Cyber Bullying Final Paper
draggon.flye
cyberbullying_bartlet.pdf
Psychology of Popular Media Culture
Anonymously Hurting Others Online: The Effect of Anonymity on Cyberbullying Frequency Christopher P. Barlett Online First Publication, November 4, 2013. doi: 10.1037/a0034335
CITATION Barlett, C. P. (2013, November 4). Anonymously Hurting Others Online: The Effect of Anonymity on Cyberbullying Frequency. Psychology of Popular Media Culture. Advance online publication. doi: 10.1037/a0034335
Anonymously Hurting Others Online: The Effect of Anonymity on Cyberbullying Frequency
Christopher P. Barlett Gettysburg College
Cyberbullying (CB) has recently become a significant issue in today’s society. Given the myriad negative consequences to the cyber-victim, it is important to determine what variables predict CB frequency. Based on broader psychological and communication theory, I predict that anonymity will (a) directly predict CB frequency, (b) moderate the relation between positive attitudes toward CB and CB frequency, and (c) mediate the relation between instant messaging frequency and CB behavior. Participants (N � 181) completed measures designed to assess these aforementioned variables. Results showed that positive attitudes toward CB, CB reinforcement, and anonymity strongly predicted CB frequency. Furthermore, moderation tests confirmed that CB was highest when positive attitudes and anonymity were both high. Finally, mediation tests revealed anonymity mediated the relation between instant messaging frequency and CB behav- ior. These results are important at elucidating what variables predict CB to hopefully inform intervention efforts aimed at reducing CB.
Keywords: cyberbullying, anonymity, attitudes
Violent and aggressive behaviors are not new behavioral phenomena. However, the method by which aggressive acts are delivered has changed with increased technology. For exam- ple, Anderson and Huesmann (2003) stated that violent behaviors rose dramatically with the ac- cessibility of handguns. In today’s technology- based culture, individuals are now turning to electronic means (e.g., Internet, texting) to harm others, termed cyberbullying (CB) (defined as, “. . . the use of information and communication technologies such as e-mail, cell phone and pager text messages, instant messaging, defam- atory personal Web sites, and defamatory online personal polling Web sites to support deliberate, repeated, and hostile behavior by an individual or group, that is intended to harm others” [cited in Li, 2007, p. 1779]). CB is a serious societal issue owing to the extensive harm it can cause
the victim. Indeed, research has shown that those who are cyber-victimized are at risk for heightened negative psychological (fearful [Be- ran & Li, 2005], depressed [Patchin & Hinduja, 2006], suicide ideation [Hinduja & Patchin, 2010], and anger [Beran & Li, 2005]) and be- havioral (drug abuse [Hinduja & Patchin, 2008] and poor school grades [Beran & Li, 2007]) outcomes.
To date, the majority of the CB literature has focused on the victim. Although important, a better understanding of the predictors of CB is needed to not only understand why people harm others using electronic methods, but also to inform interventions aimed at reducing CB. Relative to the literature focusing on the cyber- victim, there is a paucity of research testing what factors enhance or reduce the likelihood of CB. Consistent with broader aggression theory, research has shown positive correlations be- tween CB frequency and traditional bullying frequency (Smith et al., 2008), normative ag- gressive beliefs (Ang, Tan, & Mansor, 2011), and low empathy (Ang & Goh, 2010; Steffgen, Konig, Pfetsch, & Melzer, 2011). However, many additional factors can influence CB, such as one’s attitudes toward CB, reinforcement, and identification. The objective of the current study is to elucidate on these aforementioned
Christopher P. Barlett, Department of Psychology, Get- tysburg College.
Correspondence concerning this article should be ad- dressed to Christopher P. Barlett, Department of Psychol- ogy, Gettysburg College, Campus Box 0407, 300 North Washington Street, Gettysburg, PA 17325. E-mail: [email protected]
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
Psychology of Popular Media Culture © 2013 American Psychological Association 2013, Vol. 2, No. 4, 000 2160-4134/13/$12.00 DOI: 10.1037/a0034335
1
factors. In doing so, the results of this study may be important in indentifying the predictors of CB to guide intervention efforts at reducing this harmful “newer” form of aggression.
The Role of Anonymity
In the online world, an aggressor will not be face-to-face with their victim. This would imply that anonymity should increase CB behavior; however, research has shown that traditional bullying is more common than CB (Olweus, 2012), and traditional bullies can harm others using covert aggressive tactics (e.g., gossiping, rumor spreading). Therefore, it is possible that anonymity can increase both traditional bully- ing and CB. Indeed, literature from broader social psychological and communication theo- ries have shown that anonymity is related to aggressive behavior (Diener, 1976; see also An- derson & Bushman, 1997) owing to deindividu- ation processes. However, the focus of the cur- rent study was to understand what variables may predict CB behavior, and there is a paucity of research empirically testing anonymity’s contribution.
In the mediated world, anonymity is pro- nounced because a) the aggressor is not as iden- tifiable and can use fake usernames, b) the bully does not need to have a previous relationship with the victim, and c) no physical scars or marks are inflicted on the victim from the bully. In other words, the aggressor’s anonymity may enhance the frequency of which CB (vs. tradi- tional bullying) occurs. It should be noted that anonymity is not a necessary condition for CB. It is likely the case that the cyber-victim also knows the cyberbully; however, that is not al- ways the case. Furthermore, a cyberbully may not truly be anonymous. Phone numbers can be traced and IP addresses can be identified, lead- ing to low anonymity. However, feeling anon- ymous and being anonymous are not identical and even if the cyber-victim can identify their aggressor, the bully may still feel anonymous, which may predict CB frequency.
Surveys using adolescent samples have indi- cated that 29% of adolescent cyber-victims could not identify their aggressor (Patchin & Hinduja, 2006). Furthermore, in 2011, the Cen- ter for Disease Control (CDC, 2011) conducted a survey of adolescents who reported being cyberbullied. Their report indicated that of
those cyberbullied, 67% of the CB occurred on instant messenger (IM), whereas only 17% oc- curred via text messaging and 21% using e-mail (percentages were not mutually exclusive). IM may enhance anonymity to the aggressor through the use of handles (made up electronic names) rather than one’s actual name, compared with texting or e-mailing where one’s name (and phone number/e-mail address) are easily accessible, decreasing the likelihood of ano- nymity. Although, it is hypothesized that IM frequency will better predict CB than texting or e-mail frequency, a cyberbully can create a fake e-mail address to attack others online. However, using IM as a means to aggress may still feel anonymous relative to e-mail frequency. No research has explicitly tested this claim and will be tested in the current study.
Theoretical Predictors of CB
To date, several researchers have suggested that one defining characteristic that differenti- ates CB from traditional bullying is enhanced anonymity for the aggressor (Li, 2007; Smith et al., 2008; Vandebosch & Van Cleemput, 2008). Indeed, Herring (2001) stated that the perceived anonymity afforded in the mediated world in- creases the likelihood of aggressive and hostile acts, as evident by the research showing that the majority of cyber-victims do not know their aggressor (Kowalski & Limber, 2007). Despite this wealth of research and speculation regard- ing anonymity’s influence on CB, there is a paucity of research empirically testing these relations and few theoretical postulations to make informed predictions to suggest how an- onymity is related to CB.
Recently, Barlett and Gentile (2012) tested a theoretical model that focused on the distal learning processes involved in CB. Drawing on broader social–cognitive learning theories of aggression (e.g., General Aggression Model; Anderson & Bushman, 2002), their model pos- its that each successful positively reinforced instance of CB is a learning trial. Continued learning is related to the development of learned knowledge structures that predict CB. In other words, Barlett and Gentile (2012) posited that when one continues to cyberbully another, the aggressor will ascertain certain knowledge re- garding the outcomes. Barlett and Gentile (2012) postulated and found evidence to sug-
2 BARLETT
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
gest that continued and successful learned CB is related to the development of two specific learned associations: anonymity and the lack of power differential. The former construct (ano- nymity) is the focus of the current research. According to Barlett and Gentile (2012), cyber- bullies believe that they are perceived as anon- ymous in the mediated world after they success- fully harm another several times. Also, in their model, Barlett and Gentile (2012) found that perceived anonymity directly predicts both CB frequency and positive attitudes toward CB. Mediation tests of their model showed that anonymity is related to CB frequency because of positive attitudes toward CB (the mediator). In other words, perceived anonym- ity predicts the development of positive atti- tudes toward CB, which in turn predicts CB frequency (see Figure 1).
As seen in Figure 1, anonymity can serve as both a mediator and moderator when predicting CB. Notably, when learning is the primary in- dependent variable, anonymity can serve as a mediator in the path to CB frequency. However, anonymity can also serve as a moderator vari- able in the relation between positive attitudes toward CB and CB frequency. Thus, depending on what part of the model one is addressing, anonymity serves different roles. The current study will test both the mediating role of ano- nymity in the relation between media usage and CB frequency, as well as the moderating role of anonymity in the relation between positive atti- tudes toward CB and CB frequency. The inves- tigation of how anonymity is related to CB frequency is of theoretical importance. Barlett and Gentile (2012) showed that anonymity is an important predictor in CB frequency without testing whether anonymity mediates or moder-
ates the relations predicted in the Barlett and Gentile (2012) model. This is the purpose of the current study.
Overview of the Current Study
The purposes of the study was to further test and validate the Barlett and Gentile (2012) model focusing on how anonymity is related to CB frequency. It is hypothesized that instant messaging (rather than e-mail) would be related to CB and anonymity would mediate this rela- tionship. Furthermore, anonymity should mod- erate the relation between positive attitudes to- ward CB and CB frequency.
Method
Participants
Data were collected in the Fall of 2010. One hundred and eighty-one (57% female) under- graduate students from a large Midwestern Uni- versity participated in the study for partial course credit in their psychology classes. The average age of the sample was 19.47 (SD � 1.56) years. The majority (79%) were Cauca- sian. The majority of participants were in their first or second year of undergraduate education (80%).
Materials and Procedure
On completion of the informed consent, par- ticipants completed the following question- naires and then were thanked and fully de- briefed.
CB frequency. The past research in the CB literature has used dozens of different question-
Anonymity
Posi�ve A�tudes towards
Cyberbullying
Cyberbullying Frequency
Learned Cyberbullying
Behaviors
Figure 1. Extensions of the Barlett and Gentile (2012) model.
3CYBERBULLYING
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
naires to assess CB frequency. According to Rivers and Noret (2010), such diversity in mea- sures is one (of several) reason why CB fre- quency percentages vary from study to study. The diverse numbers of measures each differ in important respects. For instance, some research- ers use a Likert-like rating scale (e.g., Ybarra, Diener-West, & Leaf, 2007), whereas others use dichotomous estimates of CB (e.g., Li, 2007). Furthermore, some researchers define CB to the participant (e.g., Li, 2007), whereas others do not (e.g., Ang & Goh, 2010). Overall, there is no current “gold standard” measure of CB fre- quency. The CB frequency questionnaire used in the current study (see below for details) was used because it measured CB behavior ade- quately; it was based on a valid measure of media exposure of TV, movie, and video game violence (Gentile, Lynch, Linder, & Walsh, 2004). The CB scale used in the current study had participants indicate their level of CB with- out being explicit regarding what was being measured, akin to other measures that did not use definitions (although participants could probably infer what was being measured).
For the purposes of the current research, CB frequency was calculated using an adapted ver- sion of the Media Habits Questionnaire (Gentile et al., 2004). Participants were instructed to list their three favorite Web sites. For each Web site they listed, they rated it on how often they visited the Web site, how often they write mean messages to others on this Web site, and how often they posted mean comments about others on the Web site on a 1 (rarely) to 5 (all the time) rating scale.1 If participants only listed two, rather than three, favorite Web sites, for exam- ple, then the items designed for the third Web site were assigned a value of zero (see Anderson & Dill, 2000). The frequency rating was multi- plied by the write mean messages rating and then averaged across all three Web sites to get a CB via writing mean messages estimate. The same formula was applied to the posting mean comments ratings. These two estimates (posting mean messages and writing mean messages) were averaged to produce an estimate of CB via the Internet (� � .87). Higher scores indicate more CB. Akin to other measures of CB, the data for this were positively skewed; however, given the sample size, no corrections (e.g., log transforming) were done.
Traditional bullying. The Ybarra et al. (2007) traditional bullying scale was used to assess frequency of traditional, or face-to-face, bullying. This is a three-item questionnaire that asks participants how often they aggressed against others using a 1 (never) to 6 (everyday/ almost everyday) rating scale (� � .75). A sample item includes, “Made rude comments or mean comments to anyone.” The three items were summed such that higher scores indicated higher reported frequency of face- to-face bullying.
Positive Attitudes Toward CB. The re- searcher-created Positive Attitudes toward Cy- berbullying questionnaire (Barlett & Gentile, 2012) was used (� � .95). This is a 20-item questionnaire that asks participants their level of agreement with the items on a 1 (strongly disagree) to 5 (strongly agree) rating scale. A sample item includes, “It is OK to bully others online if they deserve it.” These items were summed, such that higher scores indicate more positive attitudes toward CB.
Anonymity. The researcher-created Atti- tudes toward Anonymity questionnaire (Barlett & Gentile, 2012) was used (� � .71). This is a five-item questionnaire that asks participants their level of agreement with the items on a 1 (strongly disagree) to 5 (strongly agree) rating scale. A sample item includes, “I feel comfort- able sending mean text messages or e-mails to anybody no matter if I know them or not.”
1 The original version of this questionnaire asks partici- pants to list their three favorite movies, TV shows, and video games. Thus, the adapted version used here only asked participants to list their three favorite Web sites. The term “Web site” was not defined for participants; however, that is a trivial absence. The primary focus for using this questionnaire was to ascertain how frequently participants visited the Web site and how often they cyberbullied using this Web site. This measure afforded the researchers a wide variety of options. For instance, a score of 0 could indicate that a) the Web site participant’s visited does not allow for CB (e.g., www.iastate.edu), or b) the Web site participant’s visited does afford possible CB opportunities (e.g., www .Facebook.com; the most favored Web site in this sample), but no such behavior is reported. Conversely, if the favored Web site does afford CB opportunities, then it is likely through these two options (posting mean comments or writ- ing mean messages to others). Other methods of CB can exist on several Web sites that are not measured here (e.g., social exclusion, posting videos, etc.); however, based on the definition of CB used in this study and knowledge that social networking Web sites would most likely be favored over other Web sites, this was a valid measure of CB.
4 BARLETT
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
These items were summed, such that higher scores indicate more positive attitudes toward anonymity in CB.
Demographic questionnaire. A demo- graphic questionnaire was used to assess sex, ethnicity, year in school, age, and other relevant demographic information.
Media frequency. To estimate weekly av- erages of instant messaging and e-mailing, an adapted version of the Media Habits Question- naire (Gentile et al., 2004) was used.2 To com- pute weekly e-mail frequency, participants in- dicated how many hours they e-mailed others (and checked e-mail) on an average weekday and weekend day during the following times: 6 a.m. to noon, noon to 6 p.m., 6 p.m. to midnight, and midnight to 6 a.m. The weekday times were added and multiplied by five. This estimate was added to the product of the week- end days multiplied by two. This formula was applied to instant messaging time. Thus, the range of possible scores was from 0 to 168 with higher scores indicating more frequency. Al- though results showed that these data were pos- itively skewed, no data transformations were conducted, because with a high sample size, the population distribution of scores approximates a normal distribution.
Results
Zero-Order Correlations
Table 1 displays the zero-order correlations between relevant variables. As expected, CB
frequency was positively correlated with posi- tive attitudes toward CB (r � .60, p � .001), perceived anonymity (r � .52, p � .001), and instant messaging frequency (r � .40, p � .001). CB was uncorrelated with e-mail fre- quency. This latter finding suggests that instant messaging is likely to be the method by which CB is manifested, rather than e-mail. Indeed, instant messaging frequency was also positively correlated with anonymity (r � .20, p � .001) and positive attitudes toward CB (r � .23, p � .001). E-mail frequency was uncorrelated with these aforementioned variables (rs � .11, ns).
Difference in Correlation Test
Prior to testing the moderating and mediating influence of anonymity in the relation between CB, it was of theoretical importance to show evidence that perceived anonymity was more strongly associated with CB than traditional bullying. A difference in correlation test for dependent samples was conducted (Cohen & Cohen, 1983). Results show a significant differ- ence, t(177) � 2.99, p � .05, in the magnitude of the relation between CB and anonymity (r � .52) and traditional bullying and anonymity (r � .30), while controlling for the colinearity of traditional bullying and CB (r � .31). This suggests that although the relation between tra- ditional bullying and anonymity was significant, this relation was stronger when CB was the predictor.
Moderation Test
Next, moderation tests were conducted to test the hypotheses of the current study. The Hayes and Matthes (2009) moderation MACRO for SPSS was used. This analysis tests the relation- ship between the independent variable (positive attitudes toward CB) and the dependent variable (CB frequency) at high (�1 SD) and low (�1 SD) level of the moderator (anonymity).
Results showed significant moderation, b � .01, se � .002, t(171) � 3.883, p � .001. The relation between positive attitudes toward CB and CB was significant when anonymity was high, b � .09, se � .01, t(171) � 6.14, p �
2 This measure was adapted by asking participants how many hours they were on IM and e-mail. The original version asked identical questions, but about video game use not IM or e-mail.
Table 1 Correlations Between Relevant Variables
1 2 3 4 5 6 7
1 — 2 .52�� — 3 .60�� .69�� — 4 .40�� .20�� .23�� — 5 .04 �.03 �.03 .53�� — 6 .31�� .30�� .40�� .09 �.02 — 7 .10 .07 .21�� .00 �.005 .17� — Mean 4.61 10.57 34.67 8.62 15.57 5.44 �0.16 SD 2.81 4.12 15.47 19.20 16.79 2.53 0.99
Note. 1 � Cyberbullying; 2 � Anonymity; 3 � Positive Attitudes toward Cyberbullying; 4 � Instant Messaging Frequency; 5 � E-mail Frequency; 6 � Traditional Bully- ing; 7 � Sex (1 � male; �1 � female). � p � .05. �� p � .01.
5CYBERBULLYING
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
.001, but not when anonymity was low b � .03, se � .02, t(171) � 1.66, p � .05 (see Figure 2).
Path Model
To test the mediating mechanisms within the Barlett and Gentile (2012) model, path analysis using MPLUS was used. The raw data were used for analysis (rather than inputting the correlation or covariance matrix). Instant mes- saging and e-mail frequency were correlated independent variables predicting anonymity, positive attitudes toward CB, and CB fre- quency. Anonymity and positive attitudes to- ward CB also predicted CB frequency. Finally, anonymity predicted positive attitudes toward CB (see Figure 3). Because all possible paths and correlations were estimated, the estimated variance–covariance matrix was identical to the actual variance–covariance matrix, making the model a perfect fit of the data (i.e., no degrees of freedom to estimate model fit indices; �2 � 0.00, Comparative Fit Index (CFI) � 1.00, Tucker-Lewis Index (TLI) � 1.00, Root Mean Square Error of Approximation (RMSEA) � 0.00, Standardized Root Mean Square Residual (SRMR) � 0.00, which is better than the base- line model in which all variables are uncorre- lated, �2 � 241.385 (df � 9), p � .00001).
Examination of the path coefficients showed that instant messaging frequency (� � .34, p � .001), positive attitudes toward CB (� � .41, p � .001), and anonymity (� � .16, p � .001) positively predicted CB frequency. E-mail fre- quency was negatively related to CB frequency (� � �.13, p � .05). Both e-mailing and instant messaging frequency predicted anonymity (� � �.18, p � .04; � � .29, p � .001, respectively); however, only instant messaging frequency (not e-mail frequency) predicted positive attitudes toward CB (� � .14, p � .04). Finally, ano- nymity predicted positive attitudes toward CB (� � .66, p � .001). The correlation between e-mail and IM frequency was significant, r � .53, p � .001. Indirect tests confirmed that the path from instant messaging frequency to ano- nymity to positive attitudes toward CB to CB frequency was significant (Indirect b � .08, t � 2.96, p � .01). The same indirect test involving e-mail frequency showed the opposite pattern (Indirect b � �.05, t � �1.97, p � .05).3
E-mail Versus Instant Messaging Frequency
The previous results show strong support for the importance of how different media outlets (i.e., IM and e-mail) influence anonymity, which, in turn, predicts CB frequency. How- ever, some may suggest that these results are driven by the fact that some individuals may simply use IM more frequently than e-mail. If true, such an explanation may provide an alter- native explanation to the previous findings. To test this, a repeated measures ANOVA was run to compare the frequency for IM to e-mail. Results showed that people spent significantly, F(1, 183) � 28.70, p � .001, partial 2 � .14, more time on e-mail (M � 15.57, SD � 16.79) than IM (M � 8.62, SD � 19.20). This pro- vides further support for the findings in the path model.
Conclusion
Overall, results support the hypotheses of the current study and show strong support that an- onymity is an important predictor of CB behav- ior. Consistent with the learning postulations of the Barlett and Gentile (2012) model, anonym- ity was both a mediator in the relation between instant messaging frequency and CB, but also a moderator in the relation between positive atti- tudes toward CB and CB frequency. This sug- gests that when individuals learn that CB is anonymous and the negative consequences are rare (given said anonymity), CB is likely to occur.
General Discussion
CB is an emerging societal problem. As sug- gested by Barlett and Gentile (2012), the liter- ature in this domain has been mostly descriptive and atheoretical. The purpose of the current research was to further our understanding of what variables predict CB behaviors in an at- tempt to (a) further advance theory by testing
3 Postexamination of the path coefficients showed one nonsignificant path (positive attitudes toward CB regressed onto e-mail frequency; see Figure 3). This path was set to 0 to estimate model fit indices. Results showed that this model was a good fit for the data (�2 � 1.38 (df � 1), p � .24, CFI � 1.00, TLI � 0.99, RMSEA � 0.05, SRMR � 0.01).
6 BARLETT
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
how anonymity is important to CB, (b) eluci- date on the predictors related to CB behaviors, and (c) use these findings to help inform future interventions aimed at reducing CB.
CB and Theory
Results from the current study support the postulations of the Barlett and Gentile (2012)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
Low (-1 SD) High (+1 SD)
Anonymity
Cy be
rb ul
ly in
g Fr
eq ue
nc y
Low CB A� (-1 SD)
High CB A� (+1 SD)
Figure 2. The moderating influence of anonymity in the relation between positive attitudes toward CB and CB frequency.
Instant Messaging Frequency
Emailing
Frequency
Positive Attitudes of
Cyberbullying Cyberbullying
Frequency
.34***
.41***
.29***
-.18*
.53***
Anonymity .67***
.16*
.14*
-.13* -.08
Figure 3. Mediated path model. � p � .05, ��� p � .001. Single headed arrows are regression coefficients whereas double headed arrows are correlations. Dashed lines indicate nonsignif- icant (p � .05) relations.
7CYBERBULLYING
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
model. This model posits that with continued CB experiences, individuals are likely to learn that they are more likely to be anonymous. Accompanied with the lack of power differen- tial, such anonymity leads to the development of positive attitudes toward CB, which predicts subsequent CB behavior. Support for the role of anonymity in CB was found in myriad ways. First, results showed that anonymity was corre- lated with both positive attitudes toward CB and CB frequency. Second, results showed that an- onymity moderated the relation between posi- tive attitudes toward CB and CB behavior. This suggests that CB is more likely when positive attitudes are high and anonymity is high. Third, anonymity significantly mediated the relation between instant messaging frequency and CB. This suggests that the reason why instant mes- saging frequency was positively related to CB was because individuals feel anonymous. Fi- nally, e-mail frequency was negatively related to anonymity and positive attitudes toward CB whereas instant messaging frequency was pos- itively related to such attitudes and behaviors. If theory is correct, this finding can be explained by the hypothesis that e-mail may be more identifiable than instant messaging.
Although these results regarding the impor- tance of anonymity are important to understand- ing the variables that predict CB behavior, the questionnaire used in the current research mea- sures anonymity attitudes. It has been argued that perceived anonymity may be more impor- tant than actual anonymity; however, that is speculative and future research should test this empirically. In the real-world, a cyberbully’s anonymity is not assured especially when be- haviors escalate to very aggressive or violent behaviors. The current study provided empirical data to suggest that perceived anonymity is an important contributing factor to predict CB, and, clearly, future work is needed to test under what conditions perceived versus real anonym- ity differentially affects such social behaviors.
Limitations and Future Research
Like all psychological research, limitations do exist that need be studied in future research. First, these data are limited by their correla- tional nature, and causal claims regarding the relations in this study cannot be made. Future research should use either an experimental or
longitudinal research design to test these hy- potheses. However, recent work by Barlett and Gentile (2012) used a longitudinal design to test similar hypotheses regarding the causal mecha- nisms of CB, and results were consistent to what was reported in this study. However, these studies have their limitations too. For example, Barlett and Gentile (2012) only used a two- month lag between scale administrations in their longitudinal study. Future research is des- perately needed in this domain.
Second, the current study used a college-aged sample. It could be argued that the frequency of CB peaks during junior high and high school years; however, no published work has tested such age comparisons. However, if significant relations can be found that test important theo- retical postulations in a population that uses CB less frequently, then it could be argued that the relations would be stronger for adolescents. However, future research should test these re- lations on adolescents to see whether the rela- tions reported herein with a college-aged sam- ple are similar or different.
Third, the measure of CB behavior was lim- ited to only measuring CB over the Internet. A complete definition of CB should consist of other means of technology (text messaging, over video game consoles, etc.). However, text messaging is not as anonymous as other forms of media (e.g., instant messaging), and the pur- pose of the study was to use a measure of CB that would have variance captured by the ano- nymity construct. Under certain circumstances, any social media communication can be anon- ymous (e.g., phone numbers can be withheld, video gamers can create avatars, people can create e-mail accounts with fake names, etc.). Future research should attempt to create and validate a measure of anonymity across various platforms to get a more comprehensive measure of anonymity in the cyber-world. Additionally, the measure of CB used in the current study is similar to other validated CB scales that ask how frequently people harm others while “on- line” (Ybarra et al., 2007). Future research should use other CB questionnaires to deter- mine whether the results replicate, and based on past research it should. For instance, Barlett and Gentile (2012) used the Ybarra et al. (2007) measure of CB and showed correlations similar to those presented in Table 1.
8 BARLETT
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
Fourth, related to the measure of CB used in the current study, no definitions or content of “mean messages” were provided. Further for ethical reasons, participants were not asked to indicate exactly what they would write if they indicated they had, in fact, written a mean on- line message. It was presumed that participants who score high on CB using this measure may not even remember every mean message or post. However, writing what participants con- sider mean messages fits the definition of CB.
Finally, future research should attempt to measure the impact of anonymity (or perceived anonymity) with other face-to-face measures. Olweus (2012) suggested that traditional bully- ing and CB are similar to one another; however, the reported frequency of CB is lower than traditional bullying. If that is true, then per- ceived anonymity should correlate with both forms of bullying to the same degree. This would specifically test whether anonymity is one defining characteristic of online versus of- fline aggression. However, caution must be used in doing such an analysis, because of (a) the high correlation between CB and traditional bullying (Barlett & Gentile, 2012), (b) the dif- ficulty in measuring anonymity given the constantly shifting online-world, and (c) the dif- ficulty in differentiating between traditional bullying from an anonymous source from CB from a known source. This is clearly an area of future work that needs attention. Furthermore, specifically related to the previous comment, future work should also attempt to develop valid and reliable measures of perceived anonymity (rather than attitudes toward anonymity) that can be assessed at the time the CB occurs.
Final Comments
Examining the predictors of CB is important. If such predictors can be tested, results repli- cated, and theory enhanced, then interventions can be formulated to reduce CB behavior. The current study is an important step in elucidating on what factors predict CB processes. It is clear that attitudes toward anonymity and CB rein- forcement are strong predictors of CB that should be targeted in future work and interven- tions.
References
Anderson, C. A., & Bushman, B. J. (1997). External validity of “trivial” experiments: The case of lab- oratory aggression. Review of General Psychology, 1, 19–41. doi:10.1037/1089-2680.1.1.19
Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53, 27– 51. doi:10.1146/annurev.psych.53.100901.135231
Anderson, C. A., & Dill, K. E. (2000). Video games and aggressive thoughts, feelings, and behavior in the laboratory and in life. Journal of Personality and Social Psychology, 78, 772–790. doi:10.1037/ 0022-3514.78.4.772
Anderson, C. A., & Huesmann, L. R. (2003). Human aggression: A social-cognitive view. In M. A. Hogg & J. Cooper (Eds.), Handbook of social psychology (pp. 296–323). London: Sage.
Ang, R. P., & Goh, D. H. (2010). Cyberbullying among adolescents: The role of affective and cog- nitive empathy, and gender. Child Psychiatry and Human Development, 41, 387–397. doi:10.1007/ s10578-010-0176-3
Ang, R. P., Tan, K., & Mansor, A. T. (2011). Nor- mative beliefs about aggression as a mediator of narcissistic explitativeness and cyberbullying. Journal of Interpersonal Violence, 26, 2619–2634. doi:10.1177/0886260510388286
Barlett, C. P., & Gentile, D. A. (2012). Attacking others online: The formation of cyberbullying in late adolescence. Psychology of Popular Media Culture, 1, 123–135. doi:10.1037/a0028113
Beran, T., & Li, Q. (2005). Cyber-harassment: A study of a new method for an old behavior. Journal of Educational Computing Research, 32, 265–277. doi:10.2190/8YQM-B04H-PG4D-BLLH
Beran, T., & Li, Q. (2007). The relationship between cyberbullying and school bullying. Journal of Stu- dent Wellbeing, 1, 15–33.
Center for Disease Control. (2011). Electronic ag- gression: Emerging adolescent health issue. Re- trieved from http://www.cdc.gov/features/ electronicaggression/. Retrieved October 24, 2011.
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ; New York: Erlbaum Assoc.
Diener, E. (1976). Effects of prior destructive behav- ior, anonymity, and group presence on deindividu- ation and aggression. Journal of Personality and Social Psychology, 33, 497–507. doi:10.1037/ 0022-3514.33.5.497
Gentile, D. A., Lynch, P., Linder, J., & Walsh, D. (2004). The effects of violent video game habits on adolescent’s aggressive attitudes and behaviors. Journal of Adolescence, 27, 5–22. doi:10.1016/j .adolescence.2003.10.002
9CYBERBULLYING
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
Hayes, A. F., & Matthes, J. (2009). Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementa- tions. Behavior Research Methods, 41, 924–936. doi:10.3758/BRM.41.3.924
Herring, S. (2001). Gender and power in online com- munication. Retrieved from https://scholarworks. iu.edu/dspace/bitstream/handle/2022/1024/WP01- 05B.html. Retrieval date: 8-22-13.
Hinduja, S., & Patchin, J. W. (2008). Cyberbullying: An exploratory analysis of factors related to of- fending and victimization. Deviant Behavior, 29, 129–156. doi:10.1080/01639620701457816
Hinduja, S., & Patchin, J. W. (2010). Bullying, cy- berbullying, and suicide. Archives of Suicide Re- search, 14, 206–221. doi:10.1080/13811118.2010 .494133
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
Li, Q. (2007). New bottle but old wine: A research of cyberbullying in schools. Computers in Human Behavior, 23, 1777–1791. doi:10.1016/j.chb.2005 .10.005
Olweus, D. (2012). Cyberbullying: An overrated phenomenon? European Journal of Developmen- tal Psychology, 9, 520 –538. doi:10.1080/ 17405629.2012.682358
Patchin, J. W., & Hinduja, S. (2006). Bullies move beyond the schoolyard: A preliminary look at cy-
berbullying. Youth Violence and Juvenile Justice, 4, 148–169. doi:10.1177/1541204006286288
Rivers, I., & Noret, N. (2010). ‘I h 8 u’: Findings from a five-year study of text and email bullying. British Educational Research Journal, 36, 643– 671. doi:10.1080/01411920903071918
Smith, P. K., Mahdavi, J., Carvalho, M., Fisher, S., Russell, S., & Tippett, N. (2008). Cyberbullying: Its nature and impact in secondary school pupils. Journal of Child Psychology and Psychiatry, 49, 376–385. doi:10.1111/j.1469-7610.2007.01846.x
Steffgen, G., Konig, A., Pfetsch, J., & Melzer, A. (2011). Are cyberbullies less empathic? Adoles- cents’ cyberbullying behavior and empathic re- sponsiveness. Cyberpsychology, Behavior, and So- cial Networking, 14, 643–648. doi:10.1089/cyber .2010.0445
Vandebosch, H., & Van Cleemput, K. (2008). Defin- ing cyberbullying: A qualitative research into the perceptions of youngsters. Cyberpsychology and Behavior, 11, 499 –503. doi:10.1089/cpb.2007 .0042
Ybarra, M. L., Diener-West, M., & Leaf, P. J. (2007). Examining the overlap in Internet harassment and school bullying: Implications for school interven- tion. Journal of Adolescent Health, 41, 42–50. doi: 10.1016/j.jadohealth.2007.09.004
Received October 31, 2012 Revision received June 17, 2013
Accepted July 2, 2013 �
10 BARLETT
T hi
s do
cu m
en t
is co
py ri
gh te
d by
th e
A m
er ic
an Ps
yc ho
lo gi
ca l
A ss
oc ia
tio n
or on
e of
its al
lie d
pu bl
is he
rs .
T hi
s ar
tic le
is in
te nd
ed so
le ly
fo r
th e
pe rs
on al
us e
of th
e in
di vi
du al
us er
an d
is no
t to
be di
ss em
in at
ed br
oa dl
y.
