group. British
Journal of Social Psychology, 33, 145-163.

Reicher, S., Levine, R. M., & Gordijn, E. (1998). More on deindividuation, power
relations, between groups and the expression of social identity: Three studies on
the effects of visibility to the in-group. British Journal of Social Psychology, 3 7,
15-40.

Resources on C onﬁdentiality for Self-Help/Mutual Aid Support Groups. (2003). Toronto,
Canada: The Self-Help Resource Centre.

Rheingold, H. (1993). The Virtual Community: Homesteading on the Electronic Frontier.
Reading, MA: Addison-Wesley.

Ridings, C. M., Gefen, D., & Arinze, B. (2002). Some antecedents and effects of trust in
virtual communities. Journal of Strategic Information Systems, 11, 271-295.

Rubin, R. B., Palmgreen, P., & Sypher, H. E. (1994). Communication Research
Measures: A Sourcebook. New York: Guilford.

Ryan, S. E. (2003). Don 't trust anyone outside your pack: Initial trust formation in an
online social activist network. Unpublished Master's thesis. University of Toronto,
Toronto, Canada.

Sassenberg, K. (2000). The two faces of anonymity: Effects on groups decision making.
In T. Postmes, R. Spears, M. Lea & S. Reicher (Eds), SIDE Issues Centre Stage:
Recent Developments in Studies of De-individuation in Groups (pp. 93-106).
Amsterdam, The Netherlands: Royal Netherlands Academy of Arts and Sciences.

Sassenberg, K. (2002). Common bond and common identity groups on the lntemet:
Attachment and normative behavior in on-topic and off-topic chats. Group
Dynamics: T heory, Research, and Practice, 6, 161-198.

158

Sassenberg, K., & Postmes, T. (2002). Cognitive and strategic processes in small groups:
Effects of anonymity of the self and anonymity of the group on social inﬂuence.
British Journal of Social Psychology, 41, 463-480.

Shaw, L. H. (2000). Liking and self-disclosure in computer-mediated andface-to-face
interactions. Unpublished Master's thesis. University of North Carolina, Chapel
Hill, NC.

Short, J., Williams, E., & Christie, B. (1976). The Social Psychology of
Telecommunications. Chichester: Wiley.

Sia, C.-L., Tan, B. C. Y., & Wei, K.-K. (2002). Group polarization and computer-
mediated communication: Effects of communication cues, social presence, and
anonymity. Information Systems Research. 13(1), 70-90.

Siegel, J., Dubrovsky, V., Kiesler, S., & McGuire, T. W. (1986). Group process in
computer-mediated communication. Organizational Behavior and Human
Decision Process, 37, 157-187.

Snyder. J. (2004). Perceived anonymity and the role of self-disclosure. Paper presented at
the International Communication Association, New Orleans, LA.

Sosik, J. J. (1997). Effects of leadership style and anonymity on group potency and
effectiveness in a group decision support system environment. Journal of Applied
Psychology, 82, 89- l 03.

Spears, R., & Lea, M. (1992). Social inﬂuence and the inﬂuence of the "social" in
computer-mediated communication. In M. Lea (Ed.), Contexts of C omputer-
Mediated Communication (pp. 30-65). Hemel Hempstead: Harvester-Wheatsheaf.

Spears, R., & Lea, M. (1994). Panacea or panopticon? The hidden power in computer-
mediated communication. Communication Research, 21, 427-459.

Spears, R., Lea, M., & Lee, S. (1990). De-individuation and group polarization in
computer-mediated communication. British Journal of Social Psychology, 29,
121-134.

Sproull, L., & Kiesler, S. (1986). Reducing social context cues: Electronic mail in
organizational communication. Management Science, 32, 1492-1512.

159

Stelzl, I. (1986). Changing the causal hypothesis without changing the ﬁt: Some rules for
generating equivalent path models. Multivariate Behavioral Research, 21 , 309-331.

Suler, J. (2002, September 2002). Psychology of Cyberspace. Retrieved September 23,
2004, from hm)://www.rider.edu/users/suler/psvcyber

Sunnafrank, M. (1986). Predicted outcome value during initial interactions: A reformation
of uncertainty reduction theory. Human Communication Research, 13, 191-210.

Tanis, M., & Postmes, T. (2003). Social cues and impression formation in CMC. Journal
of Communication.

Taylor, J ., & McDonald, J. (2002). The effects of asynchronous computer-mediated group
interaction on group processes. Social Science Computer Review, 20(3), 260-274.

Thibaut, J. W., & Kelley, H. H. (1952). The Social Psychology of Groups. New York:
John Wiley & Sons.

Tichon, J. G., & Shapiro, M. (2003). The process of sharing social support in cyberspace.
C yberPsychology & Behavior, 6(2), 161-170.

Tidwell, L. C., & Walther, J. B. (2002). Computer-mediated communication effects on
disclosure, impressions, and interpersonal evaluations: Getting to know one
another a bit a time. Human Communication Research, 28(3), 317-348.

Tourangeau, R. (2004). Survey. research and societal change. Annual Review of
Psychology, 55, 775-801.

Turkle, S. (1995). Life on the Screen: Identity in the Age of the Internet. New York:
Simon & Schuster.

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). .
Rediscovering the Social Group: A Self-categorization Theory. Oxford: Blackwell.

Utz, S. (2000). Social information processing in MU Ds: The development of friendships
in virtual worlds. Journal of Online Behavior, 1(1).

160

Valacich, J. 5., Dennis, A. R., & Nunamaker, J. F. (1992). Group size and anonymity
effects on computer-mediated idea generation. Small Group Research, 23 , 49-73.

VanLear, C ., Sheehan, M., Withers, L. A., & Walker, R. A. (2005). AA Online: The
Enactment of Supportive Computer Mediated Communication. Western Journal of
C ommunication, 29(1), 5-26.

Waldzus, S., & Schubert, T. (2000). Group norm and category norm in anonymous
situations: Two sources of social inﬂuence. In T. Postmes, R. Spears, M. Lea & S.
Reicher (Eds), SIDE Issues Centre Stage: Recent Developments in Studies of DE-
Individuation in Group (pp. 31-45). Amsterdam: KNAW.

Wallace, P. (1999). The Psychology of the Internet. Cambridge: Cambridge University
Press.

Walther. J. B. (1992). Interpersonal effects in computer-mediated interaction: A relational
perspective. Communication Research, 19, 52-90.

Walther, J. B. (1994). Anticipated ongoig interaction versus channel effects on relational

communication in computer-mediated interaction. Human Communication
Research, 40, 473-501.

Walther, J. B. (1997). Group and interpersonal effects in international computer-mediated
collaboration. Human Communication Research, 23, 342-369.

Walther, J. B., Slovacek, C., & Tidwell, L. C. (2001). Is a picture worth a thousand
words? Photographic images in long term and short term virtual teams.
Communication Research, 28, 105-134.

Walther, J. B., Anderson, J. F., & Park, D. W. (1994). Interpersonal effects in computer-
mediated interaction: A meta analysis of social and antisocial communication.
Communication Research, 21, 460-487.

Walther, J. B., & Boyd, S. (2002). Attraction to computer-mediated social support. In C.
A. Lin & D. Atkin (Eds), Communication Technology and Society: Audience
Adoption and Uses (pp. 153-188). Cresskill, NJ: Hampton Press.

Walther, J. B., & Burgoon, J. K. (1992). Relational communication in computer-mediated
interaction. Human Communication Research, 19, 50-88.

161

Wayner, P. (1999). Technology for anonymity: Names by other nyms. The Information
Society, 15, 91-97.

Weisband, S., & Atwater, L. (1999). Evaluating self and others in electronic and face-to-
face groups. Journal of Applied Psychology, 84, 632-639.

Weisgerber, C. (2000). Meeting strangers in cyberspace and real-life: A comparison of
initial face-to-face and computer-mediated communication. Paper presented at the
National Communication Association, Seattle, Washington.

Wheeless, L. R., & Grotz, J. (1976). Conceptualization and measurement of reported self-
disclosure. Human Communication Research, 2, 338-346.

Whittaker, S., Terveen, L., Hill, W., & Chemy, L. (1998). The Dynamics of Mass
Interaction. Paper presented at the ACM Conference on C omputer-Supported
Cooperative Work, Seattle, Washington.

Williams, R. (1998). Reflections on anonymity. Perceptual and Motor Skills, 67, 763-766.

Winstead, B. A., Derlega, V. J., & Wong, P. T. P. (1984). Effects of sex-role orientation
on behavioral self-disclosure. Journal of Research in Personality. 18(4), 541-553.

Worthy, M., Gary, A. L., & Kahn, G. M. (1969). Self-disclosure as an exchange process.
Journal of Personality and Social Psychology. I 3, 59-63.

Yun, H., Park, S., & Kim, H. (2003). An Invisible Leverage in the Adoption of Online
Social Support Community. Paper presented at the International Communication
Association, San Diego, CA.

Yun, H., Park, S., Kim, H., Yoo, H., Kwak, Y., & Kang, K. (2004). The Rise and Fall of
Online Social Support: Group Level Predictors. Paper presented at the
International Communication Association, New Orleans, LA.

Zhang, Y., & Hiltz, S. R. (2003). Factors that Influence Online Relationship Development
in a Knowledge Sharing Community. Paper presented at the Americas Conference
on Information Systems, Tampa, FL.

Zimbardo, P. (1969). The human choice: individuation, reason, and order versus
deindividuation, impulse, and chaos. In W. J. Arnold & D. Levine (Eds),

162

Nebraska Symposium on Motivation (pp. 237-307). Lincoln: University of
Nebraska.

163