Content uploaded by Nicholas Brody
Author content
All content in this area was uploaded by Nicholas Brody on Feb 08, 2016
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=rcmm20
Download by: [122.226.181.124] Date: 25 January 2016, At: 13:56
Communication Monographs
ISSN: 0363-7751 (Print) 1479-5787 (Online) Journal homepage: http://www.tandfonline.com/loi/rcmm20
Bystander Intervention in Cyberbullying
Nicholas Brody & Anita L. Vangelisti
To cite this article: Nicholas Brody & Anita L. Vangelisti (2015): Bystander Intervention in
Cyberbullying, Communication Monographs, DOI: 10.1080/03637751.2015.1044256
To link to this article: http://dx.doi.org/10.1080/03637751.2015.1044256
Published online: 15 Jun 2015.
Submit your article to this journal
Article views: 1424
View related articles
View Crossmark data
Bystander Intervention in Cyberbullying
Nicholas Brody & Anita L. Vangelisti
Cyberbullying incidents often occur in the presence of other bystanders. The inaction of
bystanders can augment the deleterious effects of bullying on a victim. However,
bystanders can intervene to stop a cyberbullying incident or offer support to the victim.
Two studies were conducted to examine variables that were expected to influence the
propensity of a bystander to take action in cyberbullying incidents: the number of
bystanders, the anonymity of the bystander, and the closeness between the bystander and
the victim. Results supported the diffusion of responsibility effect. A higher number of
bystanders was negatively associated with bystander intervention. Moreover, the
perceived anonymity of bystanders was negatively related to the propensity to intervene,
and closeness with the victim was associated with a greater tendency to intervene and
support the victim. Altogether, the results shed light on the interplay of context,
relationships, and technology in the behavior of bystanders to a cyberbullying episode.
Keywords: Cyberbullying; Bullying; Computer-mediated Communication; Bystander
Intervention; Anonymity
College students are voracious users of communication technology. Nearly 100% of
undergraduate students at four-year colleges access the Internet regularly, 86% are
members of a social networking site (SNS), and 96% own a cell phone (Smith,
Rainie, & Zickuhr, 2011). As individuals continue to increase their technology use,
the prevalence of hurtful online behaviors has the potential to increase, as well.
Cyberbullying—that is, bullying behavior that occurs via electronic or digital media
—has received attention from both researchers and the popular press (Tokunaga,
2010). Although bullying is often cited as a concern only among middle and high
school students, it is common regardless of age. Indeed, upward of 50% of college
Nicholas Brody (Ph.D., University of Texas at Austin) is an Assistant Professor at the University of Puget Sound.
Anita L. Vangelisti (Ph.D., University of Texas at Austin) is the Jesse H. Jones Centennial Professor of
Communication at the University of Texas at Austin. Correspondence to: Nicholas Brody, Department of
Communication Studies, University of Puget Sound, 1500 N. Warner Street, Tacoma, WA 98416, USA. E-mail:
nbrody@pugetsound.edu
Communication Monographs
2015, 1–26
ISSN 0363-7751 (print)/ISSN 1479-5787 (online) © 2015 National Communication Association
http://dx.doi.org/10.1080/03637751.2015.1044256
This article was originally published with errors. This version has been corrected. Please see Erratum (http://dx.doi.org/
10.1080/03637751.2015.1064248)
Downloaded by [122.226.181.124] at 13:56 25 January 2016
students report being bullied in an online environment, and 30% of victims say that
they were targeted for the first time during college (Kowalski, Giumetti, Schroeder, &
Reese, 2012). In the present study, almost 20% of college students report having
experienced cyberbullying within the previous three months.
Research suggests that one aspect of bullying merits closer attention: the presence of
peers or bystanders observing the bullying episode. As many as 85% of offline bullying
episodes occur in the presence of others (O’Connell, Pepler, & Craig, 1999). The
average Facebook user has 245 online contacts, and the top 10% of users have over 780
friends (Hampton, Goulet, Marlow, & Rainie, 2012). If users have not restricted their
profile, individual posts can have an audience of thousands of individuals. Some
research has examined bystander intervention in traditional (offline) bullying episodes.
Despite the fact that bystanders are sometimes present, they often do not intervene,
choosing instead to observe passively or even join in the bullying (O’Connell et al.,
1999). Whether bystanders directly intervene to stop the bullying or provide emotional
support to the victim, intervention can attenuate the negative consequences of offline
bullying (Matsunaga, 2010).
The present studies investigate the effect of several factors—including the presence
of bystanders, anonymity, and relational closeness—on bystander behavior in online
bullying. This research aims to employ the bystander effect (Darley & Latané, 1968)
and other work on bystander behavior to understand and predict communication
during cyberbullying episodes.
Bullying and Cyberbullying: Definitional Issues
Bullying is described as an intentional, aggressive, repeated act in which there is
frequently a power difference between bully and victim (Olweus, 1993). Although the
definition of cyberbullying largely overlaps with the traditional definition of bullying
(Tokunaga, 2010), there are some notable differences. First, cyberbullying occurs via
the use of an electronic device—and rather than being limited to the workplace or
school, cyberbullies can target victims whether or not they are physically present
(Patchin & Hinduja, 2006). Further, in online environments, where physical strength
holds less sway, the power difference between bully and victim often is characterized by
other factors, such as technological skill or anonymity (Vandebosch & Van Cleemput,
2008). Given the unique aspects of cyberbullying incidents, Tokunaga’s(2010)
definition, derived from previous literature, was adopted for the present studies:
“Cyberbullying is any behavior performed through electronic or digital media by
individuals or groups that repeatedly communicates hostile or aggressive messages
intended to inflict harm or discomfort on others” (p. 278). As Tokunaga ( 2010) and
Slonje and Smith (2008) note, the “repetition” element of the definition takes a unique
form online, as a single message can be viewed repeatedly and have long-lasting effects.
Research suggests that several factors that distinguish cyberbullying from other
bullying episodes may affect a victim’s ability to receive support. These include the
presence of additional bystanders, the anonymity often afforded by mediated
environments, and the quality of the preexisting relationship between victims and
2 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
bystanders. The following section considers how each of these factors might affect
bystander intervention during cyberbullying episodes.
The Bystander Effect
The present studies contribute to well-established theoretical developments sur-
rounding the bystander effect (Darley & Latané, 1968). The bystander effect is an
extensively researched phenomenon (for a review, see Fischer et al., 2011) that posits
that individuals are less likely to exhibit prosocial behavior during an emergency
situation if other bystanders are present than if they are not. Specifically, as the
number of other people present increases, any individual bystander feels less
compelled to intervene. Research on the bystander effect has suggested several other
factors that may influence an individual bystander’s propensity to intervene,
including anonymity (Schwartz & Gottlieb, 1980) and relational closeness with the
victim (Levine & Crowther, 2008). The following sections extend the bystander effect
and other research on bullying behavior by offering a series of hypotheses relating to
bystander behavior in cyberbullying episodes.
Diffusion of responsibility
Several explanatory mechanisms have been offered for the bystander effect. For
instance, Latané and Darley (1970) proposed a five-step model in which a bystander
notices the event and undergoes a conscious decision-making process before deciding
whether to intervene. This approach suggests that bystander behavior is driven by an
evaluation of the relative costs and rewards associated with nonintervention (Latané &
Nida, 1981). Other work has emphasized a subconscious element of bystander
behavior: individuals are bombarded with frequent demands on their attention and
are therefore limited in their ability to consider whether or not to act (Milgram, 1970).
Overall, the most frequently invoked explanatory mechanism of the bystander
effect is the diffusion of responsibility, which takes an exchange perspective: the
presence of others lowers of psychological costs of nonintervention (Latané & Nida,
1981). One of the initial studies on bystander intervention found that, when viewing
a confederate suffering from an epileptic seizure, 85% of participants who thought
they were alone reported the seizure in a timely fashion, but only 31% of those who
thought there were other witnesses did so (Darley & Latané, 1968).
Importantly, the diffusion of responsibility effect is not based on only the physical
presence of others. For example, Markey (2000) posted requests for help in a public,
online message board. Message board participants were more likely to respond
quickly to the requests when there were fewer active posters on the board,
particularly when the bystander was not asked directly by name for help. Another
study found that the likelihood of helping in response to an email request declined
when the email message indicated that multiple recipients were contacted (Blair,
Foster Thompson, & Wuensch, 2005).
Bystander Intervention in Cyberbullying 3
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Research examining offline bystander intervention suggests that a diffusion of
responsibility effect might be especially salient in cyberbullying episodes. More
specifically, in a study in which the immediate danger of an offline situation was
manipulated, the effect held when the potential for immediate danger was low (Fischer,
Greitemeyer, Pollozek, & Frey, 2006). However, contrary to previous findings, when
the potential for immediate danger to the victim was high, individuals intervened at the
same rate regardless of whether they were alone or in the presence of other bystanders.
Online, the bully and victim are geographically distributed, which often means that the
threat of immediate danger is low. Hence, the presence of other bystanders might have
an especially important effect on intervention in online bullying episodes.
Visual anonymity
The bystander effect sometimes functions differently based on factors beyond the
presence of bystanders, such as the perceived anonymity of the observer. In their
extension of previous work on the bystander effect in an offline context, Schwartz
and Gottlieb (1980) manipulated the number of other bystanders and whether or not
the other bystanders and the victim were aware that the observer was present.
Consistent with their predictions, the number of bystanders interacted with
anonymity, such that individuals who were anonymous (i.e., the victim was not
aware of their presence) and who knew there were other bystanders had relatively
slower reaction times (Schwartz & Gottlieb, 1980). Research on mediated commun-
ication also has indicated that people’s identity is often anonymous in online
environments, and that the perception of anonymity affects the enactment of both
pro- and anti-social behavior (Joinson, 2001). Although 40–50% of cyberbullying
victims are aware of the identity of the perpetrator, the potential anonymity afforded
by mediated communication means that victims may not know who is targeting
them (Patchin & Hinduja, 2006). Similarly, because the potential number of
witnesses to cyberbullying incidents is so large, it is likely that individual bystanders
are not identifiable by either the victim or the bully.
The most widely researched model concerned with anonymity in online contexts is
the Social Identity Model of Deindividuation Effects (SIDE; Spears & Lea, 1994). In
SIDE, anonymity is labeled as deindividuation. It is defined as a state of decreased
self-awareness that leads to fewer adherences to societal and group norms. Although
SIDE has been used to predict behavior in small groups (e.g., Spears & Lea, 1994), the
design and function of new forms of technology in which cyberbullying might occur
limit the scope of the theory in those contexts. For instance, SNS such as Facebook
include pictures and detailed identifying information about participants. Hence,
when considering whether or not individuals might intervene in a bullying incident,
deindividuation, as defined by SIDE, is not likely to affect bystander behavior. Rather,
in line with the operationalization implemented by Schwartz and Gottlieb (1980),
anonymity should be defined in terms of whether or not bystanders believe they are
visible—that is, whether their presence is apparent to others.
4 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
This operationalization of anonymity (i.e., visual anonymity) is supported by
previous work on the bystander effect. Latané and Nida (1981) summarized research
on the diffusion of responsibility and showed that it is weaker when the victim can
see the bystander (but not vice versa) than when the bystander can see the victim
(but the victim cannot see the bystander). In the cyberbullying context, these findings
suggest that bystanders should be especially likely to respond to bullying when they
think the victim knows they are present and aware of the incident.
Relationship quality
Mediated communication often entails interaction with family, friends, and acquain-
tances. Indeed, on SNS, relational development often begins offline or face to face (FtF)
before individuals interact online (Ellison, Steinfeld, & Lampe, 2007). Consequently, it
is likely that the quality of the relationship matters when predicting whether individuals
intervene during a cyberbullying episode. People might be more likely to intervene and
help a close friend or family member than an acquaintance, for instance.
Some research on the bystander effect has examined the role of relationship quality
and status in bystander behavior. This research often takes a social identity
perspective by claiming that individuals are more apt to help members of their “ in-
group” versus an “out-group” (Levine & Crowther, 2008). For instance, when people
reported feeling close to a victim, they were more likely to exhibit helping behavior,
such as defending the victim, during FtF bullying incidents (Oh & Hazler, 2009).
Middle school-aged children were more likely to report assisting a victim of offline
bullying when they had positive attitudes toward the victim (Rigby & Johnson, 2006).
In research directly investigating the bystander effect, group- and friendship-related
variables were more important than group size when determining whether a
bystander intervened in a hypothetical violent incident (Levine & Crowther, 2008).
Relational quality is also likely to influence when someone provides social support to
a victim (Dunkel-Schetter & Skokan, 1990).
The Present Studies
The two current studies build upon the bystander effect and the aforementioned
research, which have considered the influence of the presence of other bystanders,
visual anonymity, and relational quality on bystander intervention and support
behaviors. Each of these factors is suggested to play a role in the responses of
bystanders to a cyberbullying incident.
The presence of other bystanders
The bystander effect and the diffusion of responsibility (Darley & Latané, 1968)
suggest that witnesses to cyberbullying should be less likely to act in a prosocial
manner when other bystanders are present. Although to our knowledge there is no
current typology of cyberbullying bystander behavior, research on bystander behavior
Bystander Intervention in Cyberbullying 5
Downloaded by [122.226.181.124] at 13:56 25 January 2016
in traditional bullying episodes notes that a bystander can undertake several actions,
including providing support directly to the victim (Matsunaga, 2010), actively
defending the victim by responding directly to the bully (Salmivalli & Voeten, 2004),
or simply standing by, acting as a passive bystander. The diffusion of responsibility
effect suggests either an increase or decrease in certain behaviors during a
cyberbullying episode, based on the perceived number of bystanders:
H1: The perceived number of bystanders is (a) negatively related to a given
bystander’s active defending behavior, (b) positively related to a bystander’s
passive observing behavior, and (c) negatively related to a bystander’s support
behavior.
Visual anonymity
Although visual anonymity in the online environment has been linked to hostile
communication, few studies have examined the relationship between visual
anonymity (or whether the bystander’s presence is apparent) and bystander/helping
behavior. In one study of FtF behavior, individuals who perceived they were
anonymous were less likely to offer help in emergency situations (Solomon, Solomon,
& Maiorca, 1982). Given this, it is possible that in an online environment, people will
be less likely to offer assistance or support when they perceive they are visually
anonymous.
H2: Perceived visual anonymity is (a) negatively related to a bystander’s active
defending behavior, (b) positively related to a bystander’s passive observing
behavior, and (c) negatively related to a bystander’s support behavior.
Other research shows that the bystander effect can operate differently based on
perceived anonymity (Schwartz & Gottlieb, 1981). This work suggests that the
diffusion of responsibility effect predicted in H1 should be especially strong when a
bystander perceives that others are aware of his or her presence. Hence, the above
hypotheses are examined alongside the potential moderating variable of perceived
visual anonymity:
H3: Perceived visual anonymity moderates the relationships between the number
of bystanders and (a) active defending behavior, (b) passive observing behavior,
and (c) support behavior such that individuals who perceive themselves as not
visible are less likely to defend the victim, more likely to passively observe, and less
likely to offer support when in the presence of other bystanders than individuals
who perceive themselves as visible.
Relationship quality
In bullying episodes that occur offline, bystanders are more likely to exhibit helping
behavior, such as defending the victim, when they feel a sense of closeness to the
bullying target (Oh & Hazler, 2009). Additionally, individuals are more likely to help
6 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
victims whom they perceive to be their friends (Levine & Crowther, 2008). Based on
these findings, the following predictions were made:
H4: The reported closeness between a bystander and a victim is (a) positively
related to a bystander’s active defending behavior, (b) negatively related to a
bystander’s passive observing behavior, and (c) positively related to a bystander’s
support behavior.
Study 1
Method
Data were collected from 265 undergraduate students enrolled in communication
courses at a large southwestern university. Recruitment materials invited participants
who could recall at least one event in which they witnessed someone being targeted
by hurtful actions, behaviors, and/or messages on Facebook. Of the total, 199 (75.1%)
were female and 66 (24.9%) were male. Participants’ ages ranged from 18 to 42 years
(M = 20.2; SD = 1.97). Most were Caucasian (n = 163, 61.7%), followed by Hispanic
(n=40, 15.2%), Asian (n = 27, 10.2%), African-American (n = 17, 6.3%), other (n =
7, 2.7%), Middle Eastern (n = 5, 1.9%), and Native American (n = 3, 1.1%).
Respondents were offered course extra credit for their participation, and were told
that their participation was completely voluntary. One participant indicated that he
or she responded incorrectly to the instructions so that individual’s data were
removed. Ten additional participants could not recall such an event. These
participants were invited to complete another portion of the survey (reported in a
separate study).
Procedure
Participants completed an online survey. Similar to previous studies (e.g., Oh &
Hazler, 2009) they were given a brief definition of cyberbullying and asked to recall
an experience within the previous six months in which someone they knew was
targeted on Facebook. Facebook was selected as the context for the study because it is
the most widely used online social network, with over one billion worldwide
members (www.facebook.com) and is therefore likely to result in the broadest
applicability of the results.
After participants received instructions for completing the survey, they were asked
to describe the cyberbullying event in as much detail as possible, explain what led up
to the hurtful online messages or actions, and describe what they said or did
following the incident. After recounting the episode, participants completed several
additional measures, described below. Table 1 displays the mean, standard deviation,
and Cronbach’s alpha for each variable.
Presence of additional bystanders. Respondents were asked to recall the number of
additional people who witnessed or viewed the bullying episode. Similar to Oh and
Bystander Intervention in Cyberbullying 7
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Table 1 Means, SD, and correlations among the variables (N = 254).
M (SD) α Bys Anon Vic close Defend Observe Em Sup Es Sup Net Sup CB Sex
Bys 107.30 (210.58) .82 –
Anon 4.02 (2.00) .95 .18** –
Vic close 4.81 (1.95) .98 −.28*** −.27*** –
Defend 4.25 (1.89) .77 −.27*** −.47*** .42*** –
Observe 2.97 (1.63) .84 .29*** .39*** −.49*** −.46*** –
Em Sup 3.41 (1.32) .94 −.18** −.31*** .77*** .52*** −.50*** −
Es Sup 3.36 (1.32) .97 −.14* −.31*** .71*** .48*** −.48*** .93*** –
Net Sup 2.99 (1.29) .96 −.07 −.36*** .61*** .41*** −.42*** .82*** .85*** –
CB 1.29 (.50) .89 .08 .00 −.07 −.07 .08 −.09 −.08 .01 –
Sex ––.05 .01 −.08 −.12 −.01 −.18** −.15* −.12 .14* –
Hurt 5.15 (1.38) .93 .09 −.09 .08 .17** −.05 .29*** .29*** .25*** −.20*** −.25***
Bys, number of bystanders; Anon, visual anonymity; Vic close, closeness with the victim; Defend, active defending behavior; Observe, passive observing behavior; Em Sup,
emotional support; Es Sup, esteem support; Net Sup, network support; CB, past experience with cyberbullying; Sex, sex (0 = female, 1 = male); Hurt, hurtfulness of cyberbullying
episode.
*p < .05, **p < .01, ***p < .001.
8 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Hazler (2009), participants were asked to estimate “Approximately how many other
people witnessed the bullying episode?” and “How many other people were aware
that the victim was being bullied?” The mean of the two items was used to
operationalize the number of bystanders. Mean responses ranged from 0 to 2,000
(Mdn = 32.50).
Visual anonymity. Visual anonymity was operationalized in terms of whether
bystanders believed that the victim knew of their presence. Because photos and
identifying information are easily available in many online contexts, previous
measures were not applicable. Thus, several items were created for this study.
Participants were asked if, during the incident, they felt that other people, including
the victim, the perpetrator, and/or other witnesses: “were aware of me,”“knew I was
there,”“recognized my presence,” and “could see that I was online.” Participants
rated their agreement with the items on a 7-point Likert-type scale (1 = strongly
disagree;7=strongly agree). Scores were reversed so that higher scores reflected
higher perceived visual anonymity.
Relationship/closeness with the victim. Participants completed a slightly adapted
version of the Vangelisti and Caughlin (1997) closeness measure. The items were
reworded to reflect all relationship types, rather than just romantic relationships.
Sample items include: “How important is your relationship with this person” and
“How close are you to this person?” Participants responded to the items on a 7-point
Likert-type scale (1 = not at all;7=very much).
Bystander behavior. Bystander behavior was measured with a modified version of
the Participant Role Questionnaire (PRQ; Salmivalli & Voeten, 2004). Items were
adapted to refer to an individual bullying episode, rather than behavior across
multiple bullying incidents. Additionally, item wording which suggests a school
environment was removed, and the word “bully” was replaced with “perpetrator.”
The present study utilized two subscales from the PRQ. The defender subscale was
used to measure active defending, and the outsider subscale was used to measure
passive observing. Sample items include: “I told the perpetrator to stop their
behavior” (defender), and “I stayed outside the situation” (outsider). Participants
responded based on a 7-point Likert-type scale (1 = strongly disagree;7=strongly
agree).
Social support. Because the PRQ does not fully capture the range of possible social
support behaviors, the present study assessed the emotional, esteem, and network
support components of Xu and Burleson’s(
2001) typology. These components have
been used in other studies of bullying behavior (e.g., Matsunaga, 2010). Items were
modified to address non-marital relationships and to indicate provided support
rather than received support. Participants indicated their responses on a 5-point
Likert-type scale (1 = not at all;5=a great deal). Sample items include: “Told them
that they are important to me and I feel close to them” (emotional support); “Tried
Bystander Intervention in Cyberbullying 9
Downloaded by [122.226.181.124] at 13:56 25 January 2016
to reduce their feelings of guilt about the situation” (esteem support); and
“Connected them with people whom they may turn to for help” (network support).
Personal experience with cyberbullying. As a control variable, respondents
completed a slightly altered version of the Electronic Bullying Questionnaire
(Kowalski & Limber, 2007). Sample items include: “I was bullied on a SNS (such
as Facebook),” and “I was bullied through a text message sent to my cell phone.”
Participants reported their responses on a 5-point Likert scale (1 = not at all;
5=several times a week).
Degree of hurt. The perceived hurtfulness of the cyberbullying incident was
measured with a series of semantic differential items originally developed by
Vangelisti and Young (2000). Three items were added to assess the severity of the
incident. Sample items include: “Was not at all hurtful –Was extremely hurtful” and
“Was not very severe–Was very severe.” Individuals were asked to evaluate the degree
of hurt from the perspective of the victim.
Results
The hypotheses were tested with a series of hierarchical regression analyses, which
allowed for the analysis of both linear predictions and interaction effects. Five separate
regressions were conducted, one for each dependent variable (DV) (i.e., active
defending, passive observing, emotional, esteem, and network support).
1
The
Tolerance (TOL) and Variance Inflation Factory (VIF) tests showed acceptable
collinearity between the predictor variables for each regression. Table 1 displays the
correlations between the study variables.
Effect of independent variables on active defending. Because there was a significant
positive correlation between degree of hurt and active defending behavior, degree of
hurt was entered in the first block as a control variable.
2
The second block included
the number of bystanders (H1a), perceived visual anonymity (H2a), and degree of
closeness (H4a) between the participant and the victim. The variables were centered
to mitigate multicollinearity. The final block included the interaction term, which
was created by multiplying the centered number of bystanders variable with the
centered visual anonymity variable (H3a). Table 2 displays the results of the
regression.
The overall F-test indicated that the model significantly predicted active defending
behavior. In line with the hypotheses, the number of bystanders (H1a) and perceived
visual anonymity (H2a) were negatively related to active defending behavior.
Closeness with the victim (H4a) was positively linked to active defending behavior.
However, contrary to one of the four predictions, the interaction between the number
of bystanders and perceived visual anonymity (H3a) did not significantly relate to
active defending.
10 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Effect of independent variables on passive observing. None of the potential control
variables was significantly correlated with the dependent variables (DVs), so the first
block included the number of bystanders (H1b), perceived visual anonymity (H2b),
and degree of closeness (H4b) between the participant and the victim. The variables
were centered. The final block included the interaction term, which was created by
multiplying the centered number of bystanders variable with the centered visual
anonymity variable (H3b). Table 3 displays the results of the regression.
The overall F-test indicated that the model significantly predicted passive observing
behavior. In line with the hypotheses, the number of bystanders (H1b) and perceived
visual anonymity (H2b) were positively related to passive observing behavior.
Closeness with the victim (H4b) was negatively associated with passive observing
behavior. However, contrary to H3b, the interaction between the number of bystanders
and perceived visual anonymity was not significantly related to passive observing.
Effect of independent variables on social support. Because they were significantly
(or nearly significantly) associated with the DVs, participants’ sex (dummy-coded),
Table 3 Hierarchical regressions predicting passive observing behavior (N = 254).
Passive observing
Predictor variables BSEB βΔR
2
Step 1
Number of bystanders .001 .00 .13* .32***
Visual anonymity .23 .04 .27***
Closeness with Vic. −.30 .05 −.38***
Step 2
Interaction between number of
bystanders and visual anonymity
.00 .00 .02 .00
Total R
2
= .33; adjusted R
2
= .32. F(4, 250) = 30.21, p < .001.
*p < .05, ***p < .001.
Table 2 Hierarchical regressions predicting active defending behavior (N = 254).
Active defending
Predictor variables BSEB βΔR
2
Step 1
Degree of hurt .23 .09 .17** .03
Step 2
Number of bystanders −.001 .00 −.14** .31***
Visual anonymity −.34 .05 −.36***
Closeness with Vic. .26 .05 .27***
Step 3
Interaction between the number
of bystanders and visual anonymity
.00 .00 .01 .00
Total R
2
= .34; adjusted R
2
= .33. F(5, 249) = 25.22, p < .001. **p < .01, ***p < .001.
Bystander Intervention in Cyberbullying 11
Downloaded by [122.226.181.124] at 13:56 25 January 2016
closeness to the bully, and the degree of hurt were entered in the first block as control
variables. The second block included the number of bystanders (H1c), perceived
visual anonymity (H2c), and the degree of closeness (H4c) between the participant
and the victim. The final block included the interaction term, which was created by
multiplying the centered number of bystanders variable with the centered visual
anonymity variable (H3c). Table 4 shows the results of the regressions.
The overall F-test for all three regressions indicated that the models significantly
predicted social support behaviors. In line with the hypotheses, perceived visual
anonymity (H2c) was negatively associated with emotional, esteem, and network
support, and closeness with the victim (H4c) was positively related to emotional,
esteem, and network support. The perceived number of bystanders (H1c) was not
significantly associated with emotional or esteem support. Also contrary to H1c, the
perceived number of bystanders was positively related to network support. The
interaction between the number of bystanders and perceived visual anonymity (H3c)
was not significantly associated with social support behaviors.
Discussion
Number of bystanders. Consistent with research on the bystander effect, the
number of bystanders to an online bullying incident was negatively related to
participants intervening to stop the incident. Furthermore, the number of bystanders
was positively associated with participants’ tendency to passively observe the
incident. Although researchers have suggested that bystanders to bullying episodes
may decrease the likelihood of bystander intervention (e.g., O’Connell et al., 1999),
no studies have directly investigated this prediction in the bullying or cyberbullying
context. The results of the current study provide support for the diffusion of
responsibility effect in cyberbullying episodes.
It is interesting to note that, other than a positive relationship with network
support, the number of bystanders was not associated with social support behavior.
Given that previous research on the bystander effect has not investigated social
support as a DV, it is possible that the theory does not extend into contexts in
which social support is a viable option. These results suggest that closeness to the
victim and visual anonymity, but not the number of bystanders, predicted
emotional, esteem, and network-related social support. Further, given the positive
link between the number of bystanders and network support, it is likely that people
who perceive that there are many bystanders also perceive a large number of
potential outlets for network support for the victim. Indeed, one of the key functions
of Facebook use is the development of social capital, or resources gained from social
relationships (Ellison et al., 2007). Perhaps participants associated a large number of
bystanders with an increased potential for utilizing that social capital to support the
victims.
Visual anonymity. As predicted, perceived visual anonymity was positively related
to passive observing behavior and negatively associated with social support and active
12 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Table 4 Hierarchical regressions predicting social support behaviors (N = 254).
Emotional support Esteem support Network support
Predictor variables BSEB βΔR
2
BSEB β Δ R
2
BSEB βΔR
2
Step 1
Participant sex −.16 .10 −.10
‡
.11*** −.10 .10 −.06 .11** −.07 .10 −.05 .07***
Closeness with Perp. −.10 .05 −.13* −.11 .05 −.13* –.07 .05 –.09
Degree of hurt .23 .06 .24*** .24 .06 .25*** .21 .06 .22**
Step 2
Number of bystanders .00 .00 .03 .57*** .001 .00 .05 .48*** .001 .00 .12* .40***
Visual anonymity −.08 .03 −.12** −.09 .03 −.13** −.15 .03 −.23***
Closeness with Vic. .49 .03 .73*** .45 .03 .66*** .38 .03 .57***
Step 3
Interaction b/w number of
bystanders and visual anonymity
.00 .00 .04 .001 .00 .00 .03 .001 .00 .00 .05 .001
Emotional support: total R
2
= .67; adjusted R
2
= .67. F(7, 247) = 70.99, p < .001; esteem support: total R
2
= .58; adjusted R
2
= .57. F(7, 247) = 47.56, p < .001; network support:
total R
2
= .45; adjusted R
2
=.44.F(7, 247) = 30.94, p < .001.
‡
p < .10, *p < .05, **p < .01, ***p < .001.
Bystander Intervention in Cyberbullying 13
Downloaded by [122.226.181.124] at 13:56 25 January 2016
defending behavior. However, visual anonymity did not moderate the effect of the
number of bystanders on any of the DVs. Instead, it had a direct effect, regardless of
how many bystanders were present. This finding contradicts the work of Schwartz
and Gottlieb (1980), who found that the effect of the number of bystanders on
helping behavior was moderated by whether or not the bystander perceived him- or
herself as visible to the perpetrator or victim. The methodology of the present study
differed from that study in two ways. First, the number of observers in the current
study was not tightly controlled. In the Schwartz and Gottlieb experiment, there were
either one or zero bystanders. In online environments such as Facebook, there are
often at least several other bystanders. Second, the bystander was usually well
acquainted with the victim. Schwartz and Gottlieb investigated bystander interven-
tion in a context in which the victim and the bystander were strangers. In the present
study, very few participants (17.8%) indicated that the victim was a stranger or
acquaintance.
Together, the results are consistent with prior research on anonymity and
deindividuation, which suggests that decreased self-awareness allows for less
adherence to societal standards, and thus results in anti-normative behavior (e.g.,
Sproull & Kiesler, 1986). Studies also indicate that the extent to which individuals
believe they are not visible to other online communicators (sometimes labeled as
invisibility) can lead to anti-normative behavior (Lapidot-Lefler & Barak, 2012).
Although active defending and social support are prosocial behaviors, a lack of
intervention (passive observing) could be seen as anti-normative, in that the
bystander chooses not to help a victim. Consider that social isolation can exacerbate
the negative social and emotional effects of cyberbullying (Newman, Holden, &
Delville, 2005). The current findings demonstrate that visual anonymity is positively
related to passive observing behavior, a potentially hurtful behavior which may
augment the deleterious effects of the bullying episode.
Relationship with the victim. As expected, closeness with the victim was associated
with more social support and active defending, and less passive observation of the
cyberbullying episode. This finding extends previous research (e.g., Levine &
Crowther, 2008) which found that bystander closeness with a victim was positively
related to helping behavior. Moreover, in offline bullying incidents, individuals who
feel close to a victim are more likely to intervene to defend or support the victim (Oh
& Hazler, 2009). The results of this study replicate these findings and extend the
research into the online realm, in which individuals are often connected to people
they know offline as well (Ellison et al., 2007).
Limitations and rationale for Study 2. Despite the support for many of the
hypotheses, a second study was warranted for two reasons. First, the initial study
utilized recall methods. Individuals’ memories may be biased such that events that
took place after the bullying episode (e.g., whether or not someone else intervened)
influenced participants’ responses (e.g., Stafford & Daly, 1984). Second, participation
in Study 1 was limited to individuals who could recall an event in which someone
14 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
they knew was targeted. Individuals tended to recall situations that involved victims
they were relatively close to. Given these limitations, the second study was designed
to assess individuals’ responses to a victim they were close to and one they were not
close to. The four hypotheses put forth in Study 1 were tested in this second study.
The design of Study 2, which employed hypothetical scenarios, is consistent with
previous experimental research on both the diffusion of responsibility effect (e.g.,
Darley & Latané, 1968) and bystander intervention (e.g., Gini, Albiero, Benelli, &
Altoè, 2008).
Study 2
Method
Pilot study. After a careful examination of the open-ended responses generated in
Study 1, three scenarios were created for possible use in the second study. Separate
versions were generated for men and women so the victim in the scenario would be
the same sex as the participant. To select the optimal scenario for use in Study 2, 42
participants (35 women, 7 men; aged 18–34, M = 21.19, SD = 2.93) were asked to
read the scenarios and rate them for believability (Kearney, Plax, Smith, & Sorenson,
1988), hurtfulness (Vangelisti & Young, 2000), and whether the scenario constituted
cyberbullying (Kowalski & Limber, 2007)
Three separate between-subjects one-way ANOVAs were conducted to compare
the scenarios with respect to their believability, degree of hurt, and whether they
constituted cyberbullying. The ANOVAs were not significant for degree of hurt or
whether the scenario constituted cyberbullying. Although the ANOVA was not
significant for believability, F(2, 41) = 2.36, p = .11, post hoc LSD comparisons
indicated that the believability rating of the “hacking” scenario (M = 5.69, SD = 1.12)
was slightly (but not significantly) higher than the believability of the “relationship
drama” (M = 5.00, SD = 1.07) and the “burn book” scenarios (M = 4.88, SD = 1.12).
Thus, the hacking scenario was selected as the optimal vignette for Study 2
(Appendix 1).
Participants. Participants in Study 2 consisted of 379 undergraduate students (260
women and 119 men) at a large southwestern university. They ranged in age from 18
to 50 years (M = 20.69, SD = 2.98). Most of the participants were Caucasian (n = 230,
60.4%), followed by Hispanic (n=63, 16.5%), Asian (n = 48, 12.6%), African-
American (n = 27, 7.1%), other (n = 7, 1.8%), Middle Eastern (n = 2, 0.5%), and
Native American (n = 1, 0.3%).
Procedure. Similar to previous research (e.g., Gini et al., 2008), respondents read a
scenario that described an online bullying incident. They then completed an online
survey in which they were randomly assigned to one of eight conditions. The three
central independent variables (IVs) investigated in Study 1 were examined: The
number of bystanders (many, few), visual anonymity (high, low), and relationship
with the victim (close, acquaintance) were manipulated (
Appendix 1). Eight versions
Bystander Intervention in Cyberbullying 15
Downloaded by [122.226.181.124] at 13:56 25 January 2016
of the scenario were created, one for each of the conditions in the 2 × 2 × 2 design.
Participants read scenarios in which the victim was of their same sex. The
presentation of scenarios was randomized, participation was balanced among
conditions, and respondents were required to remain on the page displaying the
scenario for at least 45 seconds. Participants were told to read the scenario carefully
and to imagine themselves as the observer in the situation.
The number of bystanders was manipulated by varying the number of Facebook
friends the victim had. To determine an ecologically valid manipulation for the
variable, the results of the first study were used to calculate what might constitute a
“high” and “low” number of friends for a college student. In Study 1, the mean
number of Facebook friends for participants was 1,031.94 (SD = 863.03). For the
scenarios in Study 2, one standard deviation was added to and subtracted from the
mean to create the manipulation—a victim with many friends/bystanders (1,900) and
a victim with few friends/bystanders (170). Visual anonymity was manipulated by
altering whether the participant was logged into Facebook chat, and therefore visible
to the victim/perpetrator. Closeness was manipulated by altering whether the victim
was a “good friend” or “acquaintance.”
Participants completed the same DV and control scales described in Study 1.
Potential control variables included participant sex, experience with cyberbullying
(M = 1.31, SD = .55, α = .93), degree of hurt (M = 5.07, SD = 1.42, α = .96), and the
number of Facebook friends (M = 931.41, SD = 712.87). Seventy-four participants
(19.9%) indicated that they had been cyberbullied in the last three months. Of those,
47% told someone about the cyberbullying incident. Additionally, 62 participants
(16.67%) indicated that they had cyberbullied someone else over the past three months.
Finally, 232 respondents (62.37%) reported that they had seen or heard about a
cyberbullying episode in the past three months. On average, participants reported
seeing 2.6 cyberbullying incidents over the past three months (Mdn =1;SD = 6.66).
DVs included the possible reactions to the bullying episode: active defending (M =
4.45, SD = 1.56, α = .85), passive observing (M = 3.44, SD = 1.40, α = .83), emotional
(M = 3.41, SD = .96, α = .92), esteem (M = 3.44, SD = .98, α = .94), and network
(M = 3.23, SD = 1.01, α = .94) support. Each of these measures was described in the
Methods section of Study 1. Finally, participants answered two open-ended items:
“What do you think the purpose of this survey was?”
and “What do you think this
survey was trying to study?” Based on their awareness of the study’s goals, two
individuals were removed from subsequent analyses.
Results
Manipulation check. To ensure the scenarios activated the expected differences
relating to number of bystanders, visual anonymity, and closeness, a series of
independent samples t-tests was conducted. Individuals in the high number of
bystanders condition perceived a higher number of bystanders/witnesses to the
incident (M = 312.34, SD = 406.46) than did those in the low number of bystanders
condition (M = 74.63, SD = 217.46), t(375) = 7.05, p < .001, d = .73. Additionally,
16 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
people in the high visual anonymity condition reported feeling more visually
anonymous (M = 5.31, SD = 1.50) than did people in the low visual anonymity
condition (M = 4.69, SD = 1.46), t(375) = 4.07, p < .001, d = .42. Finally, individuals
in the close friendship condition (M = 4.65, SD = 1.33) reported feeling closer to their
friends than did those in the acquaintance condition (M = 4.01, SD = 1.35),
t(375) = 4.62, p < .001, d = .48.
Hypothesis tests. The hypotheses were tested using a 2 × 2 × 2 multivariate
analysis of covariance (MANCOVA) with number of bystanders, visual anonymity,
and relationship as the between-subjects factors. Active defending, passive observing,
and the three types of social support were the DVs. Because they were correlated with
each of the DVs, degree of hurt and participant sex were included as covariates.
Bartlett’s test of sphericity was significant, χ
2
(14) = 1133.84, p < .001, indicating that
the DVs were interrelated, and thus the use of MANCOVA was appropriate.
However, Box’s M test indicated that homoscedacity could not be assumed, F(105,
166,061.24) = 1.613, p < .001, so multivariate results should be read with caution.
The MANCOVA revealed significant multivariate effects for the number of
bystanders, Wilks’ Λ = .97, F(5, 358) = 2.43, p < .05; visual anonymity, Wilks’ Λ =
.95, F(5, 358) = 4.02, p < .001; and closeness, Wilks’ Λ = .89, F(5, 358) = 5.45, p <
.001. Additionally, there was a significant three-way interaction between the three
IVs, Wilks’ Λ = .96, F(5, 358) = 2.50, p < .05. None of the two-way interactions was
significant. The univariate results revealed significant differences between each of the
conditions and many of the DVs. Table 5 displays the results of the ANCOVA for
each independent variable. Individuals in the low number of bystanders condition
reported higher levels of active defending and network support and lower levels of
passive observing than individuals in the high number of bystanders condition. The
groups did not differ in their use of emotional or esteem support. Thus, hypotheses
1a and 1b were confirmed, and hypothesis 1c was partially confirmed.
People in the low visual anonymity condition reported higher levels of active
defending, emotional, esteem, and network support and lower levels of passive
observing than did those in the high visual anonymity condition. Hypotheses 2a, 2b,
and 2c were confirmed.
Individuals in the close friend condition reported higher levels of active defending,
emotional support, and network support and higher levels of passive observing than
did individuals in the acquaintance condition. The groups did not differ in their use
of esteem support. Thus, hypotheses 4a and 4b were confirmed, and hypothesis 4c
was partially confirmed.
Of course, each of the above effects must be considered in light of a significant
three-way interaction. The predicted interaction between the number of bystanders
and visual anonymity (H3) was not significant, Wilks’ Λ = .99, F(5, 358) = .56, p =
.89, but there was a significant three-way interaction between the number of
bystanders, visual anonymity, and closeness. Univariate results indicated that the
interaction was significant for active defending, F(1, 372) = 8.64, p < .01, partial η
2
=
.02, but not the other DVs. An examination of the marginal means indicated that the
Bystander Intervention in Cyberbullying 17
Downloaded by [122.226.181.124] at 13:56 25 January 2016
interaction between visual anonymity and the number of bystanders operated
differently based on whether participants were in the close friend or acquaintance
condition (see Figure 1).
For acquaintances, the Tukey B post hoc test indicated that individuals who were
in the high number of bystanders and high visual anonymity condition (M = 3.48, SD
= 1.55) reported less active defending than did those in the other three conditions
Figure 1 Three-way interaction between number of bystanders, relational closeness, and
visual anonymity for active defending.
Table 5 Estimated marginal means (and standard errors) for IVs associated with DVs.
High Low Fdfpartial η
2
IV: Number of bystanders
Active defending 4.31 (.10) 4.65 (.10) 6.09* 1, 372 .02
Passive observing 3.72 (.10) 3.38 (.10) 6.31* 1, 372 .02
Emotional support 3.38 (.06) 3.45 (.06) 0.75 1, 372
Esteem support 3.41 (.06) 3.50 (.07) 0.81 1, 372
Network support 3.15 (.07) 3.32 (.07) 2.94
‡
1, 372 .01
IV: Visual anonymity
Active defending 4.22 (.10) 4.74 (.10) 14.33*** 1, 372 .04
Passive observing 3.74 (.10) 3.35 (.10) 8.51** 1, 372 .02
Emotional support 3.29 (.06) 3.54 (.06) 8.84** 1, 372 .02
Esteem support 3.32 (.07) 3.59 (.07) 8.77** 1, 372 .02
Network support 3.13 (.07) 3.34 (.07) 4.77* 1, 372 .01
IV: Closeness
Active defending 4.79 (.10) 4.17 (.10) 12.03*** 1, 372 .06
Passive observing 3.31 (.10) 3.78 (.09) 11.68** 1, 372 .03
Emotional support 3.53 (.06) 3.30 (.06) 7.24** 1, 372 .02
Esteem support 3.53 (.07) 3.38 (.06) 2.57 1, 372
Network support 3.32 (.07) 3.15 (.07) 3.40
‡
1, 372 .01
‡
p < .10, *p < .05, **p < .01, ***p < .001.
18 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
(high number of bystanders and low visual anonymity: M = 4.32, SD = 1.40; low
number of bystanders and low visual anonymity: M = 4.42, SD = 1.41; high number
of bystanders and low visual anonymity: M = 4.18, SD = 1.51). For close friends, the
Tukey B test showed that people who were in the low number of bystanders and low
visual anonymity condition (M = 5.75, SD = 1.16) reported more active defending
than did those in the other groups (high number of bystanders and low visual
anonymity: M = 4.66, SD = 1.45; high number of bystanders and high visual
anonymity: M = 4.43, SD = 1.58; high number of bystanders and low visual
anonymity: M = 4.43, SD = 1.55).
The results partially confirmed H3a: Visual anonymity moderated the effect of
number of bystanders. Individuals in the high visual anonymity scenario were less
likely to defend the victim in the presence of other bystanders than were individuals
in the non-anonymous condition, but only when they were given a scenario in which
the victim was an acquaintance. When the victim was a good friend, visual
anonymity also moderated the effect of number of bystanders, but people in the
low visual anonymity and low number of bystanders condition were more likely to
report intent to actively defend the victim than each of the other conditions.
General Discussion
Bystanders to cyberbullying incidents have the ability to attenuate the social and
mental anguish of victims (Matsunaga, 2010). Observers can act immediately to stop
the bullying incident or offer social support to the victim. Despite the positive
potential for bystander intervention in cyberbullying, no research has directly
examined the predictors of intervention in this context. The results of the two
current studies suggest support for the bystander effect. Specifically, the number of
bystanders present during the incident, the bystander’s perceived sense of visual
anonymity, and the relationship between the bystander and the victim each predict
bystander behavior to varying degrees.
Number of Bystanders, Visual Anonymity, and Relational Quality
Although both Study 1 and Study 2 indicated that the number of bystanders, visual
anonymity, and relational quality were linked to people’s responses to cyberbullying,
one major difference emerged between the two studies: When participants in Study 2
read a scenario in which they were acquaintances with the victim, the influence of the
number of bystanders was moderated by visual anonymity, such that individuals who
were in the anonymous condition and who read the scenario with a high number of
bystanders reported a lower likelihood of active defending behavior than did those
who read the scenario with a low number of bystanders. However, when people were
in the low visual anonymity condition, there was no difference in active defending
behavior based on whether they read the high number of bystanders or low number
of bystanders scenario. When individuals read a scenario in which they were close
friends with the victim, those who were also assigned to the low visual anonymity
Bystander Intervention in Cyberbullying 19
Downloaded by [122.226.181.124] at 13:56 25 January 2016
and low number of bystanders condition reported the highest likelihood of active
defending behavior. In other words, people were most likely to intervene when they
were close friends with the victim, they were identifiable as being online, and there
were fewer bystanders witnessing the incident.
The results of Study 2 extend the work of Schwartz and Gottlieb (1980), who
found an interaction between bystander visibility to the victim and the presence of
bystanders. In that study, the victim was an unknown stranger to the bystander. In
Study 2, the same moderating effect occurred, but only when the victim was an
acquaintance of the bystander, not a close friend. The methodology of the second
study, which required participants to consider a scenario in which they were not
close to a victim, likely explains why the interaction emerged in Study 2.
Overall, the present results appear to support the bystander effect and prior
research indicating that people are less likely to actively defend cyberbullying victims
in incidents in which there are a higher number of bystanders, especially when the
victim is an acquaintance and the bystander perceives that the victim and/or
perpetrator cannot see him or her.
Theoretical Implications and Predictors of Bystander Intervention
The other results were largely consistent between Study 1 and Study 2 and reflected
general support for the bystander effect and the diffusion of responsibility
mechanism in cyberbullying contexts, in that the number of bystanders was
negatively related to active defending behavior and positively related to passive
observing. In line with previous research, as the number of witnesses to an incident
increased, individual bystanders were less likely to help. Moreover, the results of both
studies support the effect of visual anonymity on bystander active defending, passive
observing, and social support behaviors. Finally, Study 2 replicated the results of
Study 1 relating to closeness: individuals who read a scenario in which they were a
“good friend” with the victim were more likely to actively defend and offer social
support to the victim, and less likely to passively observe the incident, than
individuals who read a scenario in which they were “acquaintances” with the victim.
In each of these studies, the closeness variable had the largest effect size, indicating
that the relationship between the bystander and the victim is a key determinant of
bystander behavior. Oh and Hazler (2009) demonstrated that closeness with the
victim of a bullying incident was associated with actively defending the victim, and
Rigby and Johnson (2006) found that positive attitudes toward a victim of offline
bullying were associated with supportive behavior by the bystander. Similarly, Levine
and Crowther (2008) established that friendship-related variables were more
important than group size in predicting bystander intervention during emergencies.
Consistent with these findings, the present studies revealed that closeness to the
victim often demonstrated the largest effect of the three IVs hypothesized to predict
intervention behaviors.
As individuals continue to associate online with people they also know offline, the
role of relationships in online environments is likely to grow in importance (Ellison
20 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
et al., 2007). The present studies examined only the viewpoint of a single bystander.
Although there are many circumstances in which a single bystander may decline to
intervene, other members of the victim’s social network may still decide to take
action. Thus, individuals with a larger online social network and more close friends
are likely to receive more support and defending than are those with few friends or
few close friends. Because social isolation exacerbates the effects of bullying (e.g.,
Newman et al., 2005), individuals who do not have a large social network, or those
who have few close friends, may not only be less likely to receive social support, but
will perhaps suffer more extreme emotional consequences. The results of the current
study highlight the importance of the social network in cyberbullying incidents.
Because bystanders are more likely to help their close friends, individuals with fewer
close friends may be less likely to receive support when they are bullied.
Although the data presented in these studies do not differentiate between the
underlying causal factors for the bystander effect, there are several potential
explanations. Most work on the bystander effect invokes an exchange perspective
(e.g., Latané & Darley, 1970 ), in which people are assumed to be conscious decision
makers who strategically weigh the relative costs and rewards of nonintervention. In
Study 2, individuals who were anonymous, in the presence of many other bystanders,
and only acquaintances with the victim may have perceived very few costs to not
intervening, especially when compared to someone who is a close friend with the
victim, visually identifiable, and one of few bystanders.
Or, as suggested by Milgram (1970), people may not have the capacity to make
fully informed bystander intervention decisions when faced with multiple competing
goals and decisions. Bounded rationality, which examines the actual factors that
individuals use to make choices in a complex world (Simon, 1955), suggests that
decision makers face too many demands on their attention to reasonably consider
every potential choice (March, 1994). For instance, in online environments, where
people are bombarded with large quantities of information, users often employ
cognitive shortcuts (i.e., heuristics) when assessing the credibility of information
(Metzger, Flanagin, & Medders, 2010). This may explain why the closeness variable
had the largest effect size in both studies. When individuals are limited in their
decision-making capacity, they may rely on intuitive judgments (Kahneman, 2003).
One such intuitive judgment might involve the importance of helping close friends
(Levine & Crowther, 2008). Although the current studies cannot directly address this
issue, future research should elucidate the underlying mechanisms of the bystander
effect in cyberbullying.
Practical Implications
As stated by Darley and Latané (1968) in their seminal paper on the bystander effect,
“If people understand the situational forces that can make them hesitate to intervene,
they may better overcome them” (p. 383). When individuals—either victims or
bystanders—are aware of the factors that contribute to nonintervention, they can
actively seek assistance or offer support. In one study, students who learned about the
Bystander Intervention in Cyberbullying 21
Downloaded by [122.226.181.124] at 13:56 25 January 2016
bystander effect in class were more likely to intervene in an emergency later
(Beaman, Barnes, Klentz, & McQuirk, 1978). Future research should consider if
education about the situational factors that abate online intervention (e.g., visual
anonymity, number of bystanders, and relational closeness) can increase interven-
tion. Given the large number of social connections individuals have in online
environments, individuals could be encouraged to intervene regardless of whether the
victim is a close friend or an acquaintance. Additionally, victims who are aware of
these issues may be more likely to seek the social support they need to recover from
the bullying incident. Both previous research (e.g., Kowalski et al., 2012) and the
descriptive data in Study 2 suggest that cyberbullying occurs relatively frequently
among college students, which points to the need for such education programs to
target college students, specifically. Practitioners should advise college students to be
aware of the issues under examination in the current research when explaining the
importance of bystander intervention.
Limitations and Future Directions
The current studies are not without their limitations. The results are based on self-
report data. Participants may have overestimated their tendency to defend victims of
cyberbullying due to a social desirability bias. One method for addressing this
concern would be to use behavioral DVs in a laboratory setting. Future research
could use mock-up social networking site profiles and offer participants an
opportunity to act in some way to protect a confederate.
Researchers have noted the relatively atheoretical nature of cyberbullying research
(e.g., Tokunaga, 2010). The present studies take an initial step in addressing this
concern by extending the literature on the bystander effect to cyberbullying. Future
research should consider other theories which may explain or predict the behavior of
cyberbullying bystanders. For instance, in hostile online contexts where victims and
bystanders are truly anonymous (e.g., online message boards like 4chan), researchers
can apply SIDE (Spears & Lea, 1994) to better understand the interplay among
deindividuation, group membership, contextual norms, and intervention behavior.
Related to this point, future studies should explore the potential mediating and
moderating variables related to decision-making in bystander intervention. Given
that a sense of personal responsibility is associated with the tendency of bystanders to
take action (Latanè & Darley, 1970), in cyberbullying incidents, a sense of personal
responsibility could mediate the relationship between the number of bystanders and
intervention behaviors. Or, if bystander intervention behavior can be explained by
bounded rationality, individuals under a large cognitive load may be less likely to
intervene because they cannot fully consider the consequences of their inaction.
Studies suggest that people under high cognitive load are less likely to respond in a
socially appropriate manner to in-person deviant behaviors (Fonseca, Brauer,
Moisuc, & Nugier, 2013). This research should be extended to further understand
the underlying explanatory mechanisms relating to bystander intervention during
cyberbullying incidents.
22 N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Because there is currently no typology developed specifically for cyber bystanders,
the present studies adapted the measures developed by Salmivalli and Voeten (2004)
and Xu and Burleson (2001) to assess offline intervention and support behavior.
Future research should build on the results of the present studies by further
investigating which bystander behaviors are especially salient in the online
environment. Moreover, research should elucidate context-specific behaviors (such
as flagging an offending post via Facebook using a built-in reporting mechanism). An
unstated assumption of the present studies is that bystander intervention is a
prosocial behavior. However, scholars have noted the propensity for bystanders to
“join in” on the bullying and become perpetrators (e.g., Salmivalli, 1999). Future
research should consider this bystander role and explore its correlates in the
cyberbullying context.
Conclusion
The findings presented in this research suggest that several variables, including the
perceived number of bystanders, visual anonymity, and relational closeness, are
linked to bystanders’ propensity to intervene during a cyberbullying incident. The
theoretical and practical implications of these results are wide-ranging. First, these
studies extend a long line of work on the bystander effect and the diffusion of
responsibility into the cyberbullying context. Second, they utilized a novel but
conceptually necessary operationalization of anonymity (visual anonymity), which
more closely matches the experiences of modern communication technology, such as
Facebook. Furthermore, closeness with a cyberbullying victim was a key predictor in
determining bystander intervention, an important finding given the propensity for
individuals to communicate online with people they also know in offline environ-
ments. Overall, examining the variables expected to influence bystander intervention
in cyberbullying provides further insight into the interplay of technology, relation-
ships, and bullying behavior.
Acknowledgments
The authors would like to thank Jorge Peña, René Dailey, Matthew McGlone, and Timothy Loving
for their comments on a previous version of this manuscript, as well as the editor, Dr Kory Floyd,
and the two anonymous reviewers for their thoughtful comments. The authors would also like to
thank Emma Ourston and Garrett Villarreal for their assistance with data collection and coding.
Notes
[1] Because of the potential for inflated Type-1 error inherent in the use of multiple tests, the
false discovery rate (FDR) procedure developed by Benjamini and Hochberg (1995) and
described by Matsunaga (2010) was applied to the regressions. Results indicated that the
significance level for each of the beta values was below the corrected significance level
derived using the FDR procedure.
[2] Because detecting interaction effects in field research can be difficult, we included only
significant controls. However, the significance level for all variables in the active defending
and passive observing regressions remained the same when including all control variables
Bystander Intervention in Cyberbullying 23
Downloaded by [122.226.181.124] at 13:56 25 January 2016
from the social support regressions (i.e., participant sex, degree of hurt, and closeness to the
perpetrator). Similarly, the significance level for all variables in the social support regressions
remained the same when removing all non-significant controls. All beta values either
remained the same or changed by less than .01.
References
Beaman, A., Barnes, P. J., Klentz, B., & McQuirk, B. (1978). Increasing helping rates through
information dissemination: Teaching pays. Personality and Social Psychology Bulletin, 4,
406–411. doi: 10.1177/014616727800400309
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and
powerful approach to multiple testing. Journal of the Royal Statistical Society B, 57, 289–300.
doi:10.2307/2346101
Blair, C. A., Foster Thompson, L., & Wuensch, K. L. (2005). Electronic helping behavior: The
virtual presence of others makes a difference. Basic and Applied Social Psychology, 27, 171–
178. doi: 10.1207/s15324834basp2702_8
Darley, J. M., & Latané, B. (1968). Bystander intervention in emergencies: Diffusion of responsibility.
Journal of Personality and Social Psychology, 8, 377–383. doi:10.1037/h0025589
Dunkel-Schetter, C., & Skokan, L. A. (1990). Determinants of social support provision in personal
relationships. Journal of Social and Personal Relationships, 7, 437–450. doi:10.1177/026540
7590074002
Ellison, N. B., Steinfield, C., & Lampe, C. (2007). The benefits of Facebook “friends:” Social capital
and college students’ use of online social network sites. Journal of Computer-Mediated
Communication, 12, 1143–1168. doi:10.1111/j.1083-6101.2007.00367.x
Fischer, P., Greitemeyer, T., Pollozek, F., & Frey, D. (2006). The unresponsive bystander: Are
bystanders more responsive in dangerous emergencies? European Journal of Social
Psychology, 36, 267–278. doi:10.1002/ejsp.297
Fischer, P., Krueger, J. I., Greitemeyer, T., Vogrincic, C., Kastenmüller, A., Frey, D., ... & Kainbacher,
M. (2011). The bystander-effect: A meta-analytic review on bystander intervention in
dangerous and non-dangerous emergencies. Psychological Bulletin, 137, 517. doi:10.1037/
a0023304
Fonseca, A., Brauer, M., Moisuc, A., & Nugier, A. (2013). Cognitive load causes people to react
ineffectively to others’ norm transgressions. Journal of Applied Social Psychology, 43, 1518–
1527. doi:10.1111/jasp.12145
Gini, G., Albiero, P., Benelli, B., & Altoè, G. (2008). Determinants of adolescents’ active defending
and passive bystanding behavior in bullying. Journal of Adolescence, 31,93–
105. doi:10.1016/
j.adolescence.2007.05.002
Hampton, K. N., Goulet, L. S., Marlow, C., & Rainie, L. (2011). Why most Facebook users get more
than they give. Washington, DC: Pew Internet and American Life Project. Retrieved from
http://www.pewinternet.org/2012/02/03/why-most-facebook-users-get-more-than-they-give/
Joinson, A. N. (2001). Self-disclosure in computer-mediated communication: The role of self-
awareness and visual anonymity. European Journal of Social Psychology, 31, 177–192. doi:10.
1002/ejsp.36
Kahneman, D. (2003). Maps of bounded rationality: Psychology for behavioral economics.
American Economic Review, 93, 1449–1475. doi:10.1257/000282803322655392
Kowalski, R. M., Giumetti, G. W., Schroeder, A. N., & Reese, H. H. (2012). Cyber bullying amongst
college students: Evidence from multiple domains of college life. In L. A. Wankel & C.
Wankel (Eds.), Misbehavior online in higher education (pp. 293–321). Bingley: Emerald
Group Publishing.
Kowalski, R. M., & Limber, S. P. (2007). Electronic bullying among middle school students. Journal
of Adolescent Health, 41, S22–S30. doi:10.1016/j.jadohealth.2007.08.017
Lapidot-Lefler, N., & Barak, A. (2012). Effects of anonymity, invisibility, and lack of eye-contact on
toxic online disinhibition. Computers in Human Behavior, 28, 434–443. doi:10.1016/j.chb.
2011.10.014
Latané, B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn’t he help? New York,
NY: Appleton-Century-Crofts.
24
N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Latané, B., & Nida, S. (1981). Ten years of research on group size and helping. Psychological
Bulletin, 89, 308–324. doi:10.1037/0033-2909.89.2.308
Levine, M., & Crowther, S. (2008). The responsive bystander: How social group membership and
group size can encourage as well as inhibit bystander intervention. Journal of Personality
and Social Psychology, 95, 1429–1439. doi:10.1037/a0012634
March, J. G. (1994). A primer on decision making: How decisions happen. New York, NY: The Free
Press.
Markey, P. (2000). Bystander intervention in computer-mediated communication. Computers in
Human Behavior, 16, 183–188. doi:10.1016/S0747-5632(99)00056-4
Matsunaga, M. (2010). Testing a mediational model of bullied victims’ evaluation of received
support and post-bullying adaptation: A Japan-U.S. cross-cultural comparison. Commun-
ication Monographs, 77, 312–340. doi:10.1080/03637751003758235
Metzger, M. J., Flanagin, A. J., & Medders, R. B. (2010). Social and heuristic approaches to
credibility evaluation online. Journal of Communication, 60, 413–439. doi:10.1111/j.1460-
2466.2010.01488.x
Milgram, S. (1970). The experience of living in cities. Science, 167, 1461–1468. doi:10.1126/
science.167.3924.1461
Newman, M. L., Holden, G. W., & Delville, Y. (2005). Isolation and the stress of being bullied.
Journal of Adolescence, 28, 343–357. doi:10.1016/j.adolescence.2004.08.002
O’Connell, P., Pepler, D., & Craig, W. (1999). Peer involvement in bullying: Insights and challenges
for intervention. Journal of Adolescence, 22, 437–452. doi:10.1006/jado.1999.0238
Oh, I., & Hazler, R. J. (2009). Contributions of personal and situational factors to bystanders’
reactions to school bullying. School Psychology International , 30, 291 –310. doi:10.1177/
0143034309106499
Olweus, D. (1993). Victimization by peers: Antecedents and long-term outcomes. In K. H. Rubin &
J. B. Asendorpf (Eds.), Social withdrawal, inhibition, and shyness in childhood (pp. 315–341).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Patchin, J. W., & Hinduja, S. (2006). Bullies move beyond the schoolyard: A preliminary look at
cyberbullying. Youth Violence and Juvenile Justice, 4, 148–169. doi:10.1177/1541204006
286288
Rigby, K., & Johnson, B. (2006). Expressed readiness of Australian schoolchildren to act as
bystanders in support of children who are being bullied. Educational Psychology, 26, 425–
440. doi: 10.1080/01443410500342047
Salmivalli, C. (1999). Participant role approach to school bullying: Implications for interventions.
Journal of Adolescence, 22, 453–459. doi:10.1006/jado.1999.0239
Salmivalli, C., & Voeten, M. (2004). Connections between attitudes, group norms, and behaviour in
bullying situations. International Journal of Behavioral Development, 28, 246–258. doi:10.
1080/01650250344000488
Schwartz, S. H., & Gottlieb, A. (1980). Bystander anonymity and reactions to emergencies. Journal
of Personality and Social Psychology, 39, 418–430. doi:10.1037/0022-3514.39.3.418
Simon, H. (1955). A behavioural model of rational choice. Quarterly Journal of Economics, 69,99–
118. doi: 10.2307/1884852
Slonje, R., & Smith, P. K. (2008). Cyberbullying: Another main type of bullying? Scandinavian
Journal of Psychology, 49, 147–154. doi:10.1111/j.1467-9450.2007.00611.x
Smith, A., Rainie, L., & Zickuhr, K. (2011). College students and technology. Washington, DC: Pew
Internet and American Life Project. Retrieved from http://pewinternet.org/Reports/2011/
College-students-and-technology/Report.aspx
Solomon, L. Z., Solomon, H., & Maiorca, J. (1982). The effects of bystander’ s anonymity,
situational ambiguity, and victim’s status on helping. The Journal of Social Psychology, 117,
285–294. doi: 10.1080/00224545.1982.9713438
Spears, R., & Lea, M. (1994). Panacea or panopticon? The hidden power in computer-mediated
communication. Communication Research, 21, 427–459. doi:10.1177/009365094021004001
Sproull, L., & Kiesler, S. (1986). Reducing social context cues: Electronic mail in organizational
communication. Management Science, 32, 1492–1512. doi:10.1287/mnsc.32.11.1492
Stafford, L., & Daly, J. A. (1984). Conversational memory: The effects of recall mode and memory
expectancies on remembrances of natural conversations. Human Communication Research,
10, 379–402. doi:10.1111/j.1468-2958.1984.tb00024.x
Bystander Intervention in Cyberbullying 25
Downloaded by [122.226.181.124] at 13:56 25 January 2016
Tokunaga, R. S. (2010). Following you home from school: A critical review and synthesis of
research on cyberbullying victimization. Computers in Human Behavior, 26, 277–287.
doi:10.1016/j.chb.2009.11.014
Vandebosch, H., & Van Cleemput, K. (2008). Defining cyberbullying: A qualitative research into
the perceptions of youngsters. CyberPsychology & Behavior, 11, 499–503. doi:10.1089/
cpb.2007.0042
Vangelisti, A. L., & Caughlin, J. P. (1997). Revealing family secrets: The influence of topic, function,
and relationships. Journal of Social and Personal Relationships, 14, 679–707. doi:10.1177/
0265407597145006
Vangelisti, A. L., & Young, S. L. (2000). When words hurt: The effects of perceived intentionality
on interpersonal relationships. Journal of Social and Personal Relationships, 17, 393–424.
doi:10.1177/0265407500173005
Xu, Y., & Burleson, B. R. (2001). Effects of sex, culture, and support type on perceptions of spousal
social support: An assessment of the “support gap” hypothesis in early marriage. Human
Communication Research, 27, 535–566. doi:10.1111/j.1468-2958.2001.tb00792.x
Appendix 1. Study 2 Scenario
Female
You log in to your Facebook account, and you notice that your good friend Jane had her profile
hacked. Someone hacked into Jane’s profile and posted embarrassing, inappropriate pictures and a
status update that says, “I’m a slut.” You look down at the chat box, and notice that you are logged
into chat and Jane is online. Jane has only about 170 Facebook friends, so it is not likely that many
other people have noticed the status updates and pictures.
Male
You log in to your Facebook account, and you notice that your good friend Steve had his profile
hacked. Someone hacked into Steve’s profile and posted embarrassing, inappropriate pictures and a
status update that says, “I’m a pussy.” You look down at the chat box, and notice that you are
logged into chat and Steve is online. Steve has only about 170 Facebook friends, so it is not likely
that many other people have noticed the status updates and pictures.
Note: The above versions represent the close relationship, low number of bystanders, not visually
anonymous condition.
26
N. Brody and A.L. Vangelisti
Downloaded by [122.226.181.124] at 13:56 25 January 2016