ADOLESCENT SUICIDE STATEMENTS ON MYSPACE
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
CYBERPSYCHOLOGY, BEHAVIOR, AND SOCIAL NETWORKING
Volume 16, Number 3, 2013
ª Mary Ann Liebert, Inc.
DOI: 10.1089/cyber.2012.0098
Adolescent Suicide Statements on MySpace
Scottye J. Cash, PhD,1 Michael Thelwall, PhD,2 Sydney N. Peck, MS,3
Jared Z. Ferrell, MA,4 and Jeffrey A. Bridge, PhD 5
Abstract
The use of social networking sites (SNSs) has proliferated throughout the last several years for all populations,
but especially adolescents. Media reports have also identified several instances in which adolescents broadcast
their suicidal behaviors via the Internet and/or SNSs. Despite the increase in the usage of SNSs, there has been
little research conducted on how adolescents use SNSs to communicate these behaviors. The objective of this
study was to explore the ways in which adolescents use MySpace to comment on their suicidal thoughts and
intentions. Content analysis was used to identify suicidal statements from public profiles on MySpace. The
original sample consisted of 1,038 comments, made by young people ages 13–24 years old. The final sample
resulted in 64 comments, where Potential Suicidality was identified. Through content analysis, the following
subthemes (within the Potential Suicidality theme) were found: Relationships, Mental Health, Substance Use/Abuse,
Method of Suicide, and Statements without Context. Examples and discussion for each subtheme are identified. The
comments referenced a significant amount of hopelessness, despair, and desperation. This study provides
support that adolescents use public Web sites to display comments about their suicidal thoughts, behaviors, and
possible intentions. Future research is warranted to explore the relationship between at-risk behaviors and
suicidality as expressed on SNSs.
Introduction nicate with friends, and write blogs. Lenhart and Madden7
found that 55 percent of adolescents who are online use a
R ecent media reports have highlighted cases in which
the Internet, including social networking sites (SNSs), has
contributed to individual suicidal behavior,1,2 suicide pacts in
SNS, with girls using these sites primarily to reinforce exist-
ing friendships and boys using the sites mostly to make new
friends and flirt.
Japan,3 and a suicide epidemic among young subscribers to a While SNSs may provide ways to stay connected with a
popular SNS in Wales.4 While there has seen a recent increase network of friends, they may also provide individuals a place to
in research on how adolescents use the Internet, SNSs, and present ideas, feelings, and moods that are uncomfortable to
other forms of technology for communication,5–8 only one share in-person, including self-harm or suicidal thoughts.11–13
study to date,9 has examined the use of SNSs in relation to
expressions of suicidal thoughts or behavior. SNSs and at-risk behaviors
Adolescents use SNSs to display at-risk behaviors, in-
Adolescents and social networking
cluding sexual behavior, violence, and substance use. Moreno
Boyd and Ellison10 define a SNS as ‘‘web-based services et al.14 examined the display of at-risk behaviors on adoles-
that allow individuals to (a) construct a public or semi-public cent’s MySpace profile pages. They found that of a total of 500
profile within a bounded system, (b) articulate a list of other publicly available MySpace profiles for 18-year olds, 54 per-
users with whom they share a connection, and (c) view and cent contained at least one high-risk behavior, with 24 percent
traverse their list of connections and those made by others referencing sexual behaviors, 41 percent referencing sub-
within the system.’’ SNSs allow users to create and post stance use, and 14 percent referencing violence. In their 2010
profiles, upload pictures and videos, network and commu- study, Moreno et al.15 used content analysis to identify
1
College of Social Work, The Ohio State University, Columbus, Ohio.
2
Statistical Cybermetrics Research Group, School of Technology, University of Wolverhampton, West Midlands, United Kingdom.
3
Nursing Department, Elmira College, Elmira, New York.
4
Department of Psychology, University of Akron, Akron, Ohio.
5
Department of Pediatrics, The Ohio State University, Center for Innovation in Pediatric Practice, The Research Institute at Nationwide
Children’s Hospital, Columbus, Ohio.
166ADOLESCENT SUICIDE STATEMENTS ON MYSPACE 167
references to alcohol use. They found that 56 percent (out of data were more available and less restricted by privacy pro-
400 profiles of 17–20-year olds living in one Washington tection algorithms, such as access and antispidering con-
county) had references to alcohol use, more frequently made trols.14,15,18,20–23 Thelwall et al.23 demonstrated that emotions
by males (54 percent) who were white (70 percent). Moreno could be mined from data extracted from MySpace, which
et al.16 conducted focus groups to explore reasons why ado- made MySpace an appropriate source to collect data for the
lescents use SNSs to display alcohol references, finding that current study.
adolescents posted these references to look cool, therefore To find the subjects, MySpace profile owners were selected
providing some level of social gain among their peers. if they:
The research on adolescents’ use of SNSs to display at-risk (a) gave the US as their location
behaviors is increasing; however, we know very little about (b) had a public profile
their display of suicidal thoughts or intentions on these sites. (c) were not self-identified as musicians, comedians, or
The purpose of this study is to explore the content of suicidal movie makers
statements available on MySpace public profiles of adoles- (d) had 2-1000 friends and, hence, were probably active
cents aged 13–24 years old. The findings from this study can normal members.
be used to inform health and mental health professionals
After the MySpace profile owner’s profile pages were
about how adolescents have used SNSs to communicate their
identified, all publicly available comments on their MySpace
suicidal thoughts and intentions and possibly develop inter-
profile page were downloaded. The data were then parsed
ventions that use this type of technology.
based on a set of comments that had suicidal language (see
Fig. 1).
Methodology
Setting
MySpace (www.myspace.com) was the SNS chosen to
provide the data for this study. MySpace met the SNS defi-
nition provided by Boyd and Ellison10 and allowed for
downloading of information from public profiles.
Institutional review board
The Institutional Review Board (IRB) at the Ohio State
University was consulted on the project and whether or not
the project met the exemption guidelines. Moreno et al.,17
discussion of research ethics and their application to research
on SNSs was also used as a guide for conducting this re-
search. The study met the criteria for exemption for the fol-
lowing reasons: (a) all data were collected from publicly
available profiles on MySpace; (b) data were downloaded
using an algorithm and no specific identifying data were
collected (e.g., name, address, and contact information); (c)
the study was observational and no participants were ever
contacted; (d) no identifying data were included in the da-
taset; (e) any names included in the sample quotes were
changed; and (f) the specific quotes were altered slightly to
ensure that the commenter could not be identified using a
search engine.
Data
Data were collected from publicly available profiles and
their respective comments on MySpace (www.myspace.com).
MySpace was launched in 2003 and gained popularity
throughout the next several years as a place for teenagers to
post profiles and create online community.18,19 Data from the
Pew Research Project in 20066 demonstrated that 55 percent
of Internet users, ages 12–17, had a profile on a SNS. Data
collected from comScore19 on the unique visitor count to
MySpace and Facebook showed from August 2005 to May
2009, MySpace had more unique visitors than Facebook.
After May 2009, however, Facebook clearly dominated the
social networking world and has continued to increase its
presence, while MySpace’s presence has decreased. Even
though MySpace’s popularity on the Internet has waned,
MySpace has been the SNS primarily used by researchers as FIG. 1. Sample selection process.168 CASH ET AL.
Search algorithm Table 1. Suicide Phrasesa Used for Parsing Data
The downloading of profiles was undertaken automati- Never wake up Suicide
cally using purpose-built software and the downloading of Suicidal Can’t go on
comments was achieved semiautomatically using purpose- Life not worth living Kill myself
built software. Data set 1 originated with a list of 40,000 Suicide pact Suicide plans
public MySpace profile IDs from about 51,000 in steps of Ready to jump Go to sleep forever
5,193 representing a systematic sample of members with Suicide plan Want to die
various joining dates up to July 3, 2006. Data from the profile Be dead Better off without me
Better off dead Tired of living
page of each member was downloaded during March 3–4,
Rock bottom Don’t want to be here
2008, and the declared country of origin extracted as well as End my life Die alone
the number of friends of each member. Go to sleep forever
a
Suicide Phrases are based on key phrases identified from
Waves of data www.suicideforum.net
The selected subset was downloaded again in December
2008 and a further subset of comments was selected of those
who had received p4,000 comments and still had a public collecting the phrases on the site did not violate the forum’s
profile. For this subset, all comments received by each indi- policies on site usage. Specifically, the site, at that time, pos-
vidual were extracted. For each MySpace profile downloaded ted their policies stating that forum could not be used as a
as above, one commenter was selected at random, if any ex- mechanism to solicit information from the individual users
isted, that is, one person who commented to the member. All for research and/or educational projects. This allowed the
comments made to this commenting friend were then ex- researcher the opportunity to collect only information on
tracted as long as they suicidal language without ever having to contact the indi-
viduals directly or collect identifying information. Finally, the
(a) had a public profile suicideforum.net has moderators who monitor the forums to
(b) did not self identify as a musicians, comedians, or help identify individuals who are posting statements specif-
movie makers ically related to potential suicide intentions. Generating a
(c) had received p4,000 comments. suicide-specific list of phrases was necessary to create a sys-
Commenters failing this test were rejected and another tematic way to identify publicly available comments that in-
commenter selected instead until a valid commenter was cluded a reference to potential suicidal behavior.
found or all commenters had been checked. Musicians, co- Once the list of phrases was generated, the list was then
medians, and or movie makers were excluded from the used to parse the dataset for comments matching one of the
sample as their profile pages did not necessarily represent one phrases (see Table 1); these comments were selected for in-
person commenting to another person. In addition, the clusion in the study.
comments on their pages may have discussed suicide in terms Other data extracted included the age of the commenter and
of media (e.g., songs, movies, etc.) rather than actual com- the commentee, based on the age identified in the person’s
ments about suicidal behavior. A person also had to have profile. These data were adjusted to reflect their age at the time
fewer than 4,000 comments, as we found that the profiles that the comment was made, instead of the age at the time of the
where there were more than 4,000 comments often re-
presented a professional user of MySpace rather than an in-
dividual making comments to other individuals. The final
Table 2. Percentage of Comments by Matching Texta
raw data set used was a collection of all comments extracted
for Total Sample
by the above process.
Matching text Percent total sample (n = 1,038)
List generation process. A list of phrases that included
some reference to potential suicidal behaviors was collected Be dead 16.6%
from public posts on the Suicide Forum Web site (www Better off dead 0.2%
.suicideforum.net). The first author is a social work researcher Better off without me 0.2%
and has experience working with individuals who have at- Can’t go on 2.4%
Die alone 1.0%
tempted suicide. She read public comments that were posted
Don’t want to be here 0.8%
in the Peer-to-Peer Support group page for the General Sui- End my life 2.2%
cide Forum that was also designated as a place to post com- Go to sleep forever 0.1%
ments regarding ‘‘if you are suffering from suicidal feelings.’’ Kill myself 24.4%
During September and October 2009, the first author read Never wake up 3.2%
postings in the forum to identify key phrases that individuals Ready to jump 1.0%
were using to identify potential suicidal behaviors. Rock bottom 2.5%
The suicideforum.net Web site was chosen as a source to Suicidal 5.6%
identify common phrases used to indicate potential suicid- Suicide 29.4%
ality for several reasons. First, the forum is open to the public. Tired of living 0.4%
Want to die 10.2%
Second, since the forum was designed for people who are
seeking information or help about suicide, the forum seemed a
Matching text is based on suicide phrases from www.suicideforum
like a logical choice and met the needs of the project. Third, .netADOLESCENT SUICIDE STATEMENTS ON MYSPACE 169
Table 3. Age of Sender of Comment Total Sample selected for analysis (see Tables 1 and 2). Table 2 provides
support that the list of suicide-related phrases were useful in
Age at comment Percent total sample (n = 1,038) parsing the data as there was at least some representation of
13 2.2% all of the individual phrases in the total sample. Not surpris-
14 6.2% ing, the phrases that had the highest frequency were ‘‘suicide’’
15 13.9% (29.4 percent), ‘‘kill myself’’ (24.4 percent), and ‘‘be dead’’ (16.6
16 14.9% percent). The phrases with the lowest frequency ( < 0.1 per-
17 15.9% cent) were ‘‘go to sleep forever,’’ ‘‘better off dead,’’ ‘‘better off
18 11.7% without me,’’ ‘‘tired of living,’’ and ‘‘don’t want to be here.’’
19 12.4% The data structure and methodology used to extract the
20 7.0% data from MySpace limited the ability to group comments by
21 6.4% sender. For the majority of the comments, the age of the
22 2.8%
person making the comment (commenter) and the self-
23 3.3%
24 3.4% declared age of the person receiving the comment (com-
mentee) were automatically extracted from MySpace. The
initial dataset included 1,762 comments with ages of the
data download. The ages are based on the self-reported age commenter and/or commentee ranging from 13–35. After
recorded in the profile and are estimates since dates of birth applying the criterion for the commenters age (13–24-year
were not available. Data on the commenters gender was also olds), the sample size was reduced to 1,038 comments. Table
extracted when the profile included this information. 3 provides a description of the ages for the total sample.
In total, over two million comments were extracted from
MySpace to be searched for suicide-related comments. The Sample reduction process using themes. The sample
use of four different data sets controlled for differences be- was reduced in four phases, using an initial coding scheme.
tween MySpace members joining at different times. All 1,038 comments were coded using the themes and cor-
responding operational definitions (Table 4). (Note: see the
Sample Coding section for more details about coding procedures).
The unit of analysis for this study is the individual com- After Phase 1 coding, the sample was reduced to 490
ment level. Comments with one of the suicide phrases were comments, removing comments unrelated to suicidal
Table 4. Themes and Definitions for Coding All Comment Data
Theme Definition
Dealing with repercussions of suicide Comment referenced another person’s suicide and the difficulties the
commenter was facing.
Denying suicidality Comment denied that previous statements or actions were indicating suicidal
thoughts
Duplicate post Exact same comment was referenced elsewhere in the database
Fiction Comment was found to be quoting prose selection or movie lines
Foreign language Comment was not in English and could not be fully analyzed
Friend’s response to suicidal language Comment referenced earlier comment by friend, expressing concern.
Needs context Comment had no context with which to evaluate the language
Not relevant to suicide Comment used key words in a way that did not indicate suicidal discourse
Past suicidal actions Comment referenced personal previous suicide attempts, but no current
suicidal ideation
Poetry Comment was found to be quoting poetry selection
Potential suicidality Comment indicated potential serious nature of suicide threat based on context
provided
Referring to someone else’s Comment referred to another person’s suicidal actions, but not a successful
suicidal behavior suicide attempt
Hyperbole about suicide Comment made suicidal threat, but context indicated threat was not of a
serious nature, such as the use of emoticons or LOL (laugh out loud)
following threat
Seems to be military Comment referenced military use of suicidal language
Slang Suicidal language used as slang, such as referencing the suicide of an inanimate
object
Song lyrics Comment quoting song lyrics, which included a key word
Spam Comment indicating a chain mail or computer generated comment specifically
requiring the receiver to forward comment to a specified number of people
Sports Comment using a key word in a sports related manner, such as running suicide
drills
Suicide bombing Comment specifically discussing suicide bombing events
Talking about suicide Comment discussing suicide, but in a distanced, academic manner, such as
discussing a school presentation about suicide170 CASH ET AL.
Table 5. Subthemes and Quotes Within Potential Suicidality Theme
Theme Subtheme Example quote Example quote
Relationships Unknown ‘‘I’m writing to tell you I want to die. ‘‘Hey baby omg I’m so pissed and I hate
The feelings are back & I think it is life I want to f**king kill myself.. i am
time to say goodbye. I’ll miss you not saying this to get any attention .’’
and I am sorry if I cause you Any
Pain.’’
Relationships Break-ups ‘‘No more bf.no more anything.time ‘‘Hey. You are probly mad at me now. Or
to take me outside n shoot my you may hate me. But im gonna tell you
f**king brains out.I can’t get my how I feel right now and what it is I
shit together because it will fall apart want. I want you. Thats the only thing I
again.I’m gonna die alone because want. I feel miserable now. Ive hurt you
no one will love me the way I love and everything. I’m just so upset with
them .’’ myself right now. I want to kill myself
so u wont get hurt by me
anymore.. now u won’t want me any
more. Never. And im really depressed.
And so I feel suicidal and not sure about
my life.’’
Family ‘‘Im depressed mainly cause of my ‘‘Omg. that so horrible, good luck and
Problems parents. I wish I could kill myself.’’ everything, thats just really tough, im
not feeling so well either. i feel like
everyones mad at me again, and theres
even more family problems than
ever.what else is new. ive felt like this
hundreds of times before. i hate it.
everything will soon turn to crap and ill
feel again like i want to kill myself. i just
cant take it anymore.’’
Friends ‘‘I am writing to let you know that I ‘‘hey yeah well right know at this seconde
want to die. The bad feeling is back I want to die. Just want to not be here
and I think it’s now time to say good- any more. I have tried to call you but
bye. I will miss you a lot and I am you dont pick up the phone anymore.’’
very sorry if I cause you pain’’
Mental Health/ ‘‘Lol yeah. Wellfor the most part. I’m ‘‘Spending most of my time in a drunken
Substance use not doing good.I just now got out haze. I was doing ok, but the past few
of the mental hospital. I was cutting days i’ve been really sad. I dont see a
myself and thinking of suicide again. way to get back to anywhere close to
So now I’m on antidepressants but what i had with him. So . . year.feelin
they won’t kick in for another four pretty suicidal right now.’’
weeks .’’
Method of Suicide Shoot Self/Gun ‘‘Time to take me outside and shoot my ‘‘Hey I have a gun I should kill myself’’
f**king brains out’’
Knife ‘‘Fine im gunna go take Gina’s knife ‘‘I just want to walk in the kitchen grab a
and kill myself’’ knife and cut my effin throat.’’
Hit by a Car ‘‘I wanna die more than anything right
now I could just jump into the
middle of the street and wait for a car
to come and I would get down on
my knees to tell god how sorry I am
and wish everything was different
and that I could get you to
understand, then just die.’’
Statements ‘‘I am going to kill my self right ‘‘I told myself lots of times I want to die
without now...’’ that I cant even count them,’’
Context
Minor changes have been made to the quotes so that the commenters cannot easily be identified through search engines.
behaviors (e.g., a song lyrics, running suicides; see Table 5). as hyperbole, the comment included words that are associ-
During Phase 2 coding, the sample was reduced to 105 ated with suicidal behaviors, and/or the comment did not
comments by removing comments that did not reflect have enough context to rule out that it was not serious. Phase
serious/potential suicidal behaviors. The criteria for includ- 3 coding removed comments that still included potential/
ing a comment that reflected a serious/potential suicidal serious suicidal behaviors; however, these comments referred
behavior included: the comment could not be characterized to someone else’s suicidal behaviors, not the commenters.ADOLESCENT SUICIDE STATEMENTS ON MYSPACE 171
Table 6. Percentage of Comments of Matching Texta for those included Potential Suicidality subsample. Approxi-
for Potential Suicidality Subsample mately, half of the Potential Suicidality sample (46.9 percent)
was between the ages of 13 and 17.
Matching text Percent subsample (n = 64)
Be dead 0.0% Analyses
Better off dead 1.6% Content analysis was the analytical method used to gen-
Better off without me 0.0% erate themes. Content analysis allows for constant compari-
Can’t go on 0.0%
sons between different data points with the overall goal of
Die alone 1.6%
Don’t want to be here 1.6% developing themes (to saturation) to understand meaning
End my life 0.0% from text.24–26 Content analysis also provided a mechanism
Go to sleep forever 0.0% for providing counts and frequencies on the themes.
Kill myself 51.6%
Never wake up 1.6% Coder training
Ready to jump 0.0%
Rock bottom 1.6% Two researchers conducted the data coding process.
Suicidal 9.4% Themes were generated until there was saturation. As new
Suicide 14.1% themes were identified, we created an operational definition
Tired of living 0.0% to assist with consistency throughout the process (see Table 4
Want to die 15.6% for themes and corresponding operational definition). Inter-
a
rater reliability was assessed by having the two coders indi-
Matching text is based on suicide phrases from www.suicideforum
.net
vidually code the first 50 comments, and then the coders
discussed the assigned code for each comment. The inter-
rater reliability on the first fifty comments resulted in a 95
After Phase 4 coding, the sample included only serious percent theme agreement. Disagreements on assigning the
comments about the commenter’s Potential Suicidality (n = 64). theme to a particular comment were discussed before a theme
Table 6 provides the percentages for matching suicide was assigned to the comment. Throughout the coding pro-
phrases for only the Potential Suicidality sample. The per- cess, when either researcher was unsure about how to code a
centages of matching suicide phrases were somewhat differ- specific comment, the researchers discussed the comment and
ent for this subsample as compared to the total sample. For arrived at a mutually agreed upon theme. Some of the com-
example, the phrase ‘‘kill myself’’ was present in 51.6 percent ments had aspects of several of the subthemes; when this
of the matched phrases for the Potential Suicidality sample, occurred, the coders discussed the comment and decided on
whereas it was only present in 24.4 percent of matched the primary theme in the comment. Once the initial coding
phrases in the total sample. The phrase ‘‘want to die’’ was was completed, one of the researchers reviewed each code to
identified for 15.6 percent of the Potential Suicidality sample ensure the consistent coding throughout the analysis.
as compared to 10.2 percent of the total sample. The phrase
‘‘suicide’’ was used more in the total sample (29.4 percent) as Results
compared to the sample of Potential Suicidality comments Within Potential Suicidality, comments were coded to
(14.1 percent). identify more specific themes that provide additional infor-
mation on the types of comments made by users who are
Sample characteristics. Limited demographic data were expressing serious and potential suicidal behaviors. All re-
available for the sample characteristics of the Potential Sui- sults are presented only for the Potential Suicidality sample.
cidality sample. Forty percent of the samples were female, 37.5 Samples of comments for the themes with Potential Suicidality
percent were male, and 21.9 percent gender were unknown. are presented in Table 5. The presentation of these sample
Table 7 provides a distribution of the age of the commenter comments is based on a content analysis framework
Table 7. Age of Sender of Comment for Potential Table 8. Subthemes Within Potential Suicidality
Suicidality Subsample (n = 64)
Age at comment Percent subsample (n = 64) Theme Percentage (n = 64)
13 3.1% Unknown context 51.6%
14 6.3% Relationship
15 7.8% Unknown context 20.3%
16 15.6% Family 4.7%
17 14.1% Break-up 15.6%
18 10.9% Friends 1.6%
19 21.9% Mental health/substance use 6.3%
20 4.7% Identified a method
21 6.3% Knife 1.6%
22 3.1% Hit by a car and knife 1.6%
23 3.1% Gun 3.0%
24 3.1% No method identified 93.8%172 CASH ET AL.
Table 9. Comparison of Potential Suicidality Subtheme by Age
Age No context Relationship unknown Relationship family Relationship break-up Relationship friendship Mental health
13–18 51.4% 24.3% 0.0% 16.2% 2.7% 5.4%
19–24 51.9% 14.8% 11.1% 14.8% 0.0% 7.4%
previously used,15 in a study,14 to examine statements made Multiple themes
on profiles from MySpace.
One comment included two themes: mental health and a
Table 8 provides the frequencies of each subtheme within
relationship break-up. This person wrote that he/she was
Potential Suicidality.
recently released from a mental hospital for cutting and sui-
cidal thoughts and they were also going through a break-up.
Unknown context
Bivariate analyses
A little over half of the comments (33/64; 51.6 percent) did
not include any specific reference to the context in which the Additional analyses explored the relationship between age
persons were writing about. Within this theme, the com- and theme and gender and theme. Table 9 presents the
menter only indicated a desire to die or kill themselves. A few findings from a crosstab using a Fisher’s Exact Test compar-
comments reflected a previous attempt and/or previous ing age (13–18 vs. 19–24) on the type of theme. The percentage
thoughts of suicide. of themes without context was similar for both groups (13–18,
51.4 percent; 19–24, 51.9 percent). There were differences,
Relationships within the relationship theme; however, these were not sta-
tistically significant. The 13- to 18-year olds had higher fre-
The Relationships subtheme had the most prevalent rep-
quencies of comments regarding relationships with unknown
resentation (n = 27; 42.2 percent) in the data. In 20 percent of
context (24.3 percent) as compared with 19–24-year olds (14.8
the sample, the type of relationship the young person was
percent). The 19–24-year olds identified family relationship
referring to in the comment could not be determined. When
problems in 11.1 percent of comments compared to 0 percent
the type of relationship was identified, young people were
for 13–18-year olds.
most affected by break-ups (15.6 percent). The relationship
Table 10 provides a comparison of males and females on
comments, in general, reflected desperation stating that they
the Potential Suicidality subthemes. A crosstab with a Fisher’s
cannot live without the person or within the situation. Some
Exact Test was used to test for differences. Comments made
comments also reflected the commenter’s sense of not feeling
by males, on the whole, had no context (62.5 percent). The
loved.
highest percent of comments made by females were 38.5
percent no context and 30.8 percent were coded as relation-
Mental health and substance use
ship, but unknown. The differences between the two groups
Four comments (6.3 percent) were coded as a mental were not statistically significant.
health/substance use. The comments were revealing in terms
of the commenters’ acknowledgement of having a mental Discussion
illness (e.g., depression, anxiety, etc.) and/or a mental health
The results of this preliminary study document the exis-
treatment facility hospitalization. One commenter also men-
tence of serious suicidal comments on MySpace pages. The
tioned being in a drunken haze. The comments within this
majority of comments had no context. Forty two percent of
theme reflected a sense of worthlessness, hopelessness, and
comments, however, had content that was related to a rela-
that life did not matter.
tionship. The suicidal statements regarding relationship
problems were most prevalent for break-ups, followed by
Method of suicide
family relationship, and a few related to friendships. Inter-
Within some comments, the person indicated that he/she personal conflict or loss and family discord are common
was thinking about suicide and they identified a particular precipitants of suicidal behavior and suicide in young
method to commit suicide. Some comments were specific in people.27–29 The present results suggest that adolescents may
identifying an actual plan. The three methods identified in- use SNSs to reach out for help and support. However, our
cluded shooting oneself with a gun (n = 2) using a knife (n = 1), data cannot disentangle whether comments posted on My-
or stepping out in front of a car and using a knife (n = 1). The Space received a response, and if so, the nature or content of
majority of comments did not identify a method. those responses.
Table 10. Comparison of Potential Suicidality Subtheme by Gender (n = 50)
Gender No context Relationship unknown Relationship family Relationship break-up Relationship friendship Mental health
Male 62.5% 20.8% 4.2% 8.3% 0.0% 4.2%
Female 38.5% 30.8% 3.8% 19.2% 0.0% 7.7%ADOLESCENT SUICIDE STATEMENTS ON MYSPACE 173
Approximately, 6.3 percent of suicide statements on My- sites and researchers. Alternatively, with the proliferation of
Space involved references to mental health problems (mostly Facebook, researchers and practitioners could employ crea-
depression) with one comment that included a brief mention of tive approaches for reaching out to adolescents. One research
substance use. Mood disorders and alcohol/substance are risk option would be to recruit participants directly from the ad-
factors for suicide and attempted suicide in adolescents.30 The vertising section of Facebook, asking adolescents to partici-
relatively small percentage of comments that included mental pate in studies on mental health31,32. Another option is for
health references was not all that surprising, since we are not practitioners to design advertisements that inform individu-
sure if the individuals making the comments were aware of als of options for mental health or substance use assistance.32
their mental health status and/or if this is something that they Moreno et al.32 argue that ‘‘given the frequency and conse-
are willing to disclose. Moreno et al., however, have found that quences of depression as well as the inadequacy of current
SNSs, like MySpace and Facebook, provide a venue for ado- help-seeking among college students, innovative methods to
lescents to talk about sensitive or stigmatizing behaviors.31,32 identify those at risk and provide them appropriate services
A relatively small percentage (6.2 percent) of suicide are warranted.’’
statements on MySpace pages described specific methods of Facebook and Google have both created a mechanism to
suicide within the context of other themes. Specifically, the provide information on suicide support options (e.g., inter-
comment included references to using a knife when a rela- national hotline numbers). For example, if a person types
tionship theme was also identified. One person indicated that suicide into Facebook’s help field, then a list of suicide pre-
he/she would step into the road to be hit by a car or would vention hotlines are provided to the user. Google has a
use a knife to cut their throat. A gun was mentioned as a slightly different approach, where if a person searches on
method of suicide within the context of a relationship break- keywords related to suicide, the search results page lists as its
up. Finally, one comment referenced using a gun as a method, first item a red telephone, with a phone number to the
but there was no additional context provided. National Suicide Prevention Lifeline.
Foundations, organizations, and support groups are also
Limitations trying to harness the power of SNSs to connect with people
who need support. For example, Inspire Foundation USA and
This study has several potential limitations. First, data
their counterpart, ReachOut.com have a social networking
were collected from only one SNS, MySpace. MySpace at this
page and a Web site targeted to individuals (13–24) to use to
time is struggling to regain its presence on the Internet. Sec-
help that person or one of their friends get through a tough
ond, the nature of the comment and whether or not it was
time. The ReachOut.com webpage includes fact sheets on
considered serious is based on a determination made by the
mental health issues (e.g., anxiety, depression, eating disor-
two coders rather than an objective measure. The number of
der, and self-harm), drug, alcohol, and tobacco use, family
comments that were coded as serious suicide comments was
problems, and suicide). The site also includes multimedia and
significantly large, which made it necessary to only focus on
interactive approaches to provide different mechanisms to
one subset of the theme, Potential Suicidality despite the fact
reach out to people who respond to different types of help.
that many other themes were represented. Finally, the lack of
One of the unique features of ReachOut.com is their use of a
complete demographic data makes it difficult to accurately
youth advisory board, who vet all of the materials that are
describe the sample and generalize the findings, and it also
provided on the site so that the materials are youth-focused
limits the ability to look at associations between demographic
and relevant.
variables and suicide-related comments.
Other suggestions include using online groups to help in-
crease positive social support systems, while reducing the
Implications
impact of negative social interactions34 and to explore com-
These limitations notwithstanding, the findings in the mon risk factors for suicidal behaviors, substance use, sexual
present study and media reports suggest that adolescents do displays, and violence to better target individuals and inter-
use SNSs to discuss suicidal thoughts and intentions. The ventions.33,35 It is vital that we continue to try to understand
blending of computer science with social science may provide youth and the multitude of ways in which they convey their
a unique opportunity to develop ways of identifying at-risk feelings and make cries for help. As we learn more, we will be
adolescents using computer algorithms to identify the com- able to utilize this information to reach out to them and help
ments and social scientists to provide context. However, as them in ways that make sense to them.
Boyd et al.33 argue, simply using algorithms to identify self-
harm or other risk behaviors is not enough; it is necessary to Acknowledgments
identify both positive and negative instances of behaviors as The authors would like to thank Dr. Megan Moreno for her
online communities may serve as its own risk factor or as comments and feedback on an earlier version of this manu-
prosocial support. Boyd et al.33 write ‘‘in moving forward script.
to address problematic self-harm content—and youth-
generated problematic content more generally—we must Author Disclosure Statement
begin embracing visibility, both as a source of information
from which we can learn and as a potential channel through No competing financial interests exist.
which we can engage.’’
Partnerships between researchers and SNSs would require References
sites to allow researchers access to the profile data, which 1. The Associated Press. (2009) Nurse is accused of using In-
may prove controversial given concern for privacy. However, ternet to encourage suicides. www.nytimes.com/2009/10/
this should not stop conversations between social network 18/us/18suicides.html (accessed Jun. 21, 2011).174 CASH ET AL.
2. Steinhauer J. (2008) Verdict in MySpace suicide case. www 21. Hinduja S, Patchin JW. Personal information of adolescents
.nytimes.com/2008/11/27/us/27myspace.html (accessed Jun. on the Internet: A quantitative content analysis of MySpace.
21, 2011). Journal of Adolescence 2008; 31:125–146.
3. Rajagopal S. Suicide pacts and the Internet. British Medical 22. Moreno MA, Vanderstoep A, Parks MR, et al. Reducing
Journal 2004; 329:1298–1299. at-risk adolescents’ display of risk behavior on a social net-
4. Nordland R. (2009) How grim was my valley: a teen suicide working web site: a randomized controlled pilot interven-
epidemic sweeps a community in Wales. www.newsweek tion trial. Archives of Pediatric Adolescent Medicine 2009;
.com/id/117749. (accessed Oct. 19, 2009). 163:35–41.
5. Lenhart A. (2009) Teens and mobile phones over the past five 23. Thelwall M, Wilkinson D, Uppal S. Data mining emotion in
years: Pew Internet looks back. Washington, DC: Pew Internet social network communication: Gender differences in My-
& American Life Project. Space. Journal of the American Society for Information Sci-
6. Lenhart A. (2007) Social networking websites and teens: an ence and Technology 2010; 61:190–199.
overview. Washington, DC: Pew Internet & American Life 24. Krippendorf K. (2004) Content analysis: an introduction to its
Project. methodology, 2nd edition. Thousand Oaks, CA: Sage.
7. Lenhart A, Madden M. (2007) Teens, privacy, & online social 25. Miles M, Huberman M. (1994) Qualitative data analysis: an
networks: how teens manages their online identities and personal expanded sourcebook, 2nd edition. Thousand Oaks, CA: Sage.
information in the age of MySpace. Washington, DC: Pew In- 26. Weber RP. (1990) Basic content analysis, 2nd edition. Thousand
ternet & American Life Project. Oaks, CA: Sage.
8. Lenhart A, Madden M, Hitlin P. (2005) Teens and technology: 27. Beautrais AL. Risk factors for suicide and attempted suicide
youth are leading the transition to a fully wired and mobile nation. among young people. Australian Journal of Psychiatry 2000;
Washington, DC: Pew Internet & American Life Project. 34:420–436.
9. Huang Y, Goh T, Liew, CL. (2007) Hunting suicide notes in 28. Brent DA, Baugher M, Bridge J, et al. Age-and sex-related
web 2.0—Preliminary findings. Ninth IEEE International risk factors for adolescent suicide. Journal of the American
Symposium on Multimedia. Los Almitos, CA: IEEE Com- Academy of Child and Adolescent Psychiatry 1999; 38:1497–
puter Society. 1505.
10. Boyd DM, Ellison NB. Social network sites: definition, his- 29. Gould MS, Fisher P, Parides M, et al. Psychosocial risk fac-
tory, and scholarship. Journal of Computer-Mediated tors of child and adolescent completed suicide. Archives of
Communication 2008; 13:210–230. General Psychiatry 1996; 53:1155–1162.
11. Adams J, Rodham K, Gavin J. Investigating the ‘‘self’’ in 30. Cash SJ, Bridge JA. Epidemiology of youth suicide and
deliberate self-harm. Qualitative Health Research 2005; suicidal behavior. Current Opinions in Pediatrics 2009;
15:1293–1309. 21:613–619.
12. Baker D, Fortune S. Understanding self-harm and suicide 31. Moreno MA, Egan KG, Brockman, L. Development of a
websites: a qualitative interview study of young adult researcher codebook for use in evaluating social network-
website users. Crisis 2008; 29:118–122. ing site profiles. Journal of Adolescent Health 2011; 49:
13. Whitlock JL, Powers JL, Eckenrode, J. The virtual cutting 29–35.
edge: the Internet and adolescent self-injury. Developmental 32. Moreno MA, Christakis DA, Egan K, et al. A pilot evaluation
Psychology 2006; 42:407–417. of associations between displayed depression references on
14. Moreno MA, Parks MR, Zimmerman FJ, et al. Display of Facebook and self-reported depression using a clinical scale.
health risk behaviors on MySpace by adolescents: preva- Journal Behavioral Health Services Research 2012; 39:295–
lence and associations. Archives of Pediatric Adolescent 304.
Medicine 2009; 163: 27–34. 33. Boyd D, Ryan J, Leavitt, R. Pro-self harm and the visibility of
15. Moreno MA, Briner LR, Williams A, et al. A content analysis youth-generated problematic content. A Journal of Law and
of displayed alcohol references on social networking web Policy for the Information Society 2010; 7:1–32.
site. Journal of Adolescent Health 2010; 47:168–175. 34. Hirsch JK, Barton AL. Positive social support, negative social
16. Moreno MA, Briner LR, Williams A, et al. Real use or ‘‘real exchanges, and suicidal behavior in college students. Journal
cool’’: adolescents speak out about displayed alcohol refer- of American College Health 2011; 59:393–398.
ences on social networking websites. Journal of Adolescent 35. Berkman Center for Internet & Society. (2008) Enhancing
Health 2009; 45:420–422. child safety & online technologies: final report of the internet
17. Moreno MA, Fost NC, Christakis DA. Research ethics in the safety technical task force. Boston, MA: The Berkman Center
MySpace era. Pediatrics 2008; 121:157–161. for Internet & Society at Harvard University.
18. Boyd DM. (2008) Taken out of context: American teen soci-
ality in networked publics. Unpublished dissertation,
Graduate Division, University of California, Berkley.
19. comScore. (2011) Facebook domination vs. MySpace fall: data Address correspondence to:
comparison. www.onlinemarketing-trends.com/2011/06/ Dr. Scottye J. Cash
facebook-domination-vs-myspace-fall.html (accessed Dec. College of Social Work
20, 2011). The Ohio State University
20. Bonneau J, Anderson J, Danezis G. (2009) Prying data out 1947 College Road
of a social network. International Conference on Advances Columbus, OH 43210
in Social Network Analysis and Mining (ASONAM). Los
Almitos, CA: IEEE Computer Society. E-mail: cash.33@osu.eduYou can also read
