The Virtual Cutting Edge: The Internet and Adolescent Self-Injury

Developmental Psychology
Copyright 2006 by the American Psychological Association
2006, Vol. 42, No. 3, 407– 417
0012-1649/06/$12.00
DOI: 10.1037/0012-1649.42.3.407
The Virtual Cutting Edge: The Internet and Adolescent Self-Injury
Janis L. Whitlock, Jane L. Powers, and John Eckenrode
Cornell University
The 2 studies reported here use observational data from message boards to investigate how adolescents
solicit and share information related to self-injurious behavior. Study 1 examines the prevalence and
nature of these message boards, their users, and most commonly discussed topics. Study 2 was intended
to explore the correlations between content areas raised for discussion. Both studies were intended to
shed light on the role of message boards in spreading information about self-injurious practices and
influencing help-seeking behavior. More than 400 self-injury message boards were identified. Most are
populated by females who describe themselves as between 12 and 20 years of age. Findings show that
online interactions clearly provide essential social support for otherwise isolated adolescents, but they
may also normalize and encourage self-injurious behavior and add potentially lethal behaviors to the
repertoire of established adolescent self-injurers and those exploring identity options.
Keywords: adolescence, self-injury, Internet, mental health
I think my greatest fear is to be forgotten. A teacher I had last year
of exchanges that occur there, or the ways in which these ex-
doesn’t even remember my name—it makes me think that no one
changes may affect the development of adolescents and their
remembers me. How do I know I exist? At least I know I exist when
ability to cope with distress.
I cut. (Self-injury message board post)
Because adolescents use the Internet for the purpose of connect-
Adolescent Internet Use
ing with others at higher rates than any other age group (Lenhart,
Rainie, & Lewis, 2001), a better understanding of how Internet use
Computer access and use among adolescents have grown expo-
affects their social and emotional development is an important line
nentially over the past decade (Becker, 2000). More than 80% of
of scientific inquiry. Indeed, a small but growing body of research
American youth 12 to 17 years of age use the Internet, and nearly
is beginning to examine the implications of various electronic
half log on daily (Lenhart, Madden, & Hitlin, 2005). Once con-
forums for social interaction (e.g., chat rooms, news groups, mes-
nected, adolescents engage in a wide variety of activities, includ-
sage boards) on adolescent behavior (e.g., Gross, 2004; Subrah-
ing doing schoolwork, playing games, shopping, and downloading
manyam, Greenfield, & Tynes, 2004; Tynes, Reynolds, & Green-
music. Research shows, however, that adolescents use the Internet
field, 2004). The Internet may have particular relevance for
primarily for social reasons (Gross, 2004; Roberts, Foehr, & Ride-
adolescents who feel marginalized, because it provides a low-risk
out, 2005). The Internet has become a virtual meeting place where
venue for finding others who share their perceived or real differ-
teens hang out with their peers to pass time. Many adolescents
ences and exchanging information that is difficult to convey in
reportedly prefer being online to other media, including the tele-
person or when using one’s real identity (McKenna & Green,
phone, TV, and radio (2002 Gallup Survey, cited in Heitner, 2002).
2002). Adolescents who intentionally injure themselves are one
According to data from the Pew Internet and American Life
such group. Although research is nascent, adolescent self-injury
Project (Lenhart et al., 2005), the vast majority (89%) of teens use
appears to be increasingly common (Welsh, 2004; Yates, 2004)
e-mail and 75% use instant messaging (IM), which allows them to
and, as this article shows, is a practice around which many virtual
have multiple simultaneous conversations with a defined group of
communities have formed. To date, almost nothing has been
peers. More than 50% of teens possess more than one e-mail
written about the existence of self-injury Internet forums, the types
address or screen name, which they can use to send private
messages to friends or to participate anonymously in online fo-
rums, such as chat rooms (Lenhart et al., 2001).
Some studies suggest that Internet use may facilitate social
Janis L. Whitlock and John Eckenrode, Family Life Development Cen-
interaction by making it easier for individuals to connect with
ter and Department of Human Development, Cornell University; Jane L.
others they know as well as with strangers. It serves also as a
Powers, Family Life Development Center, Cornell University.
powerful resource for youth desiring information about socially
This study was supported by funding from the Cornell College of
sensitive topics such as sexuality and interpersonal relations (Su-
Human Ecology Seed and Innovation Grant. We thank Amanda Purington,
zuki & Calzo, 2004). This form of communication may be espe-
Alexis Matusiewicz, and the Cornell Self-Injury Study Team for their
cially advantageous for shy, socially anxious, or marginalized
substantial work on data collection and analysis. We also thank Elizabeth
youth, enabling them to practice their social skills without the risks
R. Woods for her thoughtful reading of an earlier version of this article.
associated with “on the ground” interactions (Heitner, 2002; Mc-
Correspondence concerning this article should be addressed to Janis L.
Whitlock, Family Life Development Center, Cornell University, Beebe
Kenna, Green, & Gleason, 2002; Subrahmanyam et al., 2004).
Hall, Ithaca, NY 14853-4401. E-mail: jlw43@cornell.edu
Additionally, online communication may encourage more truthful
407

---

408
WHITLOCK, POWERS, AND ECKENRODE
exchanges; many people report a greater willingness to share
distress through a process of self-disclosure and social comparison,
thoughts and feelings online than they would in face-to-face situ-
and serve as a venue for giving and receiving social support, it may
ations (Lenhart et al., 2001; McKenna & Bargh, 2000). Clearly, the
also provide a positive coping resource for distressed youth. Al-
Internet is transforming the social world of adolescents by influ-
ternatively, Internet use may maintain or increase distress if the
encing how they communicate, establish and maintain relation-
information exchanged reinforces negative views of self or sug-
ships, and find social support. Nevertheless, the developmental
gests destructive or otherwise ineffective coping strategies.
consequences of adolescent Internet use is an area about which
This article examines the role of the Internet among adolescents
little is known (Wartella, Caplovitz, & Lee, 2004).
who use self-injurious behavior (SIB) as a method of coping with
distress. The opportunity to explore different identities and roles
The Internet and Adolescent Social and Emotional
through the Internet may be particularly important for individuals
Development
with stigmatized identities, such as self-injurious youth, who feel
that important aspects of their selves need to be concealed in their
Three central tasks are integral to healthy social development
day-to-day lives. These youth may be especially motivated to
during adolescence: (a) to establish caring, meaningful relation-
participate in electronic forums that allow them to express them-
ships; (b) to find acceptance and belonging in social groups; and
selves in a safe and anonymous environment and to find support
(c) to establish interpersonal intimacy (Baumeister & Leary, 1995;
from others who share their sense of marginalization and under-
Reis & Shaver, 1988; Sullivan, 1953). Peers play a crucial role in
stand their behavior. To date, however, not only is very little
this process, because a positive relationship with peers is important
known about self-injury in the adolescent population, but nothing
for psychological well-being and social adjustment (Bishop &
is known about how this group uses the Internet to connect with
Inderbitzen, 1995; Hartup, 1996), whereas peer rejection is linked
others about their practice.
to serious problems, including delinquency, drug abuse, and de-
pression (Hartup, 1996; Merten, 1996).
Adolescent Self-Injury as a Developmental Phenomenon
Early studies of the influence of online interactions on adoles-
cent development suggested that high levels of Internet use may
The increasing number of stories in the mainstream press and
inhibit healthy social development by linking frequent use to social
popular media, as well as the growing number of anecdotal reports
isolation and depression, especially among teenagers (Kraut et al.,
by physicians, therapists, and school counselors, suggest that self-
1998; Nie & Erbring, 2000). However, these findings have been
injury may be “the next teen disorder” (Welsh, 2004). Although
disputed. A follow-up study conducted by Kraut et al. (2002)
operationally elusive, scholars differentiate self-injury from cul-
found that the effects documented in their earlier study had largely
turally sanctioned forms of self-mutilation, such as piercing or
dissipated. They did find, however, that effects of Internet use on
tattooing, by intention rather than form. Alternatively called de-
depression differed for introverts and extroverts: Extroverts were
liberate self-harm, self-injury, self-mutilation, or cutting, self-
more likely to feel greater social connection as a result of Internet
injurious behavior typically refers to a variety of behaviors in
use, whereas introverts became more depressed and withdrawn.
which an individual purposefully inflicts harm to his or her body
Heitner (2002) found that adolescents who use the Internet to
for purposes not socially recognized or sanctioned and without the
connect with others in real-time social exchanges tended to possess
obvious intention of committing suicide (Alderman, 1997;
higher peer status, more social skills, and greater social integration
Favazza, 1996). Self-injury, which is most often associated with
than their more socially introverted and withdrawn peers, who
the term cutting, also includes intentional carving or cutting of the
spent most of their Internet time in solitary activities. Additionally,
skin and subdermal tissue, scratching, burning, ripping or pulling
adolescents who used chat rooms exhibited lower peer status and
skin or hair, swallowing toxic substances, bruising, and breaking
had fewer social skills than those who did not. Similarly, Gross,
bones. Although not typically a suicidal gesture, self-injury is
Juvonen, and Gable (2002) found that teenagers with strong social
statistically associated with suicide and can result in unanticipated
connections use e-mail and IM to reinforce preexisting bonds,
severe harm or fatality (Claes, Vandereycken, & Vertommen,
whereas those with less developed social networks use the anon-
2003; Favazza, DeRosear, & Conterio, 1989).
ymous features of the Internet to find new friends and social
There are currently no reliable estimates of the prevalence of
outlets, perhaps compensating for what they lack offline. This
self-injury among the general U.S. adolescent population. The vast
suggests that chat rooms and similar venues in which adolescents
majority of research on self-injury has been conducted in clinical
share experiences anonymously may provide a safe forum for less
populations or using small, unrepresentative community samples.
socially adept adolescents to practice social interaction.
These studies generally find that cutting and other forms of self-
Research finds that online exchange decreases social isolation
injury are evident in approximately 20% of the clinical population
among adolescents and helps them connect with people and ex-
(Deiter, Nicholls, & Pearlman, 2000) and are linked to high levels
plore their identity (Maczewski, 2002; Suzuki & Calzo, 2004).
of pathology (Brodsky, Cloitre, & Dulit, 1995; Ross & Heath,
This helps to explain how the Internet may serve as a virtual peer
2003). The few studies that have been conducted in U.S. commu-
support group where adolescents under stress can express feelings
nity samples of young adults and adolescents are limited by small
and exchange information about modes of coping. Adolescence is
convenience-based samples and vary in estimates of self-injury
also a time of increased feelings of distress (e.g., depressed mood)
prevalence from 4% to 38% (Briere & Gil, 1998; Favazza, 1992;
and increased access to modes of coping with stress that are
Gratz, Conrad, & Roemer, 2002; Kokaliari, 2005; Muehlenkamp
independent of parents (Arnett, 1999; Compas, 1987; Petersen,
& Gutierrez, 2004). Large studies in Britain estimate that approx-
Kennedy, & Sullivan, 1991). To the extent that Internet use can
imately 10% of youth 11 to 25 years of age self-injure. A British
reduce feelings of social isolation, help normalize feelings of
report on the national scope of the problem documents a dramatic

---

SPECIAL SECTION: VIRTUAL CUTTING EDGE
409
increase in disclosures of self-injury to national children’s help
ever, see self-injury as a manifestation of mental or emotional
lines over the 5 years before the study, noting a 65% increase in the
disorders or heterotypic manifestation of childhood trauma (Con-
last 2 years (Young People and Self Harm: A National Inquiry,
ners, 2000; Conterio & Lader, 1998; Favaro & Santonastaso, 2002;
2004).
Strong, 1998; Yates, 2004).
The reasons for this apparent increase are unclear. Although
Whether a trauma-linked developmental model will hold true
general awareness of self-injury may lead to increased willingness
for nonclinical populations is unclear. However, the proximal
to disclose the behavior, it may also reflect a true increase in
causes for those who engage in self-injurious behavior in both
incidence. Social contagion has been identified by some (Yates,
clinical and community settings are strongly linked to difficulties
2004; Rosen & Walsh, 1989), because self-injury follows
in regulating strong emotions and coping with stress (Favazza,
epidemic-like patterns in institutional settings such as hospitals
1996; Rosen & Walsh, 1989; Ross & Heath, 2002; Tantam &
and detention facilities (Matthews, 1968; Ross & McKay, 1979;
Whittaker, 1993). The cumulative impact of stressful life events
Taiminen, Kallio-Soukainen, Nokso-Koivisto, Kaljonen, & Hele-
that trigger self-injurious episodes, the subsequent shame associ-
nius, 1998). The possibility that self-injury is communicable may
ated with the self-injurious response, and the secrecy surrounding
reflect a pattern similar to what Brumberg (1992) argued with
the behavior create fertile ground for the development of margin-
regard to the spread of anorexia nervosa in the 1980s, when
alized perceptions of the self. The Internet provides self-injurers
heightened cultural visibility through the mass media rendered
with an anonymous venue for sharing actual and fabricated aspects
anorexia nervosa an available emotional outlet for individuals with
of their true identity among a community of similar others. Be-
receptive predispositions.
cause young people with depressive symptomatology are much
Although no single self-injurer profile has emerged, there is
more likely to talk with strangers online and to disclose personal
general consensus that self-injury is most common among adoles-
information than those without depressive symptomatology (Ybarra,
cents. Self-injurious behavior may parallel other problem behav-
Alexander, & Mitchell, 2005), self-injurers may be particularly
iors, which begin during early adolescence, peak during mid- to
receptive to using the Internet to locate virtual communities. Al-
late adolescence, and then decline in adulthood (Briere & Gil,
though such virtual communities may provide a much-needed
1998; Favazza, 1999; Favazza & Conterio, 1989). However, there
source of support and cathartic sharing, the possibility that self-
is some evidence that self-injurious behavior may follow two
injury is communicable suggests that the Internet may also serve to
distinct patterns similar to the life course-persistent and
spread or deepen the practice in adolescent populations.
adolescence-limited trajectories evident in antisocial behavior
(Moffit, 1993). One of these begins in early childhood and persists
Aim of the Present Studies
through adulthood. The other follows a typical adolescence-limited
course that emerges in early adolescence and declines in late
Two studies were designed to explore and document adolescent
adolescence or early adulthood (Dubo, Zanarini, Lewis, & Wil-
use of online message boards to share, solicit, and receive infor-
liams, 1997; Nixon, Cloutier, & Aggarwal, 2002). Although some
mation and advice related to self-injurious behavior. Message
research finds females more likely to injure themselves than males
boards are an electronic venue in which individuals (i.e., “posters”)
(Conterio & Lader, 1998; Favazza, 1999), other studies suggest
register with the site under a chosen name and are allowed to post
that the gender gap may be narrower (Briere & Gil, 1998; Deiter
thoughts, ask questions, and respond to other posters. Unlike
et al., 2000; Dulit, Fyer, Leon, Brodsky, & Frances, 1994; Galley,
e-mail and IM, which permit private exchange between selected
2003; Martin, Rozanes, Pearce, & Allison, 1995). The difficulty in
individuals, message boards and chat rooms (in which exchanges
accurately assessing gender differences in self-injurious behavior
occur in real time) are entirely public: All postings are available to
may arise from variation in how male and female self-injurers are
all members and, in many cases, nonmembers as well, although
identified, how they injure themselves, and whether or not they
nonmembers cannot post material. When individuals log on to a
seek treatment (Alderman, 1997; Connors, 2000).
message board, the content varies but typically includes informa-
The relationship between self-injury and suicide is important but
tion about the message board purpose and rules as well as direct
not clearly understood. Persons who engage in self-injury are more
links to “threads,” posts that have been made by users sorted by
likely to consider or attempt suicide (Walsh & Rosen, 1988;
subject title. Anyone who logs on to a message board may read
Gardner & Cowdry, 1985; Hawton, Fagg, Simkin, Bale, & Bond,
posts, but only those registered may make a post. Blogs, which are
2000). Nevertheless, in the vast majority of cases self-injury is
becoming increasingly common, are essentially public electronic
used to alleviate distress temporarily rather than to signal the
diaries. Message board posters frequently link their blog site to
intention to end one’s life (Favazza, 1996; Rosen & Walsh, 1989;
their membership identity or the signature line in their posts.
Tantam & Whittaker, 1993). Indeed, some scholars see it as a
The goal of the first study was to investigate the prevalence and
highly functional alternative to suicide (Alderman, 1997; Strong,
nature of self-injury message boards and their users with the
1998). Among clinical populations, self-injury is comorbid with
intention of better understanding the general subject categories
borderline personality disorder, eating disorders, posttraumatic
most commonly raised for discussion. The goal of the second study
stress disorder, depression, anxiety disorders, and a history of
was to explore the correlations between content areas raised for
abuse or trauma (Alderman, 1997; Connors, 2000; Conterio &
discussion. Both studies were intended to shed light on the role of
Lader, 1998; Holmes & Nadelson, 2000; Sansone & Levitt, 2002;
message boards in spreading information about self-injurious prac-
Yates, 2004). Indeed, some researchers have called for a new
tices and influencing help-seeking behavior. Studies of the Internet
Diagnostic and Statistical Manual of Mental Disorders impulse-
are effectively studies of culture. Participants are free to construct
control disorder, deliberate self-harm syndrome, which would
their age and identity as they wish, and studies show that teen
include self-injurious behavior (Favazza et al., 1989). Most, how-
Internet users often construct themselves as somewhat older than

---

410
WHITLOCK, POWERS, AND ECKENRODE
they actually are (Gross, 2004). Both studies were approved by the
description of the sites; names are withheld to protect confidentiality). A
University Internal Review Board and have abided by the stipu-
total of 3,219 individual posts, examined over a 2-month period, were
lation that participant quotes would be paraphrased rather than
coded for themes of interest and used in the content analysis that follows.
exact quotes. All quotes used here comply with this requirement.
In addition to individual-level posts, we were also interested in gathering
Because the studies use observational methods, active consent was
descriptive information about message board characteristics. Although
most message boards do not contain information about date of inception,
not required.
cross-listing with other topics, and membership characteristics, one of the
Internet service providers did permit access to this information. The iden-
Study 1: Prevalence and Nature of Self-Injury Message
tity of this search engine is hidden to protect confidentiality. All 140
Board Use
message boards identified through this engine were examined for dates of
inception, number of members, and cross-listing with other conditions.
The first study was undertaken to describe: (a) the prevalence of
Measures for message board analysis.
Depending on the analysis, the
self-injury message boards, (b) the number of self-injury message
following information was obtained for each message board: date of
boards advertised jointly with message boards for another disorder
establishment, mean self-reported user age, number of active and inactive
or behavior (such as eating or bipolar disorders), (c) the ratio of
members, percentage of posters with blogs, and colisting with message
active and inactive posters, (d) the age range of users, as revealed
boards for other behaviors. Date of establishment was determined using
in their self-descriptions, (e) date of appearance on the Web, (f) the
information from the home page or, if this was absent, from the date of the
relationship between using message boards and blogs, and (g) the
first post. The mean range of self-described age was determined by
nature of and variation in the content of postings on active message
selecting 50 members at random from the site introduction board, where
boards.
basic member profiles are stored. The number of active and inactive posters
was determined by comparing the number of members who have registered
but never posted with that of members who have posted. Moderation level
Method
was determined by identifying whether there were any instructions to users
Identification and selection.
For this study, there were two units of
about allowable and unallowable posts and whether potentially triggering
analysis: the message board and individual posts in a selected set of
posts (i.e., events that provoke self-injurious behavior) were labeled as
message boards. First, to identify the prevalence of self-injury message
such. Cross-listing with other topics refers to the extent to which a message
boards, five Internet search engines were used: Yahoo, Google, MSN,
board was advertised as a virtual meeting place for individuals interested in
AOL, and Gurl.com. Terms searched included self-injury, self-harm, self-
or affected by the intersection of two or more issues (e.g., self-injury and
mutilation, and cutting. More than 400 self-injury message boards were
suicide). Cross-listing was measured by examining the formal message
identified using this method; of these, 10 were selected for in-depth content
board title and description of each board. When a cross-listed board was
analysis. These boards represented the first 10 listed in each search engine,
identified, the cross-listed topic was noted and the message board was
which are generally those Web sites with the most activity. The moderation
considered as a cross-listed board.
level of each of these boards varied as well. Moderation level refers to the
Measures for content analysis of postings.
The content analysis of
degree to which posters are actively monitored for potentially damaging
posts was conducted using a set of binary (present–absent) codes. These
content (such as sharing techniques for self-injury). For example, when a
codes were drawn from a synthesis of the research cited previously, from
particularly graphic or triggering post is made to a high- or medium-level
interviews conducted with 15 self-injurers by Janis L. Whitlock, and from
moderation board, it is blocked altogether (high moderation) or labeled as
pilot message board observations that spanned a 2-week period. A team of
a potential trigger (medium moderation). We were most interested in
three coders systematically reviewed posts and used the constant compar-
message boards with medium and low moderation, because content would
ative method (Glaser & Strauss, 1967) to inductively monitor themes. Once
be less likely to be censored. Detailed membership and moderation level
the initial set of observations had been coded, thematically grouped clusters
information for the 10 sites selected was recorded (see Table 1 for a
were identified (e.g., motivation). The 11 broad categories identified en-
Table 1
Membership Breakdown of Selected Self-Injury Message Boards
Age (years)
Members w/posts (%)
Site
% Female
No. members
% w/Blogs
Mod levela
Range
M
Never
2–10
100
A
12–44
18.7
88.1
844
30
41.0
20.0
9.5
Low
B
13–54
18.3
90.0
5,259
25
57.0
15.5
7.2
Med
C
14–47
19.4
88.0
5,082
15
14.0
44.1
17.3
High
D
12–37
17.6
74.0
NA
25
31.5
32.8
2.0
Low
E
14–22
17.5
91.5
70
15
45.7
41.5
0
Low
F
14–36
19.6
87.5
6,656
10
63.4
10.9
8.5
Med
G
16–47
20.5
90.0
153
20
65.0
9.8
9.8
Low
H
14–28
18.1
80.0
4,021
30
41.3
15.9
21.0
Med
I
15–46
23.9
80.6
1,427
15
30.4
27.8
9.5
Med
J
13–26
16.4
78.0
2,581
10
50.8
25.0
10.0
Mod
Note.
Mod
moderation; NA
not available; Med
medium.
a Moderation level was determined by the extent to which the message board moderator blocked or labeled
potentially triggering posts. In high-moderation boards, potentially triggering or disruptive posts were blocked
entirely. In medium-moderated message boards, triggering or otherwise disruptive posts were identified and
labeled. Low-moderation message boards took no steps to identify or block posts of any sort.

---

SPECIAL SECTION: VIRTUAL CUTTING EDGE
411
Table 2
cause public familiarity and dedicated server space with message
Number of Self-Injury Message Boards From One Internet
boards have also increased over time, these findings indicate that
Service Established by Year
message boards as a cultural tool for self-injury grew dramatically
between 1998 and 2000 and that interest in both establishing and
Year
No. boards
Total membership
participating in self-injury specific groups over the past 5 years has
been sustained.
1998
1
93
1999
7
949
Examination of these same message boards for colisting with
2000
26
2,831
other topics reveals links with a number of topics known to be
2001
25
703
associated with self-injury in the literature. Table 3 shows the
2002
28
1,611
extent to which self-injury message boards were cross-listed with
2003
19
952
other topics. Although listed most often alone (56%), 44% of the
2004
24
806
2005
38
1,698
message boards were cross-listed with one or more other topics.
Total
168
9,643
When linked to another topic, self-injury occurs most often with
depression (32%) and eating disorders (17%). Although less fre-
quent, it often occurs with message boards dedicated to discussions
of dissociative identity and multiple personality disorders (10%);
compassed virtually all message board exchanges and served, along with
bipolar disorder (9.2%); sexual abuse (7%); obsessive– compulsive
all specific themes they encompassed, as a coding tool (e.g., loneliness,
disorders (7%); addiction (2.8%); anxiety disorder (2.8%); lesbian,
anger, and dysphoria were all themes in the motivation category). Posts left
gay, bisexual, and transgender issues (2.8%); and autism (1.4%).
uncoded were those with little capacity to illuminate understanding of
self-injurious behavior, such as idle discussion about a current event or
Message board membership analysis.
The 10 most popular
activity. The resulting 11 areas were as follows: (a) informal support and
message boards were selected for membership analysis. Table 1
exchange, (b) motivation for self-injury, (c) concealment of self-injurious
shows the averages for self-description: member age, age range,
behavior (e.g., anxiety about exposure, methods for concealment of cuts
and percentage of females. It also shows the percentage of mem-
and scars), (d) addiction language (e.g., days self-injury free, difficulty
bers with public blogs, the percentage of registered members who
stopping), (e) formal help seeking and treatment, (f) sharing techniques, (g)
have never posted (potential viewers), percentage of posters with
links to other mental health or behavioral conditions known to be associ-
more than 100 posts, and moderation level. The average self-
ated with self-injurious behavior, (h) references to popular culture, (i)
described age of members ranged from 16.4 to 23.9 years for each
perceptions of non-self-injurers reactions to self-injurious behavior, (j)
message board, although there was large variation in ages repre-
perception of self and behavior (e.g., self-worth, lovability, dissociation),
sented and the mean tends to be negatively skewed. The majority
and (k) venting or apologizing. These areas contained a total of 70 themes
into which nearly all posts could be categorized.
of message boards had a mean stated age of 18 years; 80% of the
Coding of posts.
A content analysis was conducted on all original and
members described themselves as being between the ages of 14
follow-up responses to the original post in 10 message boards over a
and 20; 31% of all members described themselves as being 15 or
2-month period. A time frame of 2 months was selected to ensure adequate
16 years old. In all 10 boards examined, posters describing them-
breadth in content areas. A total of 3,219 posts were examined during this
selves as female were more likely to be registered and to partici-
period. Both original posts and responses to posted messages were coded.
pate actively. The number of members in each message board
Because not all posts contained content relating to the coding scheme, the
varied dramatically from 70 to 6,656. Membership data from one
total number of posts examined exceeds the total number of posts to which
of the sites were not available because members post in any of a
a code was assigned (2,942). (The criterion for leaving postings uncoded is
wide number of forums, not all of which relate to self-injury.
described later). However, if posts contained more than one thematic
reference, they were assigned multiple codes. Therefore, totals in the
Results section can add up to more than 2,942.
Table 3
Four coders were trained by Janis L. Whitlock and the study coordinator
Conditions and Behaviors With Which Self-Injury Message
to observe and code message boards during this time period. This was
Boards Are Cross-Listed
accomplished by having each coder and the principal investigator code the
same posts and discuss code assignment. Once training was completed,
Frequency
% of
each coder logged on one to two times per week during the study period to
Cross-listed condition/behavior
(n
140)
Total
identify and code new posts. To establish intercoder reliability, each coder
team was given three randomly selected threads to code. Codes for each
Depression
45
32.0
individual coder were then compared to establish agreement within each
Eating disorders
24
17.0
team. The teams were in full agreement about codes for 96% and 97% of
Dissociative identity disorder/multiple
the postings, respectively.
personalities
14
10.0
Bipolar
13
9.2
Obsessive-compulsive disorder
10
7.1
Results
Sexual abuse
10
7.1
Posttraumatic stress disorder
7
5.0
Message board prevalence, dates of inception, and colisting
Anxiety
4
2.9
with other topics.
The search procedure described previously
LGBTQ
2
1.4
revealed 406 boards. As shown in Table 2, examination of the 140
Mood disorders
4
2.8
boards available through the Internet service providing historical
Addiction
4
2.8
Autism
2
1.4
information shows an increase from 1998 to 2000 and then a
No other disorder specified
78
56.0
generally stable trend over time, with peaks in 2000 and in 2002.
Although caution is warranted when interpreting these data, be-
Note.
LGBTQ
lesbian, gay, bisexual, transgender, and questioning.

---

412
WHITLOCK, POWERS, AND ECKENRODE
The number of message board members with blogs ranged from
Table 5
10% to 30%. Preliminary examination of blog content revealed
Breakdown of Most Common Themesa
few substantive differences from message board posts other than
the fact that blogs often provided more detailed accounts of events
Percentage of
Total no. of
primary
and feelings. There was considerable variation in the extent to
Category/themes
occurrences
category
which members posted. In four of the boards, more than half of the
membership had never posted a single comment. The majority of
Triggers
those who did participate posted 2 to 10 messages, although each
Conflict with important others
212
34.8
Perceived depression
151
24.8
message board had a group of high posters with more than 100
School or work stress
63
10.3
posts.
Loneliness
44
7.2
One notable characteristic of message board membership was
Sexual abuse/rape
22
3.6
that virtually all members have images associated with their online
Other
117
19.2
identity. Characters, such as “Emily the Strange,” digital images
Concealment issues
Anxiety about concealment
149
51.8
made by posters of scars and bleeding wounds, and other graphic
Managing scars
109
37.3
and bloody icons were frequently included with posted text. Also,
Acknowledgment of dishonesty
29
10.9
lines from songs, poetry, or books were included in signature lines
Addictive elements
and were often expressions of sentiment. One example is the lyric
“I’ve been cutting free for . . .”
120
41.7
Stable pattern/cannot control urge
52
18.1
by the music artist Eminem: “Sometimes I even cut myself to see
Minimize problem
28
9.7
how much it bleeds. It’s like adrenaline, the pain is such a sudden
Liken to other drugs
27
9.3
rush for me.”
Multiple attempts to quit
27
9.4
Message board content.
A total of 3,219 posts were examined
Increased tolerance, need more over time
24
8.4
from six message boards over a 2-month period; 2,942 were
Relapse
10
3.4
Help seeking and treatment
assigned a code. Although posts could be assigned multiple codes,
Positive attitudes toward therapy
101
44.1
most were not. Tables 4 and 5 list the total number of times a post
In therapy
60
26.2
was assigned a particular content code and the proportion of all
Negative attitude toward therapy
43
18.7
posts examined that fell into this content area. As shown in Table
References to medications
25
10.9
4, the vast majority of all posts examined fell into 1 of 11 broad
a Provision of informal support, although the dominant theme, is not
thematic categories. Several of the most common categories con-
included in this table because specific categories of support were not
tained a variety of subthemes (see Table 5).
tracked.
Providing informal support to other posters was the most com-
mon type of exchange on the message boards, occurring in 28.3%
of all posts (see Table 4). Comments such as “We’re glad you’ve
trigger-related posts) followed by depression (24.8%) and stress
come here” and “Just relax and try to breathe deeply and slowly”
(10.3%; see Table 5).
were very common. Although less frequent, exchanges around
The next largest thematic category discussed involved conceal-
support were often linked with apologizing behavior (“I’m so sorry
ment of the practice and its effects (primarily scarring), accounting
to lay this on you”) and confessions of self-loathing (“I hate myself
for 9.1% of all posts examined (see Table 4). Comments focused
for doing this”). Discussion of events that triggered a self-injurious
largely on anxiety about being discovered, how to manage scars,
episode occurred in 19.5% of the posts (see Table 4). Conflict with
and the extent to which posters had to be dishonest to maintain
important others constituted the primary trigger (34.8% of all
secrecy (see Table 5).
Posts about the perceived addictiveness of self-injury were
almost as common (8.9%) (see Table 4). Typical examples include
Table 4
“It just haunts me and I don’t think I’ll ever get away from it” and
Dominant Thematic Categories in Message Board Content
“I may try and quit but even if I succeed, I’ll always dream of
razorblades and blood.” Observations starting with the phrase
% Posts
No. of category
examined
“I’ve been cutting free for [length of time]” accounted for almost
Category
occurrences
(N
3,219)a
half of posts coded with addiction elements. References to stable
patterns of self-injury, tendency to minimize the problem, similar-
Informal provision of support for others
913
28.3
ities to other drugs, multiple quit attempts, the need to self-injure
Motivation/triggers
629
19.5
more or more deeply because of increased tolerance for effects of
Concealment issues
292
9.1
Addiction elements
288
8.9
self-injury, and relapse after quitting were all areas discussed
Formal help seeking, treatment
229
7.1
within this category (see Table 5). Often this discussion co-
Requesting, sharing techniques
200
6.2
occurred with comments regarding frustration at parents, caregiv-
Link to other mental health conditions
153
4.7
ers, and others, who, according to the posters, do not appreciate the
References to popular culture
137
4.2
Interpretation of other’s perceptions
85
2.6
addictive nature of the behavior.
Perceptions of self
70
2.1
Discussion of formal help seeking from physical or mental
Venting or apologizing behavior
61
2.9
health professionals occurred in 7.1% of all posts (see Table 4).
Uncategorized
277
8.6
Attitudes toward and experiences with treatment were largely
a
positive (44.1% of all help-seeking posts reflected positive atti-
Individual posts could be assigned multiple codes; total percentages will
not equal 100%.
tudes toward therapy), and discussion of experiences related to

---

SPECIAL SECTION: VIRTUAL CUTTING EDGE
413
therapy accounted for a sizable portion of formal help-seeking
Because the extent to which posters sought and shared self-injury
posts as well (26.2%). Negative attitudes and active discourage-
techniques was of interest, low-moderation boards were selected as
ment from seeking therapy were also evident in 18.7% of the
recruitment sites to ensure that the content of message posts would not
formal help-seeking posts. References to specific medications oc-
be blocked. One medium-moderation board was selected to permit a
balanced sample by age, because the low-moderation boards did not
curred in about 1 in 10 cases.
contain an adequate number of active posters for which age was
The sixth most common category of discussion for all age
available. Each site allowed for all posts from individual posters to be
groups related to sharing details about techniques use to self-
searched readily.
injury, accounting for 6.2% of all posts. These exchanges were
Individual posters were identified by looking through threads and iden-
generally either descriptions of specific self-injury techniques or
tifying individuals in the age brackets of interest with more than 50 posts.
requests for specific technique information. The following conver-
Then 20 in each of three self-described age groups (13–15, 16 –18, 19 –22
sation exemplifies this type of exchange:
years) were selected at random, for a total sample size of 60. Once
individual posters were identified, 50 posts from each individual were
Poster 1: Does anyone know how to cut deep without having it sting
selected at random from posts made from July 2004 to January 2005. A
and bleed too much?
6-month window was chosen to ensure that each poster monitored would
have adequate time to make multiple posts and engage in a variety of
Poster 2: I use box cutter blades. You have to pull the skin really tight
interactions.
and press the blade down really hard. You can also use a tourniquet
Measures.
As in Study 1, the content analysis of posts was conducted
to make it bleed more.
using a set of binary (present–absent) codes. These codes were similar to
Poster 3: I’ve found that if you press your blade against the skin at the
those used in Study 1 but were focused on six specific types of exchange:
depth you want the cut to be and draw the blade really fast it doesn’t
(a) soliciting and sharing techniques, (b) attitudes toward and disclosed use
hurt and there is blood galore. Be careful, though, ‘cause you can go
of formal support, (c) solicitation and provision of informal support, (d)
very deep without meaning to.
disclosure to nontherapeutic others (e.g., family and friends), (e) disclosure
of other mental health conditions, and (f) disclosed self-concept. A total of
Poster 1: Okay, I’ll get a Stanley blade ‘cause I hear that it will cut
17 measures were used to indicate each of these broad conceptual domains
right to the bone with no hassle. But I’ll be careful if I do use a
(Table 6). Although many of these measures are similar to those used in
tourniquet and I won’t cut that deep.
Study 1, Study 2 measures were designed to focus on a specific type of
exchange and to permit individual-level analysis of correlations between
Mental health conditions empirically linked to self-injurious
variables. Self-represented demographics, including gender, age, and total
behavior, such as depression, eating disorders, suicidality, and
number of posts, were also recorded.
sexual abuse, were cited as either linked with or as a trigger for
Coding of posts.
A total of 3,000 individual posts were coded (50 per
self-injury in 4.7% of all posts (see Table 4). References to pop
each 60 individuals). The actual number of posts per individual ranged
culture (e.g., music, movies, books, celebrities, and characters with
from 60 to several hundred. Because coders could query for all threads
posted by individual posters, they randomly selected 50 by identifying the
special significance to posters) appeared in just over 4% of the
total number of pages and threads any one poster made and dividing by 50.
posts. Discussion and interpretation of the perceptions of others
The resulting number was used to systematically identify threads included
came up in 2.6% of the posts. Recognition of the pain their
in the analysis. For example, if a poster made 100 posts, coders coded
behavior caused or might cause others, in conjunction with diffi-
every two posts. If the resulting interval inadvertently identified a redun-
culty stopping a behavior once a pattern was formed, was often
dant thread (one already coded but that surfaced later), the next thread was
linked to the need for secrecy and feelings of shame. Similarly,
used. Posts were not coded when they did not contain content relevant to
perceptions of self, usually negative, were shared in just over 2%
the study objectives. Six coders were assigned individual posters to track
of all posts. Remaining posts were rants, aimless venting, usually
and code. To establish intercoder reliability, 40% of the observations
frustration or anger, or apologies for sharing.
(1,200 posts) were independently coded by three pairs of coders. Each pair
of coders coded 400 posts in common. No two pairs coded the same posts.
Agreement was assessed by calculating the proportion of posts each
Study 2: Variations in the Use of Message Boards Among
individual in the pair coded the same. The average intercoder agreement
across all three pairs was 93% (range
90 –96%).
Self-Injurers
This study was intended to explore the correlations between
Results
content areas raised for discussion. It focuses on exchange in four
broad areas: (a) help seeking and disclosure, (b) technique sharing,
To assess the relationship between soliciting and sharing self-
(c) comorbidity, and (d) attitudes toward self and other posters.
injurious practices and informal and formal help seeking, Spear-
These areas were selected to explore the role of message boards
man correlations were conducted between all measures. Results
exchange in spreading information about self-injurious practices
are shown in Table 6.
and influencing help-seeking behavior.
The analysis revealed two trends in the correlations: one in
which more positive exchanges were correlated and one in which
negative exchanges were correlated. For example, offering infor-
Method
mal support was correlated with disclosing that someone knows,
suggesting formal treatment, seeking advice on stopping, and
Identification and sampling.
The unit of analysis for this study was
the individual self-injury message board poster. Individual posters were
disclosing a history of trauma. Similarly, seeking advice on stop-
selected from five low- and one medium-moderation self-injury mes-
ping and harm reduction were correlated with each other and with
sage boards included in the previous study (A, B, D, E, F, and G in
seeking advice about disclosure and sharing positive remarks
Table 1).
about oneself. Encouraging formal treatment of self-injurious be-

---

414
WHITLOCK, POWERS, AND ECKENRODE
7
1
—
6
1
—
.098
5
1
—
.269*
.241
4
1
—
.261*
.103
.143
3
1
—
.234
.139
.029
.102
2
1
—
.005
.142
.171
.154
.193
1
1
—
.299*
.013
.048
.070
.144
.066
0
1
—
.169
.190
.024
.125
.008
.182
.100
—
.108
.199
.457**
.157
.478**
.092
.132
.003
—
.073
.239
.197
.167
.106
.120
.059
.301*
.237
—
.134
.088
.334**
.125
.087
.082
.058
.184
.199
.293*
status
support
attitude
support
conditions
6789
—
Techniques
.163
.160
.002
.128
.056
.031
.099
.082
.319*
.237
.114
Informal
Formal
Disclosure
Disclosure
posts)
Associated
5
—
.281*
.009
.243
.114
.156
.004
.163
.008
.096
.107
.207
.031
3,000
4
—
.319*
.228
.112
.314*
.150
.129
.191
.149
.022
.108
.061
.291*
.198
individuals;
3
—
.299*
.108
.006
.187
.126
.082
.529**
.289*
.145
.286*
.032
.068
.127
.144
60
(N
2
—
.204
.088
.085
.218
.048
.009
.145
.228
.079
.176
.105
.120
.133
.087
.010
1
—
Variables
.231
.075
.098
.042
.084
.103
.332**
.111
.025
.315*
.110
.055
.164
.185
.183
.029
Study
All
self
others
for
harm
self
knows
about
about
disclosure
about
.01.
support
stopping
reducing
treatment
someone
Measure
on
on
about
disclosure
comments
comments
p
Correlations
disclosure
trauma
disorder
techniques
treatment
formal
that
of
comments
**
techniques
informal
advice
advice
advice
6
.05.
Share
Request
Offer
Seek
Seek
Seek
Encourage
Discourage
Disclose
Suggest
Disclose
Suicidality
History
Diagnosed
Disparaging
Positive
Disparaging
p
Table
Spearman’s
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
*

---

SPECIAL SECTION: VIRTUAL CUTTING EDGE
415
havior, offering informal support, and encouraging disclosure to
tance and belonging in social groups, and to establish interpersonal
others were interrelated as well.
intimacy (Reis & Shaver, 1988; Sullivan, 1953). These develop-
A somewhat more negative cluster of exchange linked discus-
mental tasks can be especially difficult for young people struggling
sion of technique sharing with negative attitudes toward disclo-
with intense shame, isolation, and distress, particularly when the
sure. However, discouraging disclosure was also associated both
source and outcome of these feelings must be kept hidden. The
with positive views of self and with seeking advice on stopping. In
assurance of online anonymity may contribute to identity construc-
addition, disclosing that one was in treatment was correlated with
tion by providing opportunities for adolescents with marginalized
disclosing another diagnosed disorder and suicidality. Individuals
or nonmainstream proclivities to experiment with different social
who disclosed a history of trauma were likely to offer informal
roles and selves (McKenna & Bargh, 2000; Turkle, 1995).
support, and those who disclosed suicide-related behaviors were
The less positive side of our findings suggests that participation
more likely than others to disclose that someone knows about their
in self-injury message boards may also expose vulnerable adoles-
self-injurious behavior.
cents to a subculture in which self-injury is normalized and en-
couraged. For example, issues related to concealment of self-
Discussion
injurious behavior, identified as a dominant theme here, may make
self-injury message boards particularly potent agents of self-injury,
Our findings confirm that Internet message boards provide a
because sharing techniques and motives can take place anony-
powerful vehicle for bringing together self-injurious adolescents.
mously. In light of evidence here and elsewhere that self-injurious
Although the message boards examined for these studies may not
behavior may possess addictive qualities (see Yates, 2004, for
be representative of all self-injury message boards, they do provide
review), the adolescent drive to belong and the satisfaction that
a snapshot of content and exchange common in those with high
comes with associating with a community of similar others may
activity. We found that in the last 5 years hundreds of message
inadvertently feed a fundamentally self-destructive behavior for
boards specifically designed to provide a safe forum for self-
some participants. The correlation documented here between shar-
injurious individuals have come into existence. Many of these are
ing injurious techniques and discouraging disclosure lends support
populated by individuals who identify themselves as females be-
to this possibility.
tween 14 and 20 years of age. Although the strong preponderance
Although not impossible for individual self-injurers to have
of females may not accurately reflect the gender breakdown of
gathered before the advent of the Internet, easy access to a virtual
self-injurious behavior in the general population (Whitlock, Eck-
subculture of like-minded others may reinforce the behavior for a
enrode, & Silverman, in press), it may reflect the tendency for
much larger number of youth. The tendency for self-injurious
females to solicit more informal and formal help and social support
behavior to follow epidemic-like patterns in institutional settings
compared with males (Fuhrer, Stansfield, Chemali, & Shipley,
such as hospitals and detention facilities (Matthews, 1968; Ross &
1999; Saunders, Resnick, Hoberman, & Blum, 1994).
McKay, 1979; Rosen & Walsh, 1989; Taiminen, Kallio-
Once online, message board members are able to post or pas-
Soukainen, Nokso-Koivisto, Kaljonen, & Helenius, 1998) suggests
sively observe a wide variety of anonymous exchanges. Just less
that the behavior may be socially contagious in other settings and,
than half of all message boards we investigated were cross-listed
therefore, through the Internet as well. As Brumberg (1992) has
with conditions known to be comorbid with self-injurious behav-
argued for eating disorders, discussion of techniques and the
ior, such as depression, eating disorders, and suicide. Of all types
perceived benefits of self-injury may add potentially lethal behav-
of online dialogue, the giving and receiving of informal support
iors to the repertoire of established self-injurers, not yet committed
and discussion of proximal life events that trigger self-injury are
message board members, and even nonparticipating message board
most common. Posters also share casual and sometimes very
viewers who are exploring identity options. Indeed, some message
personal information related to the addictive qualities of their
boards contain links to pro-self-injury Web sites where Internet
practice, their fears relating to disclosure, experiences with ther-
users can purchase self-injury paraphernalia such as bracelets or
apy, how they self-injure, and other related health concerns. Al-
clothing that signify self-injury status and cutting clubs have been
though our findings are generally consistent with existing literature
rumored to be a growing form of friendship ritual (Booth, 2004).
on self-injurious behavior, the correlations documented among
Some self-injury Web sites host forums specifically dedicated to
informal support, encouraging disclosure, and advising formal
sharing new self-injury techniques. Discussion of technique shar-
treatment suggest that online interactions may be providing self-
ing, triggers, negative attitudes toward formal and informal help
injurers support and meaning outside the clinical setting.
seeking, and the pleasures and pains of self-injury addiction may
What the self-injurious adolescents in our study appeared to do
influence behavioral choices outside of the virtual realm that are
online is what most people who trust each other do in conversa-
later brought back, shared, and used to ensure support and mem-
tion: exchange support, share personal stories about daily life
bership. It may also make some youth targets for individuals who
events, and voice opinions and ideas. Because the anonymity of
falsely pose as supporters to accomplish other, less benevolent
the Internet inspires the most personal and trusting of exchange
aims. For vulnerable adolescents, the difficulty of ending a strat-
between individuals with little or no previous relationship, online
egy for coping with distress (self-injury) and leaving a needed
sharing may encourage greater and more truthful disclosures
source of support (individual or collective members of the virtual
(McKenna & Bargh, 2000), especially among self-injurers, many
community) may stifle the desire to find alternate ways of coping
of whom suffer from depressive symptomatology (Ybarra et al.,
with stress. Moreover, the low-sense of self-worth common among
2005). For adolescents, this support may be particularly valuable,
self-injurers may expose them to damaging online relationships.
because healthy social and emotional development hinges on their
This study supports the findings of other scholars of Internet and
ability to establish caring, meaningful relationships, to find accep-
development (Suzuki & Calzo, 2004) by suggesting that electronic

---

416
WHITLOCK, POWERS, AND ECKENRODE
forums provide a rich data source for studying issues pertinent to
Booth, S. (2004, February). Cutting clubs: What’s the latest and most
marginalized subgroups of the adolescent population that are hard
shocking new “friendship” ritual? Teen People looks at how a growing
to identify and reach. Observation of interaction as it unfolds,
number of kids are bonding with their peers by slicing themselves with
rather than in retrospective self-reports, eliminates biases relating
razor blades. Teen People, 7, 98.
to self-report and recall and allows study of actual transactions
Briere, J., & Gil, E. (1998). Self-mutilation in clinical and general popu-
between individuals. Moreover, because self-injury is typically a
lation samples: Prevalence, correlates, and functions. American Journal
of Orthopsychiatry, 68, 609 – 620.
private, secretive behavior, the Internet provides a unique opportunity
Brodsky, B. S., Cloitre, M., & Dulit, R. A. (1995). Relationship of
to study exchange between members of a group rarely assembled
dissociation to self-mutilation and childhood abuse in borderline per-
outside of a clinical setting. As such, it may provide a valuable means
sonality disorder. American Journal of Psychiatry, 152, 1788 –1792.
of accessing information and perspectives useful in clinical settings.
Brumberg, J. J. (1992). From psychiatric syndrome to “communicable”
Indeed, self-injury message boards may provide a vehicle for admin-
disease: The case of anorexia nervosa. In C. Rosenberg & J. Golden
istering a Web-based intervention to reach self-injurers.
(Eds.), Framing disease (pp. 134 –154). Piscataway, NJ: Rutgers Uni-
Our methods have limits. It was not possible to observe all 406
versity Press.
message boards identified in this study. The 10 selected for anal-
Claes, L., Vandereycken, W., & Vertommen, H. (2003). Eating-disordered
ysis, although the most active at the time of the study, may not be
patients with and without self-injurious behaviours: A comparison of psy-
generalizable to all self-injury message boards. Message boards,
chopathological features. European Eating Disorders Review, 11, 379 –396.
like all communities, possess their own culture and character and
Compas, B. E. (1987). Coping with stress during childhood and adoles-
are governed by subtle and overt norms and mores. Rules posted
cence. Psychological Bulletin, 101, 393– 403.
and enforced by the moderator is one example of this. How these
Connors, R. (2000). Self-injury: Psychotherapy with people who engage in
affect participant self-selection into specific message board com-
self-inflicted violence. Northvale, NJ: Jason Aronson.
munities and the content of their exchange is unclear. Moreover,
Conterio, K., & Lader, W. (1998). Bodily harm: The breakthrough healing
program for self injurers.