INHIBITING ONLINE AGGRESSION IN ADOLESCENTS: SELF-CONTROL AS A
MODERATOR OF NEGATIVE INTERNAL PROCESSES IN HOSTILE COMPUTERMEDIATED INTERACTIONS
By
Christina Bernice Wirth-Hawkins

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Communication Arts and Sciences - Media and Information Studies
2010

ABSTRACT
INHIBITING ONLINE AGGRESSION IN ADOLESCENTS: SELF-CONTROL AS A
MODERATOR OF NEGATIVE INTERNAL PROCESSES IN HOSTILE COMPUTERMEDIATED INTERACTIONS
By
Christina Bernice Wirth-Hawkins
Preliminary research on cyberbullying has adequately presented online aggression as a
destructive adolescent behavior of growing concern. It is now time to move beyond
existing descriptive findings toward empirically-based, theoretically rich studies that help
to shape systematic and predictive understandings of the behavior. The present research
examines aggressive Internet behaviors of high school students from the General
Aggression Model’s (GAM) theoretical perspective. Participants were exposed to an
anger-inducing provocation scenario with one of three behavioral stimulus conditions
(anti-aggressive, aggressive, or control), in which they were bullied by a fictitious
classmate through an online social-networking site. As expected, the anti-aggressive
stimulus significantly curtailed state hostility in relation to the aggressive stimulus.
However, the anti-aggressive behavioral stimulus did not significantly impede cognition.
Trait and state self-control surfaced as important potential additions for the GAM. Trait
self-control significantly moderated the negative effects of discomfort and arousal on
aggressive behavioral intentions, such that adolescents with higher levels of negative
internal states were less likely to express aggressive behavioral intentions if they
maintained higher levels of trait self-control.

Trait self-control also emerged as a

significant moderator of the negative effects of affect on aggressive behavioral intentions
for low trait aggression individuals. State self-control surfaced as a significant moderator

of the negative effects of affect on aggressive behavioral intentions and of cognition on
aggressive behavioral intentions for low trait aggression individuals. Through qualitative
analysis, the study also identified interactions of online and offline aggression.
Specifically, results suggested that online aggression can lead to potential offline
aggression. The present study’s original contribution was to examine online aggression
through a theoretical framework, to present trait and state self-control as valuable
additions to the GAM, and to ascertain the value of exposure to anti-aggressive stimuli in
restraining online aggression. The study provides a valuable platform for understanding
how online aggression functions, how it can be inhibited, and how it should be explored
in the future.

ACKNOWLEDGEMENTS
To those who have helped me throughout my graduate career, I extend my deepest
gratitude and my sincere appreciation. This dissertation is dedicated to my parents, Alice
Bowling Wirth and Michael Wirth. Words cannot express my gratitude to them. Their
unfailing support and continuous encouragement throughout my long and arduous
graduate school journey is a debt which I will never be able to repay, but for which I will
be eternally thankful. Through example, they taught me that hard work yields great
benefits, and the completion of this project would not have been possible without them.
I will always be grateful to my advisor, Dr. Robert LaRose, for taking me under
his wing during my graduate career. He not only taught me the fundamentals of research,
but he provided me with multiple opportunities to put research into action by allowing me
to develop hands-on skills in his research projects. Dr. LaRose pushed me to think in
ways that I had not thought before, and his high expectations led me to quickly develop
skills that I did not come readily equipped with. He continuously challenged me to dig
deeper, to think bigger, and to always ask the “so what” and “why” questions. I have no
doubt that I will continue to hear his voice in the back of my mind in the future,
challenging me to think harder.
I am extremely appreciative of the other members of my dissertation committee,
Dr. Nora Rifon, Dr. Cliff Lampe and Dr. Nicole Ellison. Their generous donation of time
and expertise has improved my work and I am very grateful for their contributions and
support.
I am deeply indebted to the school administrators who afforded me with the
access that I needed to collect data for this dissertation. Acquiring access to adolescents

iv

for a research study involving aggression was a trying and time consuming battle in and
of its self.

I sincerely and emphatically thank Bobby Thomas, Teresa Harper and

Lorenzo Gonzales. Without their help, this project would not have been possible and I
am forever indebted to them.
Aside from my parents, my sister Michelle Wirth was my shoulder to lean on
during graduate school. I thank her for her support, her encouragement and for the many
hours that she spent on the phone with me.

I would also like to thank the members of

my extended family, specifically my grandparents, Bernice Onuoha, Obasi Onuoha and
Kathleen Wirth, and my aunts Pateath Herndon, Pamela Church, Angelia Onuoha, Sue
Hill, Niki Onuoha, Patricia Wirth, Janet Poley and Eileen Wirth for their constant love
and support.
Finally, I offer my sincere thanks and appreciation to my husband, Craig
Hawkins. During our first year of marriage, he was subjected to the trials and tribulations
of being married to a doctoral candidate who was working on her dissertation. While
sounding somewhat appealing and glamorous, I have no doubt that the experience was
anything but. He endured the emotional stress and tribulations with me. He encouraged
me to continue when I felt like I could not spend another moment working on this
project. He frequently joked that the completion of this dissertation was as hard on him
as it was on me, and that he could not wait for me to finish.

So, I will conclude by

enthusiastically saying “Thank you for your patience and support, Craig. We are done.”

v

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................vii
LIST OF FIGURES ….....................................................................................................viii
CHAPTER 1: Introduction...………………………………………………………….…. 1
Traditional Bullying…………………..……………………..………….…4
Online Aggression…………………………………………………………5
CHAPTER 2: Theoretical Background………….………………………………………16
Media Effects…………………………………………………………….16
The General Aggression Model (GAM)…………………….…………...22
CHAPTER 3: Hypotheses………………………………………….,……………………40
Self-Control…..………………….……………………………………….40
Emoticons and Behavioral Stimulus Activation…………...…………….52
CHAPTER 4 : Method…………………...………………………………………………61
Pretest…………………………………………………………………….61
Sample………….………...………………………………………………62
Data Collection……………....…………………………………………..64
Operational Measures……………………………………………………65
CHAPTER 5: Results…………………………………………………………………....69
CHAPTER 6: Discussion..………………………………………………………………93
Limitations..………………………..…………………………………...109
Future Research…..…….………………………………………………111
Conclusion……………………………………………………………...117
APPENDICES….............................................................................................................120
REFERENCES………....................................................................................................126

vi

LIST OF TABLES
Table 1: State Self-Control Scale ………………………………………………………..67
Table 2: Correlation Matrix ……………………………………………………………..70
Table 3: Regression Statistics for the Effects of Trait Aggression and Hostile Thoughts
on Aggressive Behavioral Intentions…………...………………………………71
Table 4: Regression Statistics for the Moderating Effects of Trait Self-Control on Hostile
Thoughts and Aggressive Behavioral Intentions for Low Trait Aggression
Individuals…….………………..……………………………………………….72
Table 5: Regression Statistics for the Moderating Effects of Trait Self-Control on State
Hostility and Aggressive Behavioral Intentions for Low Trait Aggression
Individuals………………………………………………………………………75
Table 6: Regression Statistics for the Moderating Effects of Self-Control on Perceived
Discomfort and Aggressive Behavioral Intentions...…………...……………....77
Table 7: Regression Statistics for the Moderating Effects of Trait Self-Control on
Arousal and Aggressive Behavioral Intentions…………………………………79
Table 8: Regression Statistics for the Effects of Trait Self-Control and Hostile Thoughts
on Aggressive Behavioral Intentions……………...……………………………82
Table 9: Regression Statistics for the Moderating Effects of State Self-Control on
Hostile Thoughts and Aggressive Behavioral Intentions for Low Trait
Aggression Individuals………………………………………………………....82
Table 10: Regression Statistics for the Moderating Effects of State Self-Control on State
Hostility and Aggressive Behavioral Intentions……………………………...84

vii

LIST OF FIGURES
Figure 1: The General Aggression Model………………………………………………23
Figure 2: The General Aggression Model Redrawn with Hypothesized Relationships…59
Figure 3: The Moderating Effects of Trait Self-Control on Hostile Thoughts and
Aggressive Behavioral Intentions for Low Trait Aggression Individuals……73
Figure 4: The Moderating Effects of Trait Self-Control on State Hostility and Aggressive
Behavioral Intentions……………………………………………..………...…75
Figure 5: The Moderating Effects of Trait Self-Control on Perceived Discomfort and
Aggressive Behavioral Intentions………………………..…………………….77
Figure 6: The Moderating Effects of Trait Self-Control on Arousal and Aggressive
Behavioral Intentions…………………………………………………………80
Figure 7: The Moderating Effects of State Self-Control on Hostile Thoughts and
Aggressive Behavioral Intentions……………………………….…………….83
Figure 8: The Moderating Effects of State Self-Control on State Hostility and Aggressive
Behavioral Intentions………………………………………………………….85
Figure 9: Significant Tested Relationships.……...……………………………………...90
Figure 10: Stimulus Materials for Aggressive Condition………………...…………….121
Figure 11: Stimulus Materials for Anti-Aggressive Condition……...………………….122
Figure 12: Stimulus Materials for Control Condition……….………………………….123

viii

Inhibiting Online Aggression in Adolescents: Self-Control as a Moderator of Negative
Internal Processes in Hostile Computer-Mediated Interactions

Chapter 1
Despite the inclination of most adolescents to use electronic media for friendly
social exchanges, some teenagers channel their aggression through media devices.
Teenagers commonly use electronic media, such as the Internet, for entertainment
purposes related to social connection and information-seeking (Pew, 2009); however,
electronic encounters can psychologically damage and distress victims in spite of their
infrequency. From the perspective of the General Aggression Model (GAM), aggression
is defined as “any behavior directed toward another individual that is carried out with the
immediate intent to cause harm” (Anderson & Bushman, 2002, p. 28). Online aggression
specifically refers to any Internet behavior directed toward another individual that is
carried out with the immediate intent to cause harm. Concordant with many other
aggression definitions, inclusion of the term “intent” makes the behavior somewhat
difficult to fully measure without specifically ascertaining the perpetrator’s true
intentions. It also leaves open the question of whether verbal behaviors (e.g. threats or
name calling) constitute aggression.

Recognizing the unique qualities of online

interaction, here it is assumed that intention can be inferred from the perpetrators’ actions
in sending harassing messages in online venues, that psychological as well as physical
harm is relevant, and that aggression can be verbal in nature. Terms such as
cyberbullying and flaming are also used to identify aggressive Internet behaviors. They
are referred to interchangeably throughout this study.
Anecdotal incidents of cyberbullying have catapulted the issue of online

1

aggression to the forefront of public concern. “Blast sites” embody a new forum for
cyberbullying. Adolescents send public websites degrading and humiliating information
about their friends, and the websites publicly post the stories. Victims are identified by
their first names, last names and pictures. One of these sites, called “Brutal Weeks,”
maintains a Facebook account that allows people to post and share publicly humiliating
information about their friends and classmates by posting stories on its Facebook wall
(Lee, 2010). One such victim was angered by information on the Brutal Weeks Facebook
wall that claimed that she was unattractive, that she smelled bad in private areas, and that
she performed sexual acts. Despite reporting many blast sites to Facebook, she stated
that the online social networking site had neither removed the humiliating wall posts nor
had it shut down the hostile user accounts. In another case, an eighth grade girl was
suspended from her Beverly Hills middle school after posting a YouTube video featuring
fellow students bad-mouthing a classmate (Kim, 2009). Adolescents used the website to
publicly ridicule their classmate, calling her ‘spoiled,’ a ‘brat,’ and a ‘slut.’ Text and
electronic messages enabled the act of electronic aggression to become a viral
phenomenon throughout the school. In a more extreme case, a 17-year-old female from
San Jose, California tried to take her own life after becoming the victim of repeated
harassment through thousands of text messages (Gebhart, 2009). She now suffers from
severe brain damage despite surviving the suicide attempt. The most tragic cases involve
those, such as the popular teenage star athlete from Long Island who sadly succeeded in
taking her own life after becoming the victim of online abuse. She committed suicide
after falling prey to vicious online attacks on social networking sites such as
Formspring.me and Facebook (Del, Signore, 2010). People continued to post brutal

2

comments about her online after her death.
Cases of severe electronic aggression with devastating outcomes such as these are
rare. Most incidents of online aggression do not make the local news, and most are likely
only known to the aggressor and the victim. Online aggression clearly presents a cause
for concern when the result of the destructive behavior is fatal.

It also becomes

concerning when it leads to subsequent violence or when it causes psychological damage.
Whatever the outcome, the deviant behavior will likely continue with the growing
popularity of the Internet and other forms of electronic communication. The existence of
cyberbullying poses a serious and important question. What can be done to inhibit
aggressors who act through electronic media?

This question becomes increasingly

important if online aggression mirrors offline aggression. Acts of offline aggression
produce more, potentially violent acts of aggression, and the same may be true for
cyberbullying.
The purpose of this study is to examine online aggression within the theoretical
framework of the General Aggression Model (GAM). It is the first study to apply the
GAM to online aggression, and it is the first to empirically examine the antisocial
behavior from a theoretical perspective in general. The study also presents self-control as
a new variable for the GAM by investigating whether the construct successfully inhibits
online aggression. Finally, this study considers the interactions between behavioral cue
exposure, cognition, state hostility, discomfort, self-control and arousal in the inhibition
of online aggression.

3

Traditional Bullying
Although cyberbullying occurs in a dissimilar domain to traditional bullying, the
online behavior is rooted in its offline counterpart.

Like cyberbullying, traditional

bullying research is somewhat complicated by inconsistent definitions that refer to the
behavior in terms of direct physical forms of violence, or in terms of more indirect forms
of violence such as harassment and verbal attacks (Ferguson, Miguel, Kilburn &
Sanchez, 2007). Fuzzy terminology such as the meaning of “intentional” and whether
aggression should be interpreted through the eyes of an outsider or from the perspective
of the victim further convolute analysis of the behavior (Smith, 2004). Regardless of
definitional constraints, research suggests that while bullying can occur in a variety of
environments that the deviant behavior most frequently occurs in school settings among
children and adolescents (Ambert, 1994).
In a survey of school bullying among adolescents in the United States, nearly 21%
reported that they had been physically bullied at school within the last 2 months, and
almost 54% claimed that they had been verbally bullied (Wang, Iannotti, Nansel, 2009).
Verbal aggression constitutes “informing another that she or he is bad, possesses negative
qualities, or is not meeting some internal or external standard” (Kinney, 1994, p. 4). It
can involve “criticism, insults, cursing, or a negative affective reaction” (Greenberg,
1980, p. 108) or attacking the self-concept of another by “insulting the other’s character,
competence, background, or physical appearance” (Chory-Assad, 2004).

Previous

studies have found that verbal aggression is the most common form of bullying (Smith,
1994; Craig, Pepler & Atlas, 2000).
A meta-analysis of bullying among adolescents suggested that while males exhibit

4

physically aggressive behaviors more frequently than females, that differences between
the sexes in relation to indirect bullying, which refers to exclusion, social ostracism,
social isolation and gossiping, are relatively small (Scheitauer, 2002). Similarly, Wang et
al. (1999) concluded that boys were more involved in physical and verbal aggression,
whereas girls were more involved in relational aggression, which involves intentional
manipulation of peer relationships to damage another individual’s friendships and to
create feelings of insecurity. Research suggests that traditional bullies tend to come from
home backgrounds with less affection, more violence and low parental monitoring
(Olweus, 1999) compared to non-bullies, while bully victims are frequently linked to
depression, anxiety and low self-esteem (Hawker & Boulton, 2000). The proliferation of
the Internet creates a virtual environment in which adolescents can engage in behaviors
that closely align with traditional bullying.

Online Aggression
Empirical research involving online aggression is a relatively new undertaking, and the
bulk of studies involving the behavior provide little in terms of explanatory utility. Still,
descriptive findings provided through preliminary research suggest that online aggression
is a behavior of growing concern. Many accounts of online aggression refer to the
behavior as flaming.

Preliminary flaming studies recognized that many individuals

expressed themselves more strongly through the computer than they would in other
settings of communication (Kiesler, Siegel & McGuire, 1984). Another flaming study
similarly found that incidents of flaming among students increased when they worked
together through an anonymous computer-mediated group rather than when they worked

5

together in a traditional classroom environment (Reinig, Briggs, Nunamaker, 1997).
Indeed, many early accounts of flaming attributed the anti-social behavior to a
consequence

of

the

disinhibition

experienced

through

computer-mediated

communication, whereby a lack of social context and visual cues facilitate a disregard of
social standards and conventions (Kiesler, Siegel & McGuire, 1984).

Other early

accounts alternatively suggested that the behavior is a product of the social norms of the
computing subculture rather than a product of a lack of social norms (Lea, O’shea, Fung
& Spears, 1992). In an attempt to reconcile the two competing perspectives, Kayany
(1998) examined social newsgroups on a primitive Internet discussion system called
Usenet. He found that like other media of communication, the tone and emotional
content of computer-mediated communication is more strongly affected by the social and
interaction context than by the medium itself. However, online affordances provided
through the Internet such as the potential for asynchronous communication and a
significant physical barrier distinguish online aggression from its offline counterpart.
Through his account of “A Rape in Cyberspace,” Dibbell (1993) offered an early
example of the type of devastation that aggressive behavior can produce for victims
regardless of the medium in which it occurs. A user of a multiplayer computer game
used the tools and resources available within the technology to engage in a virtual rape of
another player. Despite occurring in a virtual environment, the incident demonstrated
that online aggression can lead to emotional trauma, and that the boundaries between
what is considered real-life and what is considered virtual reality may not be as clearly
distinguishable as previously considered.
Most existing cyberbullying studies describe online aggression in terms of simple

6

descriptive statistics by identifying the frequency with which the behavior occurs and by
describing gender differences. They focus mainly on the aftermath of the behavior. That
is, they highlight the behavior’s effects on the victim but give little attention to
understanding the mindset of the perpetrator and the implications of the characteristics
that distinguish online aggression from offline aggression. A clear understanding of the
processes behind electronic perpetration is needed not only to understand the behavior,
but to inhibit its persistence. Empirically-based, theoretically rich studies are needed to
systematically understand online aggression as well as its byproducts.
Motivations for engaging in aggression online differ between individuals. Miller
(2006) shed light on how various forms of motivation can lead to categorically
distinguishable types of cyberbullies. Miller described multiple categories of online
aggressors including “vengeful angels,” “revenge of the nerds,” “mean girls” and the
“inadvertent.” Vengeful angels refer to individuals who electronically bully an individual
who has bullied them or their friends. In other words, these individuals aggressively
retaliate against someone who first behaved aggressively. Revenge of the nerds refers to
individuals who bully to have power over others over whom they normally have no
power. This category suggests that features inherent to electronic media can empower
unlikely traditional bully candidates. The Internet may enable people with physical or
social limitations who would otherwise refrain from face-to-face interactions to behave
aggressively. Mean girls refer to individuals who get together in a group to cyberbully.
The term refers to girls because females more commonly engage in this type of activity;
however, males can also get together in a group to bully through electronic
communication. Inadvertent refers to individuals who do not know they are doing wrong

7

when they respond to an upsetting message, or who intend to be playful and do not
understand that others could take it seriously. However, the absence of intent in the case
of the inadvertent disqualifies the behavior from those belonging to this study’s
conceptualization of aggression. Miller’s categories illustrate potential differences and
similarities between offline and online aggression. At the same time, all of Miller’s
categories, aside from that which characterized the unlikely bully, share similarities with
traditional bullying. At the very least, the categories demonstrate that the behavior is
manifested through various and perhaps dissimilar manners, and that a variety of factors
may promote or inhibit the behavior depending on its primary motivation.
According to a recent survey, 93% of youth ages 12-17 use the Internet (Pew,
2009). This suggests that while nearly all youth in this age group enjoy the Internet’s
benefits, that they are also at risk for potential downfalls of Internet activity such as
interpersonal bullying. Research indicates that around 40-60% of adolescents have been
harassed on the Internet and that around 20-30% of them have used electronic
communication to harass others (Beran & Li, 2005). In a national survey involving youth
and the Internet, one in three teenagers (34.5%) admitted that they had been harassed
online in the last year (Ybarra, Diener-West & Leaf, 2007). According to these statistics,
many teenagers have neither engaged in nor become victims of online aggression, but an
alarming percentage have.
Previous online aggression research outlines a variety of characteristics
commonly connected to victimization and perpetration of Internet aggression. A survey
about youth and online interpersonal victimization found that 2 in 5 youth were with
peers when they harassed others online, and that behaving aggressively online was one of

8

the most strongly related antecedents to becoming a victim of interpersonal online
harassment (Ybarra, Mitchell & Finkelhor, 2007). The study revealed that youth who
intentionally embarrassed someone online 3 or more times were among the most likely
candidates for becoming victims of online aggression. An earlier survey conducted by
the same team of researchers found that those who harass online are generally White, 1517 year-olds (Ybarra & Mitchell, 2004b). These statistics present a logical cyberbully
profile considering that Caucasians in this age group maintain disproportionally high
leisurely access to the Internet when compared with most other racial groups and younger
individuals.

More access creates more opportunity, especially when the access is

unrestricted. Those who can use the Internet at their discretion at home are more likely to
harass others than those who enjoy only restricted access to the Internet in public
locations or who are monitored by their parents. Adolescents who aggress online are not
only subjected to infrequent monitoring, but they tend to have poor bonds with their care
givers (Ybarra & Mitchell, 2004b). Traditional aggression research has demonstrated a
similar pattern, whereby adolescents with highly supportive parents tend to exhibit less
aggression (Finkenauer, Engels & Baumeister, 2005). An abundance of research in
multiple fields also demonstrates that parental mediation is an important factor in
curtailing risky media behaviors (Gentile, Lynch, Linder, Walsh, 2004; Austin, Pinkleton,
Fujioka, 2000; Anderson, Berkowitz, Donnerstein, Huesmann, Johnson, Linz, Malamuth
& Wartella, 2003; Rosen, Cheever & Carrier, 2008; Livingstone & Helsper, 2008).
Adolescents with parents who monitor their activities online are likely more afraid
of the potential consequences of online aggression than are those whose parents allow
them free rein over their Internet activities. Still, results from multiple studies suggest

9

that parents often do not monitor the online activities of their children. A study involving
parental monitoring and an online social networking site found that most parents were
largely unaware of their child’s activities on the site, that they rarely viewed the site, and
that they were unsure about the type of information that their child had posted on the site
(Rosen, Cheever & Carrier, 2008). Authoritative parents, or those who monitor and
impart clear standards for their children’s conduct with warmth, have been linked to
better choices for teens on the Internet. A telephone survey regarding parenting styles
and adolescent Internet activity found that adolescents with authoritative parents are more
likely to experience evaluative and restrictive Internet content mediation than children
whose parents engage in alternative parenting styles (Eastin, Greenberg & Hofschire,
2006). Another study correlated the authoritative parenting style with the fewest risky
Internet activity behaviors, including disclosing personal information (Rosen, Cheever &
Carrier, 2008). Thus, parental intervention is an important mediator whether in the form
of direct supervision or indirect consequences; however, it is unlikely to occur in the
social media venues where cyberbullying takes place.
Most research suggests that males and females are equally likely to engage in
online aggression (Werner, Bumpus & Rock, 2009; Ybarra & Mitchell, 2004b; Hinduja
& Patchin, 2008), but some studies (Smith, Madavi, Carvalho, Fisher, Russell et al.,
2006; Kowalksi, Limber & Agatston, 2008) suggest that girls engage in online aggression
more frequently than boys. The notion that females engage in more, if not the same
amount of online aggression as males runs contrary to societal conceptualizations of
aggression as well as with most accounts of offline bullying (Kowalski & Limber, 2007).
Consistent with this perspective, the most infamous incidents of online aggression

10

involve females rather than males. Indeed, most of the faces belonging to cyberbulling
perpetrators and victims in the news media belong to females. Typically, males are not
only expected to exhibit more aggression than females, but they are conditioned to do so
from a young age through toys, social models and social expectations. Bandura, Ross
and Ross (1961) advised that the tendency for males to exhibit more aggression than
females can be explained through socialization. They suggested that society makes it
more acceptable for males than for females to exhibit trait aggression because boys are
raised to behave more aggressively. Studies involving aggression in offline environments
support the argument that males tend to exhibit more aggressive behavior than females
(Kowalski & Limber, 2007).
If females are indeed just as likely, if not more likely, than males to behave
aggressively through electronic forms of communication, gender roles in online
aggression starkly contrast those of offline aggression (Kowalski & Limber, 2007).
Studies portraying females as likely electronic aggressors do not specifically distinguish
between anonymous and private and open and public aggressive encounters.
Cyberbullying research tends to lump the behavior into one, non-specific category.
However, a fuller understanding of the behavior may illustrate that private and/or
anonymous online aggression differs substantially from public and/or identifiable online
aggression.
The indirect nature of anonymous and private aggression may partially explain
why girls are represented as both perpetrators and victims of cyberbullying more than in
traditional bullying.

Characteristics of electronic communication (e.g. potential for

anonymity) may remove some of the confining social expectations placed on females. If

11

the identity of the perpetrator is unknown, she cannot be personally judged. Likewise, if
the behavior is enacted privately, it is less susceptible to public social expectations.
Characteristics inherent to the online environment can potentially modify
dimensions of bullying because they carry a different set of perceived consequences.
Anonymity and privacy cannot only eliminate many of the normative social consequence
expectations; they can also reduce fear of retaliation. In reality, online aggression is often
perpetrated anonymously because the Internet provides a medium through which
individuals can misbehave without the fear of immediate physical retaliation (Wolak,
Mitchell & Finkelhor, 2007). Physical abilities often empower traditional bullies. A
physically tough teenager with a strong desire to fight a weaker teenager can accomplish
his goal with few impediments.

Conversely, fear of physical retaliation or lack of

bullying self-efficacy may restrain a physically weak teenager from engaging in the same
type of behavior. Electronic communication greatly diminishes the utility of physical
strength, especially when aggressive behavior is enacted anonymously.

Anonymity

creates the potential not only for unlikely candidates for traditional bullying to become
online aggressors, but for victims in the offline world to become bullies in the virtual
world without fear of retaliation. Research has yet to empirically ground the notion that
electronic communication devices reach their undeniable potential for enabling nontraditional bullies to become cyberbullies.

While certain characteristics separate

electronic communication from face-to-face communication, the notion that these
features play a large role in modifying aggressive tendencies is merely speculative until it
is scientifically tested. Online aggression, like offline aggression, may reveal itself to be
a derivative of trait aggression rather than a distinctly altered behavior.

12

To what extent does online aggression truly differ from traditional forms of
aggression? Many characteristics of electronic harassment run parallel with their offline
counterparts.

A majority of bullying occurs in public locations such as the school

cafeteria, hallways and playgrounds in the offline world (Swearer & Doll, 2001), but
instant messaging, chat rooms, e-mail messages and websites are among the most
reported cyberbullying platforms in the online world (Kowalski & Limber 2007;
Agstston, Kowalski & Limber, 2007). Traditional bullying is also usually confined to the
hours of the school day (Rigby, Smith & Pepler, 2004), but electronic bullying eliminates
the barrier that allows bully victims to find solace in the comfort of their own homes. At
the same time, virtual environments create a physical barrier to the bully from which
some may find comfort.

Adolescents can engage in open and public electronic

aggression just as easily as they can engage in private or anonymous aggression. Online
affordances also facilitate and accelerate group bullying by allowing adolescents to
collectively gang up on another person, even when they are not together in spatial
proximity.

The effects of anonymity and privacy afforded through electronic

communication carry little weight when the behavior is perpetrated openly.

The

behavior likely more closely aligns with offline aggression in these instances.
Despite the ease with which one can anonymously convey hostility through the
Internet, harassment by a known perpetrator can incite a more personal form of
victimization for the message’s receiver. Ignoring an act of online hostility becomes
more difficult when the victim knows the attacker personally outside of the Internet.
Indeed, online aggression likely aligns more closely with its offline counterpart when it is
committed by a known perpetrator because the hostility is more difficult to ignore. The

13

question then becomes: how will the victim of the online attack respond to the
provocation?
Results from multiple studies indicate that online aggression is potentially
problematic because it can become part of an ongoing aggressive cycle. Ybarra and
Mitchell (2004a) connected traditional bullying victimization with online aggression,
indicating that adolescents who are bullied at school may become prime candidates for
becoming bullies on the Internet.

Numerous studies have presented past Internet

victimization as an antecedent to future online aggression (Ybarra & Mitchell, 2004b;
Kowalski & Limber, 2007; Wolak, et al., 2007; Werner, Bumpus & Rock, 2009).
According to these reports, online aggression often creates a vicious retaliatory cycle in
which an aggressive act by one peer can cause an equally, if not more aggressive act of
retaliation by the so-called victim and so forth. In some cases, past Internet victims do
not necessarily retaliate against their original perpetrator, but have learned that the
Internet is a viable medium through which one can effectively engage in aggression.
Wolak et al.’s (2007) study revealed that those who were harassed by peers over the
Internet were more likely to have used the medium to harass someone they were angry
with. Traditional bullying research has also demonstrated that this cyclical pattern is
common in offline environments. An offline adolescent aggression study found that
individuals bullied at school were significantly more likely to become bullies (Barboza,
Schiamberg, Oehmke, Korzeniewski, & Post et al., 2009).
Much work needs to be done to flesh out the different types of cyberbullying. At
this point, many of the ideas surrounding the implications of aggression through
electronic media rely mainly on descriptive statistics or assumptions about characteristics

14

of the medium. Thus, most conclusions about the behavior rely mainly on speculation.
Anonymous and/or private online aggression may significantly differ from open,
identifiable online aggression. This study notes the importance of examining anonymous
and private cyberbullying, but it looks specifically at open, identifiable online aggression
within the context of computer-mediated provocation from a known perpetrator. It seeks
to understand this particular type of aggressive behavior and to identify factors that
promote and inhibit the behavior. Its main goal is to identify manners through which to
impede online aggression, and thus the vicious retaliatory aggression cycle demonstrated
in previous research. The present study’s original contribution is to examine online
aggression through a theoretical framework, to present trait and state self-control as
valuable additions to the GAM, and to ascertain the value of exposure to anti-aggressive
behavioral stimuli in restraining online aggression. Chapter 2 aims at grounding the
behavior of cyberbullying within an explanatory and predictive theoretical framework.

15

Chapter 2
Theoretical Background
Media Effects
Two gunmen shot and killed twelve of their fellow classmates and one teacher
before killing themselves at an affluent Colorado High School in 1999. Heartbreaking
episodes such as the Columbine massacre have called attention and harsh criticism to a
culture of media violence produced from movies and video games.

Attempts to

understand the tragic anger and rage of two seemingly unlikely killers led many people
throughout the nation to point fingers of culpability at parental guidance and the negative
outcomes of social exclusion. The search for an answer also left many fingers pointing
directly at the media. According to one media critic, “Investigators found a strong
connection between violent video games and the Columbine shootings, as well as other
high school shootings here in the U.S. and Europe” (Tomeo, 2007).
Public television and radio networks saturated news coverage with stories
revolving around uncanny similarities between the violence depicted in the video games
and the movies that contributed to the Columbine killers’ mediated experiences. For
instance, the protagonist in The Matrix, a motion picture that had been recently released,
wore a black trench coat and carried multiple automatic pistols as he raged against his
antagonists. Many argued that the real-world scene that unfolded in the unsuspecting high
school halls of Columbine appeared too eerily similar to be discounted as completely
coincidental. Subsequent school violence incidents such as the tragedies that unfolded at
Virginia Tech and Ecole Polytechnique drew similar comparisons. On a smaller scale,
people continually question the effects of violent video games on individuals who spend

16

much of their free time absorbed in severely aggressive interactive content. They argue
that the unique interactive capabilities of video games, more than television or film, may
make them even more influential (Children Now, 2001).
Associations between violent episodes such as school shootings and media
content are easily drawn, yet few conclusions made by the greater public are grounded in
scientific research or theory. For most individuals, the media do not exert the direct and
powerful influence widely postulated after the Columbine incident. Potential connections
between media content and aggressive behavior present a clear call to understand the
implications of media effects. Some media effects literature argues a more restrained
interpretation of the effects of violent media on aggressive behavior (McGuire, 1986;
Freedman, 1988; Ferguson, 2009). They advise that aggressive media stimuli, while
potentially influential, are not always as directly powerful as some may believe. On the
other side of the debate, meta-analytic reviews of violent stimuli and aggressive
behaviors strongly advise that the connection between media violence and aggression is
undeniable (Wood, Wong & Chachere, 1991; Paik & Comstock, 1994; RoskosEwoldsen, Roskos-Ewoldsen & Carpentier, 1994; Bushman & Anderson, 2001).
The Magic Bullet Theory (or Hypodermic-Needle), a primitive explanation of
media effects, emphasized the potent persuasive powers of the media in shaping
perspectives and actions. The idea that media are so strong that individuals largely
absorb messages without filtering or selectively interpreting the content has been largely
discredited (Sproule, 1989; Greenberg & Salwen, 1996). For some individuals, such as
children, magic bullet-type effects are more likely. Mediated information has more
potential to exert influence over people who have had little to no direct contact with a

17

matter because they lack a sufficient method for evaluating information (Liebert &
Schwartzberg, 1977; Fujioka, 1999). For instance, a television character who complains
that everyone has a monster under their bed would more likely terrify a young child than
a seasoned adult. Likewise, a television character who suggests that hitting one’s brother
results in a delicious candy bar would doubtfully convince a mature, experienced
individual. The media’s strong influence is more likely to affect children because their
knowledge structures closely resemble a blank slate or a sponge.
Children develop their understanding of the world in which they live during the
critical stages of childhood and adolescence. Various life experiences educate them in
the social roles, societal constructions, and behaviors that their environments generally
accept.

Bandura and colleagues (1961) demonstrated the potentially detrimental

influence of violent media on children through social learning.

Children in their

experiment viewed one of three films with an adult attacking a blow-up doll with a stick.
The first film featured the adult hitting the doll without reward or consequence. The
adult was rewarded and praised after hitting the doll in the second film, and the adult was
punished after hitting the doll in the third film. Children were left alone with a similar
doll after viewing the film.

Those who viewed the punishment condition behaved

significantly less physically and verbally aggressively towards the doll than did children
in the other treatment groups. These experiments provided support for Bandura’s social
learning theory, which advises that children can learn behaviors such as aggression from
watching and imitating others.
Aggressive scripts learned during childhood often form the basis for the
individual’s knowledge structure and can become activated throughout the individual’s

18

lifetime.

Repeated exposure to parallel messages increases the knowledge structure’s

strength and accessibility. Through social learning, magic bullet-type explanations of the
effects of media content on children are somewhat more viable. However, by the time
children reach adolescence and young adulthood, years of socialization, interaction with
family and friends, and life experiences create much more complex knowledge structures
from which they interpret media messages.

Mediated content can still affect the

adolescent or adult, but knowledge structures and personalities filter the effects of the
content on behavior.
A meta-analysis of media violence (including television, movies, video games
music and comic books) and aggressive behavior considered the links between media
violence and short and long-term effects (Bushman & Huesmann, 2006).

Through

combining the results of multiple studies, it found that short-term media effects were
greater for adults than for children, whereas long-term effects were greater for children
than for adults. The analysis argued that short-term effects through priming are more
likely to affect adults because they rely on well encoded scripts, schemas or beliefs which
adults have had more time to develop than have children. On the other hand, long term
media effects require the encoding of scripts, schemas or beliefs to which children are
more receptive through observational learning.
Along the same lines, one probable explanation for how media violence affects
individuals involves desensitization through cultivation. For example, repeated exposure
to portrayals of women in the media has been linked to sexual harassment, violence
against women, and eating disorders (Lavine, Sweeny & Wagner, 1999). Cultivation
theory advises that continual exposure to mediated reality can cultivate beliefs about

19

violence and thereby conduct (Gerbner & Gross, 1976; Dixon & Linz, 2000; Bandura,
1986). The media create pictures in viewers’ heads that serve as knowledge of reality.
Media messages can portray the world as being more violent than reality would suggest.
They can also depict violence and aggression as acceptable behaviors if the individual
possesses a meaningful motivation.
Many films feature antagonists who shoot, kill or fight in the name of revenge or
to fulfill a basic need. Bandura (1986) proposed that drawing from media experiences
may cause children to develop false beliefs about the real world. Insufficient information
derived from media such as television, movies and video games may produce thoughtprocessing errors. Television messages and video game content can be internalized
through massive and continuing exposure. Cultivation provides a logical explanation for
the manner through which media content can affect individuals, yet providing empirical
support for the theory proves somewhat difficult.

By definition, testing the theory

requires a long-term longitudinal research design that lasts throughout a long portion of
the participants’ lifetimes (Bushman & Huesmann, 2000, p231). Tests are susceptible to
many confounds as well as to attrition.
The General Aggression Model (Anderson & Bushman, 2002), the successor of
the General Affective Aggression Model (Anderson, Deuser & DeNeve, 1995), has
become the resident theory used to explain and predict the effects of media on aggressive
behavior. The theory is more robust than previous media effect theories because it can be
used to understand all types of aggressive behavior, not only those confined to the media.
Through its incorporation of many previous theories of aggression and human behavior,
the GAM has the ability to elegantly combine and account for many previous media

20

effect theories.

Models integrated into the GAM’s theoretical framework include

Bandura's social learning theory, Berkowitz's Cognitive Neoassociationist Model,
Dodge's social information-processing model, Geen's affective aggression model,
Huesmann's social-cognitive model of media violence effects, and Zillmann's excitation
transfer model (Anderson & Dill, 2000).
The GAM proposes that aggressive behavior, or lack thereof, is affected by
various input variables (e.g. situation cues and individual differences), internal state
variables (e.g. cognition, affect and arousal), and appraisal processes. It accounts for one
of the Magic Bullet Theory’s primary deficiencies through its proposition that factors and
differences related to the individual influence the effects of violent media on behavior.
Although most GAM applications are cross-sectional in nature, it also accounts for one of
the cultivation theory’s major propositions, that exposure to messages and images can
affect behavior through long-term internalization and memory.

These effects are

discussed further within the discussion of the GAM in the next section.
The GAM has become well established within the field of media research. The
primary concern of several GAM media studies is to address societal criticisms relating
to detrimental media by explaining the effects of video game violence (Anderson &
Bushman, 2002; Kirsh, 2003; Giumetti & Markey, 2007) and violent films (Bushman &
Geen, 1990; Anderson, 1997; Anderson, Dueser & DeNeve, 1995) on aggressive
behavior. Few GAM studies focus on the implications of aggression and how it relates to
the Internet.

However, the GAM’s utility in understanding aggression involving

alternative forms of media leads naturally to its application to the Internet.

21

The General Aggression Model
The GAM calls attention to interactions between factors related to the individual
and to the situation to understand aggressive behavior. Specifically, it proposes that
aggressive behaviors result from a variety of input variables, present internal state
variables, appraisal processes and outcome variables. Early tests of the theory focused
primarily on the effects of aggressive stimuli (Bushman & Geen, 1990) and unpleasant
conditions (Anderson et al., 1996; Anderson & Bushman, 2002) on aggressive behavior
through cognition, affect, and arousal. Many of these tests illustrated the potential of
exposure to aggressive behavioral stimuli to adversely affect aggressive behaviors, but
they did not fully account for important personality factors that the aggressor brought to
the situation. The interaction between person factors (e.g. trait aggression) and situation
factors (e.g. aggressive stimuli) have become the primary concern for the theory’s later
development (Joireman, Anderson & Strathman, 2003).
The GAM provides an ideal framework for understanding online aggression
although most of the studies to which the theory has been applied involve media
environments such as video games or television. While the theory’s utility in terms of
understanding bullying and other offline situations has been discussed above, the GAM
has yet to be fully explored outside of the media realm. However, the type of aggression
depicted in this study is verbal in nature, and previous studies have applied the GAM to
verbal aggression. A longitudinal study that utilized components of the GAM found that
students who reported a high level of exposure to violent video games at the beginning of
the school year scored higher on both verbal and physical aggression at the end of the
year (Moller & Krahe, 2009). Media violence exposure was associated with all types of

22

aggression for boys, but only for verbal aggression for girls.

In another study,

participants who were exposed to a verbally aggressive television sitcom produced a
significant number of verbally aggressive cognitive responses and produced more
verbally aggressive cognitive responses than did participants exposed to a crime drama
(Chory-Assad, 2004).

Examination of the online aggression phenomenon requires a

dynamic theory such as the GAM that can effectively account for and incorporate both
the situational components of the mediated environment and the characteristics of the
individual that ultimately substantially contribute to the behavior. The GAM’s proposed
relationships (Anderson & Bushman, 2002) are depicted in Figure 1.

Figure 1 The General Aggression Model (Source: Anderson & Bushman, 2002)

Inputs

Person

Situation

Present Internal State:
Routes

Outcomes

Affect
Cognition
Arousal

Appraisal
& Decision
Processes

Social
Encounter

Thoughtful
Action
Impulsive
Action

Present Internal State and Input Variables
The GAM suggests that an individual’s present internal state mediates the input
23

variables’ impact on the final outcome behavior. For example, situational input variables
such as pain, uncomfortable temperatures and exposure to violent movie clips have all
been shown to result in increased aggression (Anderson & Bushman, 2002). Activating
these situational factors can lead to increased levels of hostile affect, cognition and
arousal, which namely account for the internal present state components of the GAM.
Cognition involves highly accessible aggressive scripts and hostile thoughts. Scripts
refer to concepts stored in memory that develop over time from life experiences and that
guide plans, actions, and social interactions (Schank & Abelson, 1977). Hostile thoughts
directly refer to an individual’s active contemplation of enacting hostile behavior.
Studies examining the effects of viewing weapons (Berkowitz & LePage, 1967;
Bartholow et al., 2005) or violent media (Bushman, 1998) on aggressive behavior have
demonstrated that imagery with violent associations often stimulate aggressive
cognitions.
Affect refers to an individual’s situational mood and emotional state. Affect in
the GAM directly involves state hostility which refers to the degree of momentary anger
felt by the individual. Many studies examining the effects of state hostility on aggressive
behavior have found that increased anger leads to aggressive behavior. A study involving
workplace aggression demonstrated that angry individuals, or those in negative emotional
states, were more likely to behave aggressively towards customers (Grandey, Dickter &
Sin, 2000).

Another study similarly found that negative mood states resulted in

aggressive behavior when combined with the belief that aggressive behavior is good for
one’s mood state (Bushman, Phillips and Baumeister, 2000). Similarly to the effects of
stimuli on cognition, studies examining the effects of violent media (Bushman & Geen,

24

1990) on aggressive behavior have demonstrated that imagery with violent associations
often stimulate state hostility.
Input variables include two components: situational factors and person factors.
Situational factors refer to any important aspect of the situation including presence of a
provocation, aggressive cues, frustration, pain, induced discomfort, drugs, and incentives
(Anderson & Bushman, 2002).

Acts of provocation not only make certain individuals

angry and produce hostile thoughts; they also provide an impetus for aggressive behavior
which may otherwise not exist. The GAM specifies that provocation primes aggression
related cognition, affect, and arousal. Vasquez, Denson, Pedersen, Stenstrom and Miller
(2005) manipulated provocation by provoking some, but not all, participants through
degradation and insult and by subsequently exposing all participants to a neutral, mild or
moderately strong triggering event. Results indicated that aggressive behavior is greatest
for previously provoked individuals who are exposed to mild triggering events.
Individuals in another study exhibited escalated aggressive behavior if they were
strongly, rather than weakly provoked (Kramer, Jansma, Tempelmann & Munte, 2007).
The studies differ on whether strong or weak provocation is likely to result in aggressive
behavior. Vasquez et al. (2005) share the perspective with a previous study (Pederson,
Gonzales, Miller, 2000) that similarly concluded that mild provocation may increase
aggression beyond strong, repeated provocation because the latter increases motivation
for escape or avoidance. The studies’ divergent findings could also be explained through
the type of provocation used to incite aggression. Vasquez and colleagues provoked
individuals through direct insult and degradation, while Kramer and colleagues provoked
individuals in a more passive manner by “punishing” them with high noise levels. The

25

former provocation circumstance challenged individuals on a personal level while the
latter was part of a more trivial game. Despite potentially conflicting conclusions, results
from both studies are consistent with the GAM’s theoretical framework. Provocation
affected aggressive behavior through cognition, affect and arousal.

Kramer et al.’s

(2007) study demonstrated these effects by monitoring which parts of the brain were
activated by provocation.
Aggressive cues are objects that prime aggression-related concepts in memory and
are particularly applicable to studies involving the channels through which media impact
aggressive behavior.

The GAM indicates that input variables such as aggressive

behavioral cues may influence later aggression by temporarily augmenting hostile
feelings and hostile thoughts.

Berkowitz and LePage (1967) found that the mere

presence of guns (versus badminton rackets and shuttlecocks) increased the aggressive
behavior of angered research participants. Individuals administered more electric shocks
to another person if they were exposed to the weapon than if they were not exposed to the
weapon. Other studies have also found that the presence of guns or other aggressive cues
increase the accessibility of hostile thoughts through automatic priming (Anderson et al.,
1998; Bartholow, Anderson, Carnagey & Benjamin, 2005; Meier, Robinson &
Wilkowski, 2007). Bushman and Geen (1990) similarly found that violent media can
affect hostile thoughts. After exposure to either a violent or a non-violent movie clip
from popular movies, participants were asked to write down their thoughts as they
watched the tape. Those exposed to the non-violent movie wrote fewer hostile thoughts
than those exposed to the violent movie.
Although the GAM provides an avenue through which aggressive behavioral cues

26

can affect behavior through both cognition (hostile thoughts) and affect (hostile feelings),
many studies have only empirically supported the latter. Some studies (e.g. Berkowitz &
LePage; 1967; Anderson et al., 1998; Bushman, 1998) looked for a purely cognitive route
to aggression from aggressive primes without considering affect, which may partially
account for the discrepancy. However, some studies that tested for both cognitive and
affective effects found that aggressive behavioral cues (e.g. a photo of a weapon or a
violent film) influence the cognitive route, but not the affective route to aggression
(Anderson et al., 1996; Anderson et al., 1998).

Despite this finding, other research has

revealed an important link between comfort and aggressive affect. A GAM study by
Anderson et al. (1996) found that uncomfortable temperatures increased state hostility but
did not affect hostile thoughts.
Some of the literature on this topic revealed both affective and cognitive effects.
A GAM study demonstrated through multiple experiments that violent song lyrics can
detrimentally affect internal processes. After exposure to violent song lyrics, individuals
felt more hostile and maintained more hostile thoughts than those who listened to a
similar but non-hostile song (Anderson, Carnagey & Eubanks, 2003). The effects of
exposure to violent film clips on both state hostility and hostile thoughts have been
demonstrated through multiple studies (Bushman & Geen, 1990; Anderson, 1997;
Anderson et al., 1995). After exposure to film clips varying in degree of violence,
individuals who viewed the most violent of the videos, which depicted killing and guns,
maintained more hostile thoughts and state hostility than those who viewed the other
videos (Bushman & Geen, 1990).
Two possibilities emerge through these discrepant findings. The first is that the

27

aggressive cues used in the studies that only measured cognition may also have
influenced affect but were unable to include the effect within the analysis because they
did not test the relationship. The second possibility is that the behavioral cues in question
would have only affected cognition, and that the route affected by the input variable
depends on the type of behavioral stimulus.
For example, studies finding only a cognitive effect (Berkowitz & Lepage, 1967;
Anderson et al., 1998) primed aggressive internal states through weapon cues while
studies finding both cognitive and affective outcomes (Bushman & Geen, 1990;
Anderson, 1997; Anderson, Dueser & DeNeve, 1995) primed cues through violent movie
clips. Hostile thoughts are frequently measured through self-reported desires for revenge,
and exposure to weapons could easily amplify this desire by activating aggression-related
memory concepts by visually providing a mean through which one can enact revenge.
Alternatively, exposure to a weapon cue would not necessarily increase an individual’s
angry feelings. Violent films, on the other hand, may affect both cognitive and affective
channels because people often identify with and become attached to media characters
through multiple mechanisms including parasocial interactions.

In many cases,

individuals may have formed revenge-related desires and feelings of anger on behalf of
the media character or because they shared common life experiences with the media
character.
While most aggression-related research uses either violent or non-violent stimuli,
the current study proposes that similarly to the manner through which exposure to
aggressive stimuli influence aggressive behaviors, exposure to anti-aggressive stimuli
may prove useful for curtailing aggressive behaviors by activating anti-aggressive

28

concepts in memory. Previous literature that has demonstrated the utility of pro-social
stimuli in affecting behavior provides support for this proposition (Huston-Stein and
Friedrich, 1973; Greitemeyer, 2009; Greitemeyer & Osswald, 2010).
Person factors, or individual differences, comprise the second component of the
GAM.

They include the characteristics that an individual brings to a potentially

aggressive situation, such as personality traits, attitudes, beliefs, values and genetic
predispositions. Recent GAM applications have explored how person factors influence
the effects of situational factors such as violent imagery on aggressive behaviors. Several
studies have highlighted the importance of trait aggression as an individual difference for
its influential power on multiple aggression-related mechanisms. Trait aggression, or
trait hostility, refers to “an individual’s characteristic pattern of aggressive behavior
across development” (Kirsh, 2006, p. 286).
Previous research has demonstrated that trait aggression increases state hostility
as well as hostile thoughts (Bushman & Geen, 1990; Lindsay & Anderson, 2000;
Anderson, 1997). Research not only emphasizes the importance of trait aggression in
predicting aggressive behavior, it also suggests that factors within the GAM’s theoretical
framework may disproportionately affect individuals based on their level of trait
aggression.

For example, exposure to aggressive stimuli may not affect high trait

aggression individuals. In a particular instance, watching violent movie clips increased
hostile thoughts for low trait aggression individuals, but had little to no impact on those
classified as high in trait aggression (Anderson, 1997). Anderson suggested that high
trait aggression individuals may be so chronically primed for aggressive stimuli that
exposure to aggressive primes has little impact on them. In other words, aggressive

29

stimuli may generate hostile thoughts and anger for low trait aggression individuals,
whereas high trait aggression individuals may not need the added stimulus.
Attitude and self-efficacy represent other important individual difference variables
for aggressive behaviors in the GAM. Attitude refers to “a psychological tendency that is
expressed by evaluating a particular entity with some degree of favor or disfavor (Eagly
& Chaiken, 1993, p. 1). Similarly to the Theory of Planned Behavior (Ajzen, 1985), the
GAM proposes that an individual’s attitude significantly impacts behavior, such that
individuals who maintain positive attitudes towards violence are more likely to engage in
violent behaviors. To illustrate, a study involving parental aggression against umpires at
baseball games found that parents who maintained hostile attitudes were more likely to
behave aggressively towards umpires (Hennessy & Schwartz, 2007). Attitudes have also
demonstrated predictive utility in bullying research. Survey-based school studies have
found that participants who report negative attitudes towards bullying are significantly
less likely to emerge as bullies (Pellegrini, Bartini & Brooks, 1999; Salmivalli & Voeten,
2004).
Not surprisingly, studies investigating traditional bullying have found that most
school-aged children and adolescents think that bullying is an inappropriate course of
action and they maintain attitudes in clear opposition of the behavior (Boulton, Bucci &
Hawker, 1999; Randall, 1995). However, these studies did not measure attitude in terms
of potential outcomes of the behavior, but in terms of prescriptive social norms that
reflect the individual’s perceived morality of bullying.

Aspects of more general

behavioral theories such as the Social Cognitive Theory (Bandura, 1986) and the Theory
of Planned Behavior (Ajzen, 1985) paint a more detailed picture of the motivation behind

30

attitudes.

These theories suggest that attitudes about behaviors are influenced by

outcome expectations in which individuals evaluate likely rewards or consequences of
engaging in a particular behavior. For example, an individual will more likely develop a
positive attitude towards aggressive online behavior if she believes that her actions will
produce no consequences or that the beneficial outcomes will outweigh potential
consequences.
An outcome expectation thought process resulting in a positive conclusion will
likely result in a positive attitude toward online aggression. This process is extremely
useful in understanding the route through which an individual’s attitude may affect their
behavior, yet it is potentially problematic in measuring the construct by any method other
than through reflective self-report. The fact that many aggressive behaviors are not
planned but impulsive, may account in part for why some studies involving traditional
bullying have found only moderate correlations between pro-bullying attitudes and
bullying among school children (Salmivalli & Voeten, 2004). Especially in the case of
retaliatory aggression, an individual may maintain a generally negative attitude towards
aggressive behaviors, but a specific, time-sensitive situation in which the individual has
been provoked may produce a dissimilar attitude from that which their general attitude
would predict.
The heart of this study focuses on retaliatory, rather than unprovoked aggression.
Topalli and O’Neal (2003) refer to retaliatory aggression as a transient aggressive state in
which the perceiver must “1) experience anger, 2) judge the egregious behavior of the
provoker to have been deliberate, and 3) believe the act of retaliation to be achievable and
pending.” The combination of anger and perception of deliberate hurtful behavior by the

31

provoker may temporarily overshadow the individual’s general attitude towards
aggressive behavior. Furthermore, impulsive responses to provocations likely do not
account for an individual’s general attitude towards aggression as fully as do carefully
planned responses.
Similarly to Bandura’s Social Cognitive Theory, the GAM proposes that selfefficacy beliefs are important antecedents to aggressive behavior.

Individuals who

believe that they can successfully engage in aggressive acts and that those actions will
produce desired outcomes are more likely to engage in aggressive behaviors than those
who believe otherwise (Anderson & Bushman, 2002).
For example, someone who believes that he cannot win a fight against a bigger or
stronger attacker will likely resist the temptation to engage in physical brawl. The same
individual may, however, engage the attacker in a verbal dispute if he believes that he can
outwit or emotionally hurt the individual. While electronic communication eliminates
some of the physical barriers presented by traditional bullying, low levels of self-efficacy
may persist in deterring a physically weak adolescent who has considered cyberbullying
as a means of retaliation against a real-world bully from performing the behavior. In all
likelihood, self-efficacy has the potential to both promote and restrain online aggression.
It can protect individuals from becoming recurring victims of bullying, but it can promote
the belief within potential bullies that they can successfully engage in the aggressive
behavior. Although self-efficacy is generally described as an important antecedent to
aggressive behaviors, most GAM literature in which it appears reference its importance
without empirically testing its utility.
Existing GAM literature has identified trait aggression, attitude and self-efficacy

32

as important individual differences for impacting aggression.

The theory clearly

recognizes person factors such as these as important antecedents to aggressive behaviors.
However, to the author’s knowledge, an individual’s overall ability to successfully
engage in self-control has been neither explicitly identified nor fully explored within the
GAM. Trait self-control refers to the capacity to exert control over the self by the self
(Baumeister & Exline, 2000) and involves an individual’s ability to consider alternatives
before acting, to engage in self-discipline, and to forgo present pleasurable experiences
that will prove detrimental in the long-run. This study posits that an individual’s self
regulatory capacity should impact aggressive behavior through influencing multiple
components of the GAM, including the individual’s ability to engage in state self-control.
Specifically, high levels of self-control should facilitate effective self-observation, such
that the GAM components that lead to aggressive behavior, such as hostile cognition and
affect, are counteracted by self-discipline and the individual’s ability to consider potential
future consequences and restrain violent impulses accordingly.

Outcome Variables
The final stage of the GAM model is outcomes, which includes appraisals and
decision processes. In this process, an individual may engage in an immediate appraisal
which is automatic, or a reappraisal which involves searching for an alternative view of
the situation. The former should result in an impulsive action, while the latter should
result in a thoughtful action (Anderson & Bushman, 2002). The current study proposes
that state self-control occurs during the reappraisal process. The study also proposes that
individuals who are high in trait self-control will more likely engage in the reappraisal

33

process than in the immediate appraisal process, and that they will more likely refrain
from aggressive behaviors online.

A New Variable for the GAM: Self-Control
The concept of self-control closely relates to sensation-seeking, impulsivity and
consideration of future consequences which previous GAM research has linked to
aggression. In a survey-based methodology guided by the GAM, Joirman and colleagues
(2003) found that aggression was positively related to sensation-seeking and impulsivity
and negatively related to consideration of future consequences. They also experimentally
manipulated anticipated consequences of an aggressive act. After experiencing criticism
from an experimenter, participants were offered the opportunity to negatively evaluate
the individual with whom they were told they would have no future interaction, possible
future interaction, certain future interaction or certain immediate interaction. Aggression
was most evident in cases with diminished perceived consequences, such that individuals
believed they would have little to no future interactions with the evaluator.
Perceived consequences likely play a substantial role in anonymous forms of
online aggression. Individuals who aggress anonymously online most likely consider
future consequences to be relatively low or non-existent because the victim is unaware of
their aggressor’s identity. Perceived consequences generate potential motivation for selfcontrol, yet the self-discipline dimension of self-control may still inhibit aggression in
cases with few consequences for the perpetrator. In instances of online aggression when
adolescents contemplate whether to openly bully their classmates, consideration of future
consequences likely inhibits more individuals. For example, when cyberbullies perceive

34

their identities as known, they may perceive the risk of potential retaliation, parental
consequences, school punishments, or even possible legal penalties depending on the
severity of the behavior. The perception that engaging in socially undesirable behaviors
could ultimately produce seriously negative future consequences can significantly impede
aggressive behavior if the individual can control impulsive inclinations.

However,

behaviors are occasionally outside of an individual’s volitional control.
Impulsivity and disinhibition present two constructs that undermine an
individual’s volitional control. Together they embody two related factors that are said to
typify adolescence (Pechmann, Levine Loughlin & Leslie, 2005).

Applying these

measures to behavior, highly impulsive individuals frequently take actions without
carefully thinking through or considering possible consequences. Convincing impulsive
cyberbullies to modify their behaviors may prove to be particularly challenging because
rationally appealing to their fears with potential consequences will unlikely obstruct
negative behaviors.
Disinhibition is a dimension of impulsivity which is characterized by a low
capacity to control response inclinations (Colder & Stice, 1998) and weak inhibitory
control leads adolescents to pursue risky behaviors (Pechmann et al., 2005).
Understanding the role that impulsivity and sensation-seeking play during adolescence
may be of particular importance to understanding online aggression. Both of these
characterizations of impulsivity suggest that an individual's level of control or regulatory
abilities over their own actions are key to understanding why some individuals tend to be
more impulsive than others.

Indeed, disinhibition has emerged as one of the most

significant predictors of physical aggression in prior research (Joireman et al., 2003).

35

Some adolescents are more impulsive than others and are therefore more likely to engage
in problematic behaviors (Pechmann et al., 2005) such as cyberbullying. However, the
asynchronous nature of the Internet, which affords individuals with time to think without
the pressures of a real-time audience, may suppress impulsivity in ways that it is not
suppressed in traditional bullying.
Adolescents are less likely than adults to use cognitive strategies to control their
behaviors and are less able to regulate emotions (Gross & Levenson, 1997). Pechmann
and colleagues (2005) suggested that weak inhibitory control leads adolescents to pursue
risky behaviors and that some adolescents are more likely to engage in problematic
behaviors because they maintain high levels of impulsivity. Their review of adolescence
and impulsivity noted that neurobiological changes occurring during adolescence cause
adolescents to be more impulsive than adults.
Impulsivity should be considered in the case of cyberbullying if for no other
reason than that the majority of perpetrators are young individuals. Colder and Stice
(1998) measured impulsivity through multiple items including the following: “Before
making a decision I usually try to consider all sides of an issue,” and “I rely on careful
reasoning before making up my mind.” Measuring dispositional self-control in a similar
manner, Finkel, DeWall, Slotter, Oaten and Foshee (2009) found that high self-control
related to significantly fewer acts of interpersonal violence.
Pechmann et al. (2005) make a clear case for understanding the importance of
impulsivity in behavior during childhood and adolescence. However, impulsivity and
disinhibition do not affect all adolescents equally. Therefore, it cannot be assumed that
all adolescents, or even the majority, frequently behave aggressively online for impulsive

36

or irrational reasons. Like self-control, impulsivity can vary across settings and domains
of behavior; however, many impulsivity studies treat the construct as a trait.

For

example, Colder and Stice (1998) found that impulsivity was more strongly associated
with delinquent behavior for males than for females. This gender-based difference in
impulsivity may also be closely related to why most aggression studies (e.g. Bandura et
al., 1961; Kowalski & Limber, 2007) identify males as exhibiting more aggressive
behaviors than females. Colder and Stice’s research suggests that males are less likely to
control their impulses, which results in more gender specific delinquent and aggressive
behavior. Multiple descriptive studies (Agstston et al., 2007; Smith, Madavi, Carvalho,
Tippet, 2006; Kowalksi, Limber & Agatston, 2008) have found that a significant
proportion of online aggressors are females. Interpreting Colder and Stice’s findings
through this perspective suggests that females’ problematic behavior likely does not
result from impulsivity alone.
Furthermore, impulsivity does not equally affect all adolescent age groups.
Research related to impulsivity and age has shown that impulse control and self-control
improve gradually with age and experience, developing fully in late adolescence and
early adulthood (Pechmann et al., 2005). Age implications are particularly relevant to
online aggression because cyberbullies tend to be high school-aged rather than middle
school-aged which contrasts much of the bullying that occurs in traditional environments
(Ybarra & Mitchell, 2004; Hinduja, 2008). Under this assumption, many if not most
cyberbullies are older and in the latter stages of adolescence.

At this stage of

development, impulsivity should play a less substantial role because an individual’s level
of self-control should be able to better counteract many impulsive desires.

37

The affordance of time after a provocation likely promotes self-control and
consideration of possible consequences, thereby restraining impulsivity and aggressive
behaviors. A study involving the effects of self-control on violence in interpersonal
relationships manipulated response time after provocation by asking participants to
respond either immediately after being provoked or to respond after 10 seconds (Finkel et
al., 2009). Participants were more likely to exhibit signs of aggressive interpersonal
violence when they responded immediately after the provocation than if they were
afforded a 10 second delay.
Time pressure situations that require immediate responses likely restrict one’s
ability to effectively engage in self-control if self-control is external to the individual’s
automatic tendency.

For example, the immediate response demanded from a time

sensitive offline provocation may obstruct one’s self-regulatory capabilities. Finkel and
colleagues’ research clearly demonstrates that people select more constructive courses of
action when they are afforded time to think. Aside from instances in which individuals
engage in forms of synchronous Internet communication, the self-regulatory capabilities
of those who are provoked online have the potential to benefit from the asynchronous
nature of the online environment if they do not immediately respond to their attackers.
The Internet provides an opportunity that is generally not granted to individuals who are
provoked offline: the opportunity to think through alternatives before acting and to
consider potential future consequences.
This chapter focused specifically on understanding online aggression through
theories of media research with a particular focus on the General Aggression Model. It
looked specifically at why the construct of self-control should be added to the theoretical

38

model, and it made a case for why the concept is important to aggressive behaviors.
Chapter 3 discusses hypotheses related to the GAM by taking a closer look at the roles of
self-control and the activation of visual behavioral stimuli in online aggression.

39

Chapter 3
Hypotheses

Self-Control
Civilization is often defined by the ability of citizens to conform to societal
demands by inhibiting their undesirable, antisocial impulses.

Without self-control

reinforced by personal and collective standards, an organized society would experience
extreme chaos. Societal confines and individual self-control stop people from stealing,
killing, cheating and engaging in other forms of socially undesirable behaviors. Despite
societal boundaries and social desirability, the capacity to control one’s self varies
between individuals. Most people tend to lose control when they fail to actively monitor
their actions. Some people succumb to antisocial behaviors because they fail to control
their impulses, but others are better able to regulate their instinctual inclinations.
Literary accounts of the behavior use the terms self-control and self-regulation
interchangeably (Baumeister & Vohs, 2003; Finkenauer et al., 2005). Self-control refers
to one’s capacity to inhibit and override socially unacceptable and undesirable impulses
through altering and regulating one’s behavior, thoughts and emotions (Baumeister,
Herton & Tice, 1994; Carver & Scheier, 1998; Muraven & Baumeister, 2000 &
Finkenauer, Engels & Baumeister, 2005). Consideration of its definition alone provides a
potential avenue through which self-control may interact with existing GAM variables to
inhibit socially undesirable behaviors such as online aggression. Situational factors such
as provocation affect aggressive behaviors through cognition and affect. While the GAM
proposes that thoughts and emotions contribute to increased aggressive behavior, a

40

conceptual understanding of self-control provides that one can regulate their thoughts,
emotions and behavior. The relationship has not been tested; however, in theory, selfregulatory capacity should assist individuals with negative cognitive and affective states
by inhibiting detrimental thoughts and emotions leading to aggressive behaviors.
While some people possess a natural inclination to control undesirable tendencies,
others tend to exhibit their self-regulatory deficiencies whenever an opportunity arises.
Rigid societal confines help individuals regulate potential desires for extreme forms of
socially undesirable behavior such as murder. Many of the same individuals may find it
harder to regulate their desires to stay on the couch with a bag of potato chips when they
believe that they should be exercising, which illustrates that self-control is domain
specific.
Successful exertion of self-control requires the acquisition of resources that can
promote desirable behaviors by inhibiting a variety of undesirable ones. These resources
include personal standards, an active monitoring process and the capacity to alter one’s
behavior (Bandura, 1991; Baumeister, 2002). For example, developing and maintaining
strong personal standards against undesirable behaviors such as online aggression can
help to inhibit undesirable behaviors. The ability to actively monitor personal standards
in the presence of frustrating situations reinforces the strength of personal standards. The
monitoring process alerts the individual to when personal standards need to be adhered
to. Finally, the capacity to alter one’s behavior is needed to successfully adhere to
personal standards after becoming aware. Failure of any of these variables can undercut
self-regulation.
As a person factor, self-control may effectively impede aggressive behavior by

41

modifying the relationships between multiple components of the GAM. For example, the
GAM specifies that situation factors, such as violent stimuli or provocation, can lead to
increased hostile cognition, affect, and arousal through the activation of aggressive
scripts. Someone with a high self-regulatory capacity, whose personal standards run
contrary to exhibiting anger and highly hostile thoughts, may use self-control to
counteract the effects of the aggressive situation by maintaining low levels of hostile
thoughts and anger. Further down the line in the GAM model, self-control may also be
used to moderate the effects of the present internal state variables on aggressive behavior.
The GAM suggests that present internal state variables such as state hostility and hostile
thoughts lead to aggressive behavior. An individual may find it difficult to completely
block the development of hostile thoughts and anger after experiencing an aggressive
provocation or after exposure to violent stimuli. However, for the individual whose
personal standards run contrary to aggressive behavior, the monitoring process
component of self-control that alerts the individual to when personal standards need to be
adhered to may help the individual to refrain from aggressive behavior, despite exhibiting
high levels of hostile thoughts and anger.

Self-control is considered an important

precursor to many behaviors that require the inhibition of impulses.

Many people

instinctually respond to aggressive provocations with equally aggressive behavior.
Walking away from a situation after provocation or confronting the perpetrator in a
constructive, rather than destructive manner requires the exertion of willful self-control.
While lacking empirical support within the tradition of GAM research, selfcontrol likely contributes substantially to aggressive and impulsive media-related
behaviors such as online aggression. Previous research has revealed a negative link

42

between self-control and anger. Individuals with high levels of self-control were less
angry in general than low self-control individuals in Tangney et al.’s (2004) study.
Indeed, trait aggression has emerged as one of the most important predictors of
aggression related processes (Bushman & Geen, 1990; Lindsay & Anderson, 2000;
Anderson, 1997). Baumeister et al.’s finding supports the proposition that the effects of
self-control on the processes leading to aggression may be significant.
Offline aggression research has demonstrated the utility of self-control in the
domain of hostility by linking bullying to cost and benefit perceptions in combination
with self-control (Archer & Southall, 2009), but the literature suggests that the ability to
engage in successful self-control is not domain specific. An individual who maintains a
high self-regulatory capacity should be able to exert control not only over their
aggressive behavioral impulses but also over impulses that promote dissimilar behaviors.
Highly impulsive individuals exhibit low levels of trait self-control by frequently
engaging in actions without carefully thinking about or considering possible
consequences.

Indeed, self-regulatory failure may be difficult to avoid for many

adolescents who are more impulsive than their more mature counterparts. Impulsivity is
sometimes viewed as the simple inverse of self-control. However, from Baumeister and
Bandura’s conceptualization of self-control, impulsivity, along with others such as
sensation-seeking, comprises elements of a larger self-regulatory framework.

While

interrelated, the totality of self-control is more complex than any of its singular
components. For example, a study investigating low self-control and imprudent behavior
reported that for some of the deviant behaviors examined, the sensation-seeking
component of self-control was more predictive than was the more inclusive self-control

43

scale (Arneklev, Grasmick, Tittle & Bursik, 1993). Another study alternatively found
that self-regulation predicted child adjustment problems while impulsivity alone did not
(Lengua, 2003).

Accordingly, studies that have measured both self-control and

impulsivity present impulsivity and self-control as separate and distinguishable factors, or
they present impulsivity as one of multiple dimensions of self-control (Piquero & Rosay,
1998).
Self-control becomes increasingly difficult when provoked.

Participants in

DeWall, Baumeister, Stillman and Gailliot’s (2007) study who were low in trait selfcontrol were more likely to express intentions of behaving aggressively after provocation
than those who were high in trait self-control. Similarly, Avakame’s (1998) research
involving the effects of self-control on conjugal violence measured both physical
violence and psychological aggression, where psychological aggression constituted
emotional or verbal maltreatment.

Self-control emerged as a powerful predictor of

physical as well as psychological aggression in the study, where both types of aggression
increased for males and females as their self-control weakened. Despite occurring in the
offline world, these cases indicate that aggression is more likely to occur when
individuals do not actively engage in self-control. Furthermore, those who monitor their
online behaviors should be able to overcome impulses to engage in aggressive behaviors
online.
Research has demonstrated that a high self-regulatory capacity can counter a
variety of personal and interpersonal problems associated with low self-control (Tangney
et al., 2004). Aggression is among these behavioral problems. A growing body of
research suggests that low self-control leads to aggressive behavior; however, the

44

relationship has yet to be fully examined within the context of media effects on
aggression and has yet to be embedded within a larger theoretical framework such as the
GAM.
Self-control has been linked to a variety of interpersonal problems such as
destructive relationships, poor interpersonal skills and less optimal emotional responses
such as aggression (Finkel & Campbell, 2001; Tangney et al., 2004). Combating the urge
to destructively fight with someone after being wronged is difficult. It often requires the
consideration of potential long-term consequences and setting aside pride to
constructively respond to the situation.
To illustrate, Finkel and Campbell (2001) demonstrated that high self-control can
help people overcome many of the impulsive tendencies that lead to romantic break-ups.
In the face of relationship problems, high self-control individuals in their study were
more likely to work against break-up or avoidance tendencies by confronting the
situation. They expressed more of a desire to rationally confront problems through
compromise than to exit the relationship. High self-control participants in Tangney and
colleagues’ (2004) study similarly expressed more of an inclination to constructively
approach anger by engaging in rational discussion with their target and less of an
inclination to grow increasingly angry by ruminating about anger.
In short, high trait self-control individuals exhibit higher capacities for managing
anger and frustration. Mulling over anger likely increases the GAM’s present internal
state variables of hostile thoughts and state hostility, which increase aggressive behavior.
The literature has identified control of thoughts as one of the major domains of selfcontrol (Finkenauer et al., 2006). Tangney et al.’s (2004) findings present a potential

45

inhibitory relationship for the GAM. If high self-control can impede the progression of
hostile thoughts, state hostility and other internal state factors that contribute to
aggressive behavior once they develop, then high self-control may inhibit the ability of
these factors to contribute to aggression. More simply, there should be a) a positive
relationship between the present internal state variables and aggressive behavioral
intentions and a greater positive relationship between b) hostile thoughts, c) state
hostility, d) perceived discomfort, e) arousal and aggressive behavioral intentions among
those with lower trait self control than among those with higher trait self-control
(Hypotheses 1a-1e).

State Self-Control
A study involving the influence of impulsive precursors on behavior identified
self-control as an important moderator.

Across three experiments, self-control

significantly moderated the effects of automatic affective reactions on impulsive eating
and drinking behaviors (Hofmann, Friese & Strack, 2009). Within the context of the
study, self-control was examined as a product of temporary ego-depletion rather than as a
stable personality trait.
Many current self-regulation scholars consider self-control not only as a stable
trait but as a finite resource which can be depleted. Self-control can vary not only as a
trait but as a state. Studies that examine low self-control as a function of ego-depletion
tend to focus on the latter rather than the former. The construct is often divided into two
categories, and this study similarly proposes that the behavior should be considered as
two separate, but interacting components: long-term trait self-control and the immediate

46

behavior of state self-control.
State self-control involves inhibiting and overriding socially unacceptable and
undesirable impulses through altering and regulating one’s behavior, thoughts and
emotions in a particular moment. Within the context of the current research, state selfcontrol refers to the willful exertion over the self to refrain from one’s inclination to
engage in retaliatory aggressive behavior despite provocation. The literature tends to
isolate state self-control from trait control by examining one or the other without
regarding their relationship to each other, but trait self-control should affect what an
individual does in a situation in which effective self-regulation is required. Most selfregulatory depletion research relies on measures of observed trait self-control, such as an
individual’s persistence in completing a frustrating puzzle after being subjected to prior
self-regulatory depletion.
According to a strong and convincing line of research, exerting self-control in one
area depletes one’s ability to exert self-control in another area (Baumeister, 2002; Vohs,
Baumeister & Ciarocco; 2005; Stucke & Baumeister, 2006; Vohs, Baumeister,
Schmeicel, Twenge, Nelson & Tice, 2008). For instance, two out of three experiments in
a study designed to test the implications of self-regulatory depletion found that
participants who exerted self-control in one area performed worse than their counterparts
in subsequent tests of self-control (Muraven & Slessareva, 2003). Self-control depletion
activities frequently involve engaging in cognitive tasks such as thought or memory
suppression.

Subsequently, self-control is often measured through persistence to

complete a frustrating task.

Despite relating to disparate self-regulatory behaviors,

exercising self-control in one domain usually depletes one’s ability to engage in self-

47

control in the second domain.
In light of prior research that demonstrated that suppressing a natural emotional
response is more difficult than expressing the natural emotional response (Baumeister et
al., 1994; Baumeister et al., 1998), Finkel and Campbell exposed participants to a
hypothetical dilemma initiated by a hypothetical partner and experimentally manipulated
self-regulatory strength depletion based on emotional suppression (high depletion) and
emotional expression (low depletion). Consistent with self-regulatory depletion research,
those with high self-regulatory depletion were less likely to accommodate their
hypothetical partner’s behavior by working to constructively accommodate the behavior.
They were instead more likely to meet their partner’s destructive behavior with additional
unhelpful behavior.
Self-control requires that individuals avoid responding to immediate stimulus
environments so that they can pursue long-term strategies that produce significant, but
delayed benefits.

However, research continuously illustrates that individuals with

depleted self-regulatory strength are less able to avoid responding to stimulus
environments so that they can reap delayed benefits. If self-control can indeed be easily
depleted, as much of the research in the field of self-regulation suggests, what are the
implications for individuals’ ability to control their own behavior in difficult situations?
Self-control can be effectively used to curtail incidents of online aggression only if
individuals who have experienced self-regulatory depletion can overcome their depleted
states. Research suggests that this goal can be accomplished through at least two avenues:
training and motivation. Although not specifically addressed within the GAM, training
and motivation may work behind the scenes through the model’s cyclical nature, as

48

depicted in Figure 1. The model posits that the outcome of each potentially aggressive
social encounter feeds back into the individual’s personality traits, thereby influencing
the individual’s behavior in future similar situations.

The decisions made by the

individual after each potentially aggressive social encounter feed back into the
individual’s personality traits such as self-control. Uncomfortable consequences of a
previous social encounter may motivate an individual to engage in more successful selfregulatory control in future potentially aggressive situations.

From the training

perspective, the GAM would imply that trait self-control is strengthened when the
individual uses willful restraint in social encounters, whereby individuals essentially train
and strengthen their self-regulatory capacity.
Recent evidence suggests that engaging in self-regulatory exercises can strengthen
self-regulatory abilities (Baumeister, Gaillot, DeWall & Oaten, 2006).

Finkel et al.

(2009) assigned participants to one of two 2-week self-regulation bolstering regimens or
to a control condition in which they were not exposed to an intervention.

The

intervention sought to improve self-regulatory capabilities by bolstering ego strength over
time. The first intervention consisted of a physical regulation task that asked individuals
to use their non-dominant hand for mundane tasks such as eating and brushing their teeth.
The second intervention consisted of a verbal regulation task that asked individuals to
regulate aspects of their habitual speech processes such as avoiding sentences that begin
with the word ‘I’, saying the word ‘yes’ instead of ‘yeah’ and refraining from use of
abbreviated speech and shorthand terms. Inclinations of interpersonal violence declined
for individuals who participated in both the physical and the verbal self-regulation
regimens, but not for those who belonged to the control condition.

49

The implications of Finkel et al.’s (2009) self-control interventions and others that
have presented similar conclusions (Baumeister et al., 2006) are profound.

They

collectively demonstrate that by consciously and deliberately engaging in activities that
counteract impulses, that individuals can effectively increase their self-regulatory
capacity thereby potentially decreasing their involvement in risky behaviors through state
self-control.

Like strengthening a muscle, giving sufficient attention to the self-

regulatory resource will increase its power and its utility.
One of the most intriguing dimensions of these interventions is that the regulatory
regimens required of the participants were not behavior specific such that they only
improved self-control over the behaviors performed during the intervention. Eating with
a non-dominant hand and avoiding certain types of sentences are completely unrelated to
abstaining from interpersonal violence after provocation other than that all of the
behaviors in question required the individual to successfully engage in self-control.
Participation in a comparable self-regulatory regimen should similarly inhibit incidents of
online aggression after provocation for individuals with aggressive inclinations. Thus,
individuals who continually and consciously engage in effective state self-control should
also be able to build their long term self-regulatory capacities. Through this avenue,
individuals can overcome incidents of self-regulatory depletion because they will possess
more of the resource to deplete.
The second avenue through which self-regulatory depleted individuals can
overcome their depleted state involves motivation.

Studies have shown that when

properly motivated, an individual can overcome self-regulatory depletion and engage in
effective self-control.

Muraven and Slessareva (2003) proposed that self-control is

50

jointly determined by depletion and motivation. They demonstrated across three studies
that when depleted individuals were properly motivated, such that they believed that the
task at hand would help others or themselves that they were more likely to engage in
subsequent self-control than depleted individuals with low motivation.
These two avenues provide important implications for the utility of self-control
within the context of online aggression.

Individuals involved in online aggression

maintain varying degrees of trait self-control, and depending on the day or the moment,
situational factors may deplete their state self-regulatory abilities.

However, these

findings suggest that by practicing efforts of self-control, individuals may be able to
strengthen their trait self-control and overcome future depletions.

Furthermore, if

supplied or instilled with proper motivation, individuals are even more likely to control
themselves. For adolescents who openly engage in online aggression, motivation could
manifest itself through a variety of channels including, but not limited to, repercussions
by parents, school authority figures or friends, or isolation from their online community.
However, aggressive behavior online should be more prevalent when adolescents fail to
control their actions and self-regulatory capacity should impact one’s ability to engage in
state self-control. Specifically, aggressive behavioral intentions should be a) negatively
related to trait self-control, b) negatively related to state self-control and c) trait and state
self-control should be positively related to each other (Hypotheses 2a-2c).
Additionally, state self-control should moderate the relationships between the
GAM’s present internal state variables and aggressive behavioral intentions. In other
words, there should be a greater positive linear relationship between a) hostile thoughts,
b) state hostility, c) perceived discomfort and d) arousal and aggressive behavioral

51

intentions among those with lower state self-control, than among those with higher state
self-control (Hypotheses 3a-3d).

Emoticons and Behavioral Cue Activation
Computer mediated communication lacks many of the non-verbal cues associated
with face-to-face communication. When someone receives a provoking message on the
Internet, the message likely lacks many of the non-verbal cues (e.g. facial expression and
body stance) that normally accompany similar face-to-face encounters. Non-verbal cues
often help to remove some of the ambiguity of a message and they assist the message’s
receiver with interpreting the provoker’s intention.
Internet users have learned to enhance many of the limitations of text-based
applications such as email to achieve communication that parallels face-to-face
communication (Walther & Burgoon, 1992). Research indicates that computer mediated
communication users have overcome many of the obstacles that limit expression through
electronic media through emoticons. Emoticons refer to pictographics that are used in
computer mediated communication “to express emotion or as surrogates for nonverbal
communication” (Thompson & Foulger, 1996, p.226).

Emoticons are graphic

representations of facial expressions that are embedded in computer mediated
communication.

They depict emotion, such as happiness or anger, through visual

representation. Social Information Processing (SIP) Model suggests that users who are
unfamiliar with each other can achieve normal relationships online, but that it takes more
time than with traditional communication because the lack of nonverbal cues within the
environment limit the range of communication (Walther & A’ddario, 2001). Emoticons

52

serve to restore some of the emotion to what may otherwise come across as cold and
textual communication. They can also reinforce the message’s intended meaning. For
example, participants in a study that paired a happy facial expression with a positive
verbal message rated the combined image and message as more positive than the positive
message on its own (Derks and Grumbkow, 2007).
Other studies have examined the effects of emoticons on computer mediated
communication. Walther & A’ddario (2001) studied the impact of text-based emoticons
by presenting participants with either a negative or positive message coupled with a
happy facial expression, a sad facial expression, a winking facial expression or no facial
expression. They found that messages with any negative element, either verbal or visual,
rated significantly more negatively than messages without negative elements. Messages
accompanied by the wink or smile were rated as significantly more positive than any
other message combination, while emoticons did not significantly impact negatively
worded combinations. By pairing an unspecified emoticon (presumably a happy face)
with verbal hostility online, Thompson and Foulger (1996) found that an emoticon can
reduce a message’s perceived hostility. At the same time, the emoticon increased the
perceived hostility of more hostile messages. In a similar study, Derks and Grumbkow
(2007) found that when coupled with a smile emoticon, negative verbal computer
mediated messages conveyed less negativity than a negative message without the
emoticon for high school students. Their study also revealed that in combination with a
negative message, a happy facial expression generated more ambiguity than either a
positive message or negative message alone. The same was true for frowning facial
expressions paired with positive messages.

53

In combination, these studies suggest that the incorporation of facial expression
emoticons can be extremely effective in conveying the sender’s emotion, but that
emoticons may not be able to counteract a message perceived as highly negative.
Participants exposed to highly antagonistic messages in Thompson and Foulger’s study
may have perceived the emoticon as a mocking rather than as a playful symbol. In both
cases, the emoticons were created through sideways text-based symbols rather than
through the symbols in current use such as yellow smiley faces. Derks and Grumbkow
did not vary the degree of negativity in negative messages, thus the messages may not
have been perceived as antogonistic enough when coupled with a happy emoticon to
increase participant anger. Discrepancies between Derks and Grumbkow and Thompson
and Foulger’s studies suggest that the ambiguity created by conflicting message and
emoticon valence will be perceived as positive when the message is mildly negative but
as negative when the message is perceived as highly negative. This perception is likely
based on the message receivers attempt to interpret the sender’s seemingly conflicting
meaning.
Much of the ambiguity generated through messages accompanied by contradicting
messages from emoticons and verbiage likely stemmed from the fact that the emoticon
came from the message’s source. The counteracting impact of the positive emoticon
paired with the negative message described by Derks and Grumkow and Thompson and
Foulger to some extent, may be more effective when the emoticon comes from a neutral
source. Outside of research on computer mediated communication, literature indicates
that facial expressions can serve not only as emotion messengers that reveal specific
information about a sender’s intention, but as emotion elicitors that stimulate positive

54

affect because they simply represent positive stimuli in the same way that an image of a
puppy may stimulate positive affect (Ruys & Stapel, 2008). Priming through facial
expressions has been examined to a certain extent, and priming in general has been
examined extensively within media research. Priming embodies an important component
of the GAM through the proposition that situational priming cues can affect an
individual’s internal state and subsequently their behavior.
A study investigating the subliminal effects of priming emotional facial
expressions found that exposure to angry, disgusted, or fearful faces elicited negative
moods as well as increased the likelihood of performing actions corresponding to the
expression (Ruys & Stapel, 2008). Negative emotions were more likely to develop when
facial expressions flashed super-quickly (40 milliseconds) rather than quickly (120
milliseconds), and actions corresponding to the emotion (e.g. kicking the lamppost after
exposure to the angry face) were more likely when the facial expression flashed quickly
rather than super quickly. This facial expression study differs to an extent from the
computer mediated communication facial expression studies because they were not
combined with negative messages.

However, findings from the subliminal facial

expression study suggest that when facial expressions are not linked to a particular
message or sender, that they can independently affect emotion and possibly behavior.
From the perspective of the GAM, behavioral cues such as facial expressions should
impact aggressive behavior by influencing affective state and/or cognition.
Studies involving physical violence and face-to-face aggression propose that
violence is usually triggered by salient aggressive and contextual cues (Denson et al.,
2007; Siever, 2008). Research suggests that schematized views of the world formed to

55

help people cope with the flood of new information that they constantly receive account,
at least partially, for the impact that violent media have on aggressive behavior.
Developing stereotypical models is not merely a subconscious action; it occurs
automatically and inescapably (Coleman, 2003). Kawakami, Spears and Dovidio (2002)
referred to automatic processes as mental activities that occur outside of awareness in a
largely involuntary, unintentional, and effortless manner. Mere exposure to a symbol or
representative of a social category can activate schematic associations and can
subsequently influence behaviors (Bargh, Lambardi & Higgins, 1998). Behavioral cue
activation refers to the extent to which an associated schematic belief is accessible in
one’s mind (Kunda & Spencer, 2003). Schematic associations are easily made because
they are so well-learned that they become part of the status-quo. Conscious subscription
to a schematic knowledge structure can potentially fuel associated behaviors. However, a
general consensus among contemporary attitude researchers concedes that implicit
attitudes may be automatically updated by the attitude object or by cues associated with
the object, even when the person does not have conscious access to an attitude (Eagly &
Chaiken, 1993).
Behavioral stimulus exposure does not influence all individuals equally.
Behavioral stimuli used in stereotype research clearly demonstrate this notion. Devine
(1989) found that despite possessing an equal knowledge-base regarding cultural
stereotypes, people higher in prejudice endorsed the activated stereotypes more strongly
than did low prejudice people. While most members of a culture are generally exposed to
the same deeply vested societal schematic knowledge structures, a person’s willingness
or lack thereof to endorse these schemas may determine the extent to which they apply

56

them to behaviors. Similarly, a recent study found that hostile thoughts were primed by
hunting guns for non-hunters, whereas they were primed by assault guns for hunters
(Bartholow, Anderson, Carnagey & Benjamin, 2005). Weapon imagery primed hostile
thoughts in both cases; however, individual differences determined which type of gun
most strongly affected cognition. One would expect that the individual’s experience with
or level of endorsement of a symbolic cue, which result from life experiences among
other factors, would moderate the relationship between the schema activation and
intentions to engage in online aggression.
As outlined in the GAM discussion, many aggression studies have manipulated
schematic associations through behavioral cues without particular regard to individual
differences.

These studies have demonstrated that stimuli can activate aggressive

components that influence hostile behaviors. Violent music, films, video games and
television have been shown to increase the likelihood of both short- and long-term
aggressive behavior (Anderson et al., 2003). Physically aggressive behavior, verbally
aggressive behavior, hostile thoughts and aggressive emotions have all been linked to
short-term exposure of violent stimuli (Anderson et al., 2003). However, what happens
after exposure to nonviolent, rather than a violent, stimuli? Behavioral stimuli can just as
easily reflect positive associations as they can reflect negative associations. A peace
symbol, for example, is associated with non-aggression rather than with aggression.
A long line of research indicates that violent and aggressive stimuli can adversely
affect aggressive behavior, but some studies testing anti-aggressive or pro-social stimuli
have found that exposure to certain types of stimuli can promote pro-social behavior.
Huston-Stein and Friedrich’s (1973) work on the impact of aggressive vs. pro-social

57

television viewing on children represents an early example of media effects on pro-social
behaviors. After exposure to 12 half-hour episodes of Batman or Superman cartoons or
Mister Rogers’ Neighborhood, low socio-economic children exposed to the latter
increased pro-social helping behavior in the classroom and on the playground. More
recently, a series of four experiments testing the effects of video game content on
behavior demonstrated that exposure to pro-social video games promotes, at least in the
short-term, pro-social behavior (Greitemeyer & Osswald, 2010). Playing a pro-social
game such as City Crisis led individuals to help an individual who spilled pencils on the
floor, to assist a master’s student who needed help with her thesis and to intervene when
an angry boyfriend yelled at his girlfriend. Another study demonstrated an analogous
effect through pro-social song lyrics rather than through video games (Greitemeyer,
2009). Individuals who listened to songs with pro-social, rather than neutral lyrics were
more likely to exhibit helping behavior. The pro-social effect in Greitemeyer’s (2009)
study was mediated by interpersonal empathy, pointing to the importance of the cognitive
route in influencing behavior. This study lends credence to previous studies that have
similarly demonstrated that hostile song lyrics have the potential to affect an individual’s
internal state and their subsequent behavior by increasing the accessibility of state
hostility, hostile thoughts (Anderson et al., 2003), empathy or mood.
Prior research clearly suggests that behavioral cues, whether positive or negative,
can influence an individual’s behavior through activating the internal state. Depending
on the content, video games can lead to aggression or pro-social behavior as can song
lyrics. Images of weapons have activated aggressive behavioral processes in the past,
indicating the potential for anti-aggressive behavioral images to restrain aggressive

58

behavior.

In line with previous research, this study posits that those exposed to

aggressive behavioral cues should have a) the most hostile thoughts and b) the most state
hostility, followed by those who were not exposed to behavioral cues, and finally by
those who were exposed to anti-aggressive behavioral cues (Hypotheses 4a and 4b).
Trait self-control should also moderate the effects of the behavioral stimulus on hostile
thoughts and state hostility such that there should be a positive relationship between
aggressive stimulus exposure and c) hostile thoughts and d) state hostility among those
with low trait self-control, but no relationship among those with high trait-self-control
(Hypotheses 4c and 4d). Figure 2 illustrates the present study’s proposed hypotheses.

Figure 2 The General Aggression Model Redrawn with Hypothesized Relationships.

Inputs

Trait Self-Control
(Person Factor)

H4c & H4d

Behavioral Stimuli
Exposure
H4a &H4b

Routes
H2c

H1b- H1e

Present Internal State:
Affect Cognition
Arousal

H2a
H2a
H2a

State SelfControl

H3a – H3d H3d
H3a –

H1a
H1a

H2b
H2b
Behavioral
Intention

Outcomes

This study’s goal is to test whether an increase in state hostility and hostile
thoughts occur for those exposed to aggressive stimuli, and whether there is a decrease in

59

the effect on aggressive behavior for those exposed to anti-aggressive stimuli. Websites
contain many areas within their screen space that can be used to display behavioral
stimuli should they demonstrate utility in curtailing aggressive behavior.

Should these

symbols effectively curtail aggressive behaviors, they can be effectively incorporated into
online applications such as Facebook, online forums and email.
The present study’s original contribution is to examine online aggression through
a theoretical framework, to present trait and state self-control as valuable additions to the
GAM, and to ascertain the value of exposure to anti-aggressive behavioral stimuli in
restraining online aggression. The study outlined in the following chapters investigates
new GAM variables and studies the impact of pro-social priming with anti-aggressive
stimuli. Specifically, it examines the impact of trait self-control on aggressive online
behavior through the GAM’s present internal state variables. It explores the relationships
between state self-control, trait self-control and other GAM variables. It investigates the
effects of behavioral stimulus activation on state hostility and hostile thoughts and
finally, it explores potentially differences between off- and online aggression. Chapter 4
describes the method and operational variables used to test the hypotheses discussed in
this chapter.

60

Chapter 4
Method

Pretest of Stimulus Materials
To augment confidence in the internal validity of stimulus materials and to ensure
that the provocation scenario invoked a sufficient level of anger, a group of 20 midwestern high school students were presented with 4 different anger-inducing scenarios.
The exercise’s goal was to select the online aggression scenario that produced the most
anger in the target population. Participants were given screenshots of 4 messages posted
on a Facebook profile. They were then asked to imagine that after signing into their
Facebook account, they see that one of their classmates has posted a harassing comment
about them on their friend’s Facebook wall for everyone to see. A screen shot of their
friend’s Facebook wall with a harassing comment about them was affixed below the
instructions. After reading each message, participants rated the scenario based on how
the provoking situation would make them feel on a 5-point Likert-type scale from 1 (not
angry at all) to 5 (extremely angry). The scenario that produced the most anger in
participants (M = 3.93) was selected for the remaining study. Stimulus materials are
displayed in Appendices A, B, and C.
Selecting the behavioral stimulus symbols was the next pretest task. After rating
the anger-inducing scenarios, participants were subsequently presented with 11 symbols
and were asked to write a few sentences relating to the meaning of each symbol, the
emotions the symbol made them feel or brought to mind, and if those emotions were
strong or weak. The happy and angry facial expression symbols were determined to have

61

the highest level of face validity for the target population because participants agreed that
the meaning of the happy face symbol was a strong, happy emotion in all but two cases.
All except for one participant agreed that the angry face symbol represented an angry
emotion. In many cases, individuals suggested not only that the meanings of these
symbols were happy or angry, but that they also invoked the corresponding emotion
within themselves as they viewed the symbols. Responses provided for the remaining
symbols were much more convoluted and dissimilar and were therefore excluded from
the subsequent study.

Sample
Data were collected from a sample of 464 high school students from two
demographically diverse school districts in Colorado. High school students were deemed
the most relevant target population because studies presenting descriptive statistics on
online aggression suggest that individuals belonging to this age group engage in the
behavior more frequently than any other age group (Ybarra & Mitchell, 2004; Hinduja &
Patchin, 2008). The week of the study, all students received a parental consent form
explaining the nature of the study.

Additionally, an automated message by an

administrative staff member was sent to homes to inform parents that a study was being
conducted at their child’s school and that they could choose to opt out of participation.
The day of the study, students were provided with an informed assent form before
completing the survey, which explained the nature of the study and indicated that
participation was voluntary. After removing cases with incomplete information, data
from 434 of the 464 participants was useable.

62

The sample consisted of 50.69% males and 49.31% females, which is consistent
with the U.S. census bureau’s record of this age group of 51.2% males and 48.8%
females, respectively (U.S. Census Bureau, 2008). Of those who reported their racial
background, 51% identified themselves as exclusively White, 11.48% as Black, 15.80%
of Hispanic origin, .04% as Asian or Pacific Islander, and .01% as Native American.
About 13.16% identified themselves as belonging to two or more racial groups. The
remaining .03% of the sample indicated that they were from an unlisted race, but did not
specify which one.
Consistent with previous research which indicated that 40-60% of adolescents
have been harassed on the Internet (Beran & Li, 2005), 41.70 % of those in the present
sample indicated that they had been bullied through some form of electronic technology
such as a text message, an Internet chat-room or e-mail. A larger percentage, 68.7%,
reported that they knew someone who had been bullied through some form of
technology.

Likewise, the portion of those who admitted to using some form of

technology to bully others in this sample, 31.8%, closely corresponded to the portion of
those who claimed to have used technology to bully others (20-30%) in previous research
(Beran & Li, 2005). A large percentage of those who admitted to using technology to
bully others were females (61.4%) while males comprised of only 35.6% of past online
aggressors. Not surprisingly, only 5.3% admitted that they had bullied someone on
Facebook specifically. This small percentage is understandable when considering that
only 41.24% of the sample had Facebook accounts, well below the national average of
73% for teenagers (Pew, 2009). In the current study, 24% of females and 20.5% of males
reported aggressive online behavioral intentions.

63

The majority of participants maintained GPAs between 2.0 and 4.0. Specifically,
1.61% reported a GPA below 1.0, 8.29% reported a GPA between 1.0-1.99, 46.54%
reported a GPA between 2.0 and 2.99, and 42.62% reported a GPA between 3.0 and 4.0.

Data Collection
Using an online scenario-based survey, participants were systematically assigned
to one of three conditions; aggressive (Appendix A), anti-aggressive (Appendix B) or
control (Appendix C). A separate URL was assigned to each treatment condition. URLs
were individually displayed on a piece of paper and the pieces of paper were mixed in a
box to facilitate random assignment to treatment condition. Upon entering the classroom,
students were directed to grab a piece of paper from the box and to choose a computer to
work from. After all participants were seated in front of a computer, they were informed
that they were participants in a study about teenagers and the Internet, that the study was
voluntary, and that they did not have to answer questions that they did not want to
answer. They were subsequently instructed to go to the URL on their piece of paper, to
read the informed consent, and to follow the instructions on the website. Finally, they
were asked to raise their hands if they had any questions during the course of the study.
Before exposure to the scenario, participants responded to a range of survey-based
personality questions, including trait reactive aggression, trait self-control and attitude
towards retaliatory aggression.

Subsequently, the online application prompted

participants with the stimulus materials.

Participants were asked to imagine that after

signing into their Facebook account, they see that one of their classmates has posted a
harassing comment about them on their friend’s Facebook wall for everyone to see. A

64

screen shot of their friend’s Facebook wall with a harassing comment about them was
affixed below the instructions. A behavioral cue visual stimulus was embedded in the
Facebook page outside of the comment box for the aggressive (angry facial expression)
and anti-aggressive (happy facial expression) conditions. The control group was not
exposed to a symbol. Finally, participants completed the remaining survey items
pertaining to hostile thoughts, state hostility, state self-control and aggressive behavioral
intentions.
Analyses were performed using the Statistical Package for the Social Sciences
(SPSS) version 17.0 (SPSS, 2007).

Data were cleaned before testing hypotheses,

including the removal of scale items exhibiting basement or ceiling effects. Scale items
contributing to the most valid and reliable measures were used in subsequent analyses.
Moderation effects were tested through multiple regression analyses, while treatment
effects were analyzed through factorial univariate analysis of covariance procedures.

Operational Measures
A complete list of scale items used for measurement is indexed in Appendix D. State
hostility and State arousal were measured with subscales of the Profile of Mood of States
(POMS) short form (McNair, Lorr & Droppleman, 1992).

The POMS anger-hostility

subscale used to measure state hostility determined the individual’s level of momentary
anger through 10 items (e.g. I feel angry) on a scale ranging from 1 (strongly disagree) to
5 (strongly agree). The scale consistently demonstrates internal reliability, typically
yielding alphas of around .90 and it yielded a Cronbach’s alpha of .81. The POMS vigorarousal subscale measured state arousal through 3 items (e.g. I feel lively) on a scale

65

ranging from 1 (strongly disagree) to 5 (strongly agree). It generally yields alphas of
around .87 and it yielded a Cronbach’s alpha of .91 in this data set.
A 5-point Likert scale ranging from 1 (never) to 5 (often) measured trait reactive
aggression. The scale was developed by Raine, Dodge, Loeber, Gatzke-Kopp, Lynam et
al.(2006), and measures trait reactive aggression through 12 items (e.g. I react angrily
when provoked by others). It typically yields an alpha within the range of .81 to .86. The
present scale yielded an alpha of .87. The brief version of Tangney, et al.’s (2004) SelfControl Scale was used to measure trait self-control. Through 13 items (e.g. Sometimes I
can’t stop myself from doing something even if I know it’s wrong), the 5-point Likerttype scale ranging from 1 (strongly agree) to 5 (strongly disagree) measured trait selfcontrol.

It has demonstrated alphas of around .85 and the present scale yielded a

Cronbach’s alpha of .74. The state self-regulation scale was adapted from the trait
regulation scale. Previous research has demonstrated that creating a state measure by
slightly rewording a trait scale can produce an equally valid and reliable measure (Farrar
& Krcmar, 2006). Participants were asked to consider how they acted in the present
situation.

A confirmatory factor analysis (CFA) was conducted to assess internal

consistency and factor loadings, using the statistical software program AMOS. The CFA
determined that seven factors best represented state self-control. A Cronbach’s alpha of
.89 and substantial, significant factor loadings were obtained (p < .001), indicating
2

internal consistency. The model also obtained a good model of fit (X = 10.82, df = 6, p
= .09, AGFI = .971, CFI = .996, RMSEA = .04). Scale items are displayed in Table 1.

66

Table 1 State Self-Control Scale
β

Scale Item
Circle the number next to each item that best describes your thought
process when reacting to your classmate …
I could not resist the temptation to treat my classmate the same way I
was treated
I could not stop myself from reacting angrily towards my classmate
I had a hard time breaking the habit of reacting in a negative way
I wanted to say inappropriate things
I thought that I should humiliate my classmate even if it was bad for
me in the long run
I could not stop myself from treating my classmate the same way they
treated me even if I thought that it was wrong

M

SD

.77

3.50

1.32

.86 3.54
.75 3.42
.68 3.09
.70 3.61

1.27
1.23
1.32
1.23

.69

1.24

3.53

Perceived discomfort was measured with Anderson, Anderson et al.’s (1996) 5point Likert-type Perceived Comfort Scale (PCS) ranging from 1 (not at all) to 5
(extremely). Nine items (e.g. I feel comfortable) measured perceived discomfort. This
scale typically yields an alpha of .90 and it yielded an alpha of .89 in this case. Through
9 items (e.g. I want to get revenge), the revenge/retaliatory thinking subscale of the
Hostile and Hostile thoughts Scale (HAT) measured hostile thoughts on a 5-point Likerttype scale ranging from 1 (strongly disagree) to 5 (strongly agree). This subscale has
demonstrated internal reliability with alpha’s of .93 (Deffenbacher, Petrelli, Lynch,
Oetting & Swan, 2003) and yielded a Cronbach’s alpha of .95 in the present study.
Finally, aggressive behavioral intentions accounted for the primary dependent
1

variable of interest . Previous studies have measured aggression through scenario-based

1

The aggressive behavioral intention scale consisted of the likelihood that individuals
would engage in the following behaviors: 1. Send an angry private message back to the
classmate who wrote on my friend’s wall, 2. Post a humiliating comment about my
classmate on my friend’s Facebook wall, 3. Embarrass my classmate in front of other
people on Facebook, 4. Curse back at my classmate in person, 5. Ask my friend to
67

aggression scales (Leifer & Roberts, 1972; Yagil, 2001; O’Connor, Archer & Wu, 2001).
In the present case, participants were asked the likelihood that they would engage in
online and offline retaliatory aggressive behaviors (e.g. post a humiliating comment
about my classmate on my friend’s Facebook wall and Humiliate my classmate in front
of other people while at school) on a 5-point Likert-type scale ranging from 1 (very
unlikely) to 5 (very likely). Combined aggressive behavioral intentions, online and
offline aggressive behavioral intentions were measured and yielded Cronbach’s alphas of
.95, .94 and .89 respectively. Two open-ended questions (1. Is there anything else that
you would do to respond to your classmate and 2. Would you have reacted differently if
your classmate had harassed you in a similar manner in person) determined what other
types of behaviors participants would engage in and differences between online and
offline interactions. Based on a meta-analysis of short and long-term effects of violent
media on aggression in children and adults, this study expected to yield modest, but
significant effect sizes for exposure to stimulus on state hostility and hostile thoughts
(Bushman & Heusmann, 2006). Chapter 5 summarizes the test of each hypothesis and
research question.

remove the insulting comment from their wall, 6. Use Facebook to treat my classmate the
same way that I was treated, 7. Threaten my classmate on Facebook, 8. Humiliate my
classmate in front of other people while at school, 9. Insult my classmate to their face, 10.
Curse back at my classmate online, and 11. Humiliate my classmate in front of other
people while at school.
68

Chapter 5
Results
In consideration of traditional bullying studies that indicate that males are more
likely to bully than females (Wang et al., 2009), and conflicting previous research that
found that females are more likely to participate in cyberbullying than males (Smith et al,
2006; Kowalski, Limber & Agatston, 2008) or that males and females are equally likely
to engage in the aggressive behavior (Werner, Bumpus & Rock, 2009; Ybarra &
Mitchell, 2004b; Hinduja & Patchin, 2008), sex was included as a control variable in all
analyses.

Trait aggression was also included as a control variable throughout the

analyses because previous research has demonstrated the important association between
the construct and multiple aggression-related mechanisms (Kirsh, 2006; Bushman &
Geen, 1990; Lindsay & Anderson, 2000; Anderson, 1997).

VIF and tolerance

collinearity statistics were calculated for each regression equation to ensure that results
were not adversely affected by multicollinearity. Table 2 shows a matrix of Pearson
product-moment correlation coefficients, Cronbach alphas, means and standard
deviations.

Hypotheses 1a – 1e
The first set of hypotheses predicted a) a positive relationship between the present
internal state variables and a greater positive relationship between b) hostile thoughts, c)
state hostility, d) perceived discomfort and e) arousal and aggressive behavioral
intentions among those with lower trait self-control than among those with higher trait
self-control. Linear regression analyses controlling for sex and trait aggression were

69

conducted to determine the moderating effects of trait self-control. Interaction terms were
added to models as crossproducts of the independent predictors. The sex variable was
dummy coded as 1 (male) and 2 (female). Variables were centered before analysis to
reduce the potentially problematic occurrence of multicollinearity.

Table 2 Pearson Product-Moment Correlation Matrix of Independent and Dependent
Variables
Variable

1. Trait
Aggression
2. Attitude
3. Trait SelfControl
4. State SelfControl
5. State
Hostility
6. Hostile
Thoughts
7. Arousal
8. Discomfort
9. Behavioral
Intentions

2

3

8

.50
- -.31*
.44
- -.42* .30*
.51
.43 .24* -.30* -.53*
.53

.44*

-.32* -.70*

.71*

.16
.00
.45

.13
-.01
.45*

.06 -.09*
-.13* -.004
-.26* -.69*

-.07
.15*
.48*

.06
.06
.68*

.69*
.16

.07

.78

2.75
2.99

.86
.59

3.44

1.01

2.92

.98

.95

7

2.55

.91

6

SD

.89

5

M

.87
.74

4

α

.87

1

2.69

1.16

.80
.88
.95

3.12
2.89
2.33

1.09
.88
1.11

Trait Self-Control and Hostile Thoughts
A multiple linear regression was calculated to determine if trait self-control significantly
moderated the effects of hostile thoughts on aggressive behavior when controlling for sex
and trait aggression. The regression equation was significant (F(5,426) = 79.47, p < .001)
2

2

with an R of .69 and an R of .48. However, hostile thoughts (t = 14.79, p < .01, R =

70

2

.26) and trait aggression (t = 2.61, p < .01, R = .01) emerged as the only significant
predictors of aggressive behavioral intentions while trait self-control, sex, and the trait
self-control-by-hostile thoughts variable were not significant predictors. Table 3 displays
these relationships.

Table 3 Regression Statistics for the Effects of Trait Aggression and Hostile Thoughts on
Aggressive Behavioral Intentions
Variable
b
β
t
p
VIF
Aggressive Behavioral Intentions
Constant
Hostile Thoughts
Trait Self-Control
Trait Self-Control-by-Hostile Thoughts
Sex
Trait Aggression

-.01
.59
-.03
-.04
-.03
.16

.62
-.01
-.03
.01
.12

-.13
14.79
-.34
-.70
.36
2.61

.90
.001**
.73
.48
.72
.001**

1.04
1.32
1.27
1.00
1.01

Trait self-control did not moderate the relationship between hostile thoughts and behavior
as predicted; however, further analysis suggested that this relationship may be more
complex. In consideration of research that suggests that factors within the GAM’s
theoretical framework may disproportionately affect individuals based on their level of
trait aggression (Anderson, 1997), an additional multiple linear regression was calculated
after splitting the data set between individuals with low and high trait aggression. Trait
aggression was dichotomized by splitting the data set at the trait aggression scale’s
median (3), to derive a more accurate representation of high and low trait aggression than
would be derived from splitting the data set in half at the respondent’s median (2.46).
The trait self-control-by-hostile thoughts interaction variable nearly reached significance
for low trait aggression individuals when controlling for sex, trait self-control and hostile
71

thoughts (t = -1.79, p = .07), but fell short of arriving at the .05 significance level. The
regression equation was significant (F(4,317) = 79.273. p < .001) with an R of .71 and an
2

R of .49. Once again, hostile thoughts emerged as a significant predictor of aggressive
2

behavioral intentions (t = 17.24, p < .001, R = .46) while sex and trait self-control were
not significant predictors. In this case, low trait aggression individuals were almost
significantly less likely to admit to aggressive behavioral intentions if they maintained
highly hostile thoughts. Table 4 displays the regression equation and Figure 3 illustrates
the relationships. The interaction between trait self-control and hostile thoughts did not
approach significance (t = .140, p = .89) for high trait aggression individuals.

Table 4 Regression Statistics for the Moderating Effects of Trait Self-Control on Hostile
Thoughts and Aggressive Behavioral Intentions for Low Trait Aggression Individuals
Variable
b
β
t
p
VIF
Aggressive Behavioral Intention
Constant
Hostile Thoughts
Trait Self-Control
Trait Self-Control-by- Hostile Thoughts
Sex

-.05
.66
-.09
-.12
-.05

72

.70
-.06
-.08
.03

-1.23
17.24
-1.22
-1.79
.67

.22
.001**
.22
.07
.50

1.04
1.32
1.27
1.00

Figure 3 The Moderating Effects of Trait Self-Control on Hostile Thoughts and
Aggressive Behavioral Intentions for Low Trait Aggression Individuals.

Hypothesis 1c was partially supported. A multiple linear regression was calculated to
determine if trait self-control significantly moderated the effects of state hostility on
aggressive behavior when controlling for sex and trait aggression.

The regression

equation was significant (F(5,426) = 38.18, p < .001), with a multiple R of .56 and an R
2

2

2

of .31. Trait aggression (R = .05) and state hostility (R = .10) significantly predicted
aggressive behavior when controlling for sex, trait self-control, and the interaction
between trait self-control and state hostility. Sex, trait self-control and the interaction
73

between trait self-control and state hostility were not significant predictors when
controlling for trait aggression and state hostility. However, the data set was split once
again between low and high trait aggression individuals in consideration of research that
suggests that factors within the GAM’s theoretical framework may disproportionately
affect individuals based on their level of trait aggression (Anderson, 1997). When the
data were split between high and low trait aggression individuals, trait self-control
moderated the effects of state hostility on online aggressive behaviors when controlling
for sex, state hostility and trait self-control. A significant regression equation was found
2

(F(4,317) = 24.29, p < .001), with a multiple R of .484 and an R of .235. Trait self2

2

control (t = -2.31, p = .02, R = .013) and state hostility (t = 9.18, p < .001, R = .20)
were significant predictors; however, sex did not significantly predict aggressive
behavior.

The trait self-control-by-state hostility variable emerged as a significant
2

predictor (t = -2.39, p = .02, R = .02), indicating that trait self-control moderates the
effects of state hostility on online aggressive behaviors for low trait aggression
individuals when controlling for sex, trait self-control and state hostility. Low trait
aggression individuals who were high in state hostility were less likely to admit to
aggressive online behavioral intentions if they were higher in trait self-control compared
to high state hostility individuals who were lower in trait self-control. The regression
equation is displayed in Table 5 and Figure 4 illustrates these relationships.

74

Table 5 Regression Statistics for the Moderating Effects of Trait Self-Control on State
Hostility and Aggressive Behavioral Intentions for Low Trait Aggression Individuals
Variable
b
β
t
p
VIF
Aggressive Behavioral Intention
Constant
Sex
State Self-Control-by-State Hostility
State Hostility
Trait Self-Control

-.14
-.13
-.21
.49
-.22

-.07
-.13
.47
-.13

2.80
-1.33
-2.40
9.18
-2.31

.001*
.185
.02*
.001*
.02*

1.0
1.2
1.1
1.2

Figure 4 The Moderating Effects of Trait Self-Control on State Hostility and Aggressive
Behavior for Low Trait Aggression Individuals

75

Trait Self-Control and Perceived Discomfort
Hypothesis 1d was supported. A multiple regression equation was calculated to
determine if trait self-control moderates the effects of perceived discomfort on aggressive
behavior. A significant regression equation was found (F(5,426) = 26.68, p < .001) with
2

2

a multiple R of .47 and an R of .23. Trait aggression (t = 8.63, p < .001, R = .14), and
2

perceived discomfort (t = 2.28, p = .05, R = .01) were significant predictors of
aggressive behavioral intentions when controlling for sex, trait self-control, and the
interaction between trait self-control and perceived discomfort. Sex did not emerge as a
significant predictor. As hypothesized, the trait self-control-by-perceived discomfort
variable emerged as a significant predictor, indicating that trait self-control significantly
moderated the relationship between perceived discomfort and aggressive behavior (t = 2

2.28, p = .02, R = .02).

The lower individuals were in comfort, the more likely they

were to express aggressive behavioral intentions.

Furthermore, individuals who were

lower in comfort were significantly less likely to express aggressive behavioral intentions
if they were higher in trait self-control. Table 6 presents the significant regression
equation and Figure 5 illustrates the moderating influence of trait self-control on the
effects of perceived discomfort on aggressive behavioral intentions.

76

Table 6 Regression Statistics for the Moderating Effects of Self-Control on Perceived
Discomfort and Aggressive Behavioral Intentions.
Variable
b
β
t
p
VIF
Aggressive Behavioral Intention
Constant
Sex
Trait Control
Comfort
Trait Self-Control-by- Comfort
Trait Aggression

.008
.03
-.14
-.10
-.20
.58

-.02
-.07
-.08
-.09
.41

.319
.35
-1.476
-1.904
-2.28
8.63

.75
.73
.14
.05*
.02*
.001*

Figure 5 The Moderating Effects of Self-Control on Discomfort and
Aggressive Behavioral Intentions

77

1.0
1.0
1.0
1.0
1.0

Trait Self-Control and Arousal
Hypothesis 1e was supported. A multiple linear regression was calculated to
determine if trait self-control significantly moderated the effects of arousal on aggressive
behavioral intentions. A significant regression equation was found (F(5,426) = 25.22, p <
2

2

.001), with a multiple R of .48 and an R of .23. Arousal (t = 2.40, p = .02, R = .01)
2

and trait aggression (t = 7.84, p < .001, R = .11) were significant predictors of
aggressive behavioral intentions when controlling for sex, trait self-control, and the
interaction between trait self-control and arousal; however, sex and trait self-control were
not.

The trait self-control-by-arousal variable emerged as a significant predictor

suggesting that trait self-control significantly moderates the effects of arousal on
aggressive behavioral intentions. Individuals who were more aroused were more likely
to express aggressive behavioral intentions if they were lower in trait self-control (t = 2

2.43, p = .02, R = .01). Table 7 displays the regression equation and Figure 6 illustrates
the moderating influence of trait self control. There was a greater positive relationship
between arousal and aggressive behavioral intentions among those with low trait self
control than among those with high trait self-control. Individuals who were highly
aroused and who were lower in self-control were the most likely to express aggressive
behavioral intentions.

78

Table 7 Regression Statistics for The Moderating Effects of Trait Self-Control on
Arousal and Aggressive Behavioral Intentions.
Variable
b
β
t
p
VIF
Aggressive Behavioral Intention
Constant
Sex
Trait Self-Control
Arousal
Trait Self-Control-by-Arousal
Trait Aggression

.01
.02
-.16
.10
-.18
.54

-.01
-.08
.11
-.12
.38

.18
.17
-1.73
2.40
-2.43
7.84

.86
.87
.08
.02*
.02*
.001*

1.0
1.0
1.0
1.2
1.0

Hypotheses 2a-2c
The second set of hypotheses, which predicted that aggressive behavioral intentions are
negatively related to a) trait self-control, and to b) state self-control and that c) trait and
state self-control are positively related, was supported. A Pearson correlation coefficient
was calculated to test the relationship between trait self-control and aggressive behavioral
intentions. A weak but significant negative correlation was found (r(432) = -.26, p <
.001). A Pearson correlation coefficient was calculated to test the relationship between
state self-control and aggressive behavioral intentions. A strong negative correlation was
found (r(432) = -.69, p < .001), indicating a significant negative linear relationship
between the two variables. Individuals who reported low levels of state self-control
tended to express more aggressive behavioral intentions.

A Pearson correlation

coefficient was also calculated to test the relationship between trait self-control and state
self-control. The analysis revealed a weak correlation (r(432) = .30, p < .001), indicating
a significant positive linear relationship between trait self-control and state self-control.

79

Figure 6 The Moderating Effects of Trait Self-Control on Arousal and Aggressive
Behavioral Intentions.

Hypotheses 3a-3d
The third set of hypotheses proposed that there should be a greater positive linear
relationship between a) hostile thoughts, b) state hostility, c) perceived discomfort and d)
arousal and aggressive behavioral intentions among those with lower state self-control,
than among those with higher state self-control. A series of multiple linear regressions
were calculated to determine the influence of state self-control.

80

Hostile Thoughts
A multiple linear regression was calculated to determine if state self-control significantly
moderated the effects of hostile thoughts on aggressive behavioral intentions.

A

significant regression equation was found (F(5,426) = 109.05, p < .001), with a multiple
2

R of .75 and an R of .56. However, the state self-control-by-hostile thoughts interaction
variable did not emerge as a significant predictor of aggressive behavioral intentions
when controlling for sex, state self-control and hostile thoughts, indicating that state selfcontrol did not significantly moderate the effects of hostile thoughts on aggressive
behavioral intentions (Table 8). However, a multiple regression calculated after splitting
the data set between low and high trait individuals, suggested that state-self control was a
significant moderator for certain people. A significant regression equation was found
2

(F(4,317) = 118.73, p < .001), with a multiple R of .77 and an R of .60. The state selfcontrol-by-hostile thoughts interaction variable emerged as a significant predictor of
aggressive behavioral intentions when controlling for sex, state self-control and hostile
2

thoughts (p = .005, R = .01), indicating that state-self control significantly moderated the
effects of hostile thoughts on aggressive behavioral intentions for low trait aggression
individuals. State self-control and hostile thoughts significantly predicted aggressive
behavioral intentions while sex did not. Table 9 displays the regression equation and
Figure 7 displays the moderating relationship between trait self-control, hostile thoughts
and aggressive behavioral intentions.
The figure demonstrates that low hostile thought individuals were the least likely
to express aggressive behavioral intentions, but that high hostile thought individuals who

81

were lower in state self-control were the most likely to express aggressive behavioral
intentions.

Table 8 Regression Statistics for the Effects of State Self-Control and Hostile Thoughts
on Aggressive Behavioral Intentions
Variable
b
β
t
p
VIF
Aggressive Behavioral Intention
Constant
Sex
Trait Aggression
State Self-Control-by-Hostile Thoughts
State Self-Control
Hostile Thoughts

-.132
-.01
.06
-.04
-.60
.35

-.01
.04
-.04
-.41
.37

-2.96
-.16
-1.13
-1.02
-8.73
7.43

.003
.88
.26
.31
.001*
.001*

1.0
1.47
1.18
2.09
2.35

Table 9 Regression Statistics for the Moderating Effects of State Self-Control on Hostile
Thoughts and Aggressive Behavioral Intentions for Low Trait Aggression Individuals
Variable
b
β
t
p
VIF
Aggressive Behavioral Intention
Constant
Sex
State Self-Control
State Self-Control-by-Hostile Thoughts
Hostile Thoughts

-.07
-.02
-.50
-.09
.38

82

-.01
-.45
-.10
.40

-1.67
-.28
-9.01
-2.81
8.11

.096
.78
.001*
.005*
.001*

1.02
1.99
1.04
1.94

Figure 7 The Moderating Effects of State Self-Control on Hostile Thoughts
and Aggressive Behavioral Intentions.

State Hostility
A multiple linear regression was calculated to determine if state self-control significantly
moderates the effects of state hostility on aggressive behavioral intentions. A significant
regression equation was found (F(5,426) = 89.21, p < .001), with a multiple R of .72 and
2

2

2

an R of .51. State hostility (R = .01), state self-control (R = .20), and trait aggression
2

(R = .01) significantly predicted aggressive behavioral intentions, when controlling for
sex and the state self-control-by-state hostility variable. However, sex was not a
83

significant predictor. State self-control emerged as a significant moderator of the
relationship between state hostility and aggressive behavioral intentions when controlling
2

for sex, trait aggression, state self-control and state hostility (R = .01). Individuals with
higher levels of state hostility were significantly less likely to admit to aggressive
behavioral intentions if they maintained higher levels of state self-control. Table 10
depicts the significant regression equation and Figure 8 illustrates the relationships.

Table 10 Regression Statistics for the Moderating Effects of State Self-Control on State
Hostility and Aggressive Behavioral Intentions
Variable
b
β
t
p
VIF
Aggressive Behavioral Intention
Constant
Sex
State Self-Control
State Self-Control-by-State Hostility
State Hostility
Trait Aggression

-.189
-.06
-.84
-.09
.14
.14

84

-.03
-.56
-.07
.12
.10

-4.34
-.83
-13.19
-1.92
2.78
2.45

.001
.41
.001*
.05*
.006*
.02

1.03
1.39
1.02
1.43
1.04

Figure 8 The Moderating Effects of State Self-Control on State Hostility and Aggressive
Behavioral Intentions

Comfort
A multiple linear regression was calculated to determine if state self-control significantly
moderated the effects of perceived discomfort on aggressive behavioral intentions. A
significant regression equation was found (F(5,426) = 88.22, p < .001), with a multiple R
2

2

of .71 and an R of .51. State self-control (t = -15.91, p < .001, R = .30), perceived
2

2

discomfort (t = 2.93, p = .004, R = .01), and trait aggression (t = 3.23, p = .001, R =

85

.01) significantly predicted aggressive behavioral intentions when controlling for sex and
the state self-control-by-perceived discomfort variable. While the state self-control-byperceived discomfort variable significantly moderated the effects of perceived discomfort
on aggressive behavior when trait aggression was not part of the equation, the state selfcontrol-by-perceived discomfort variable and sex were not significant predictors when
controlling for trait aggression. Therefore, although the state self control-by-discomfort
variable approached significance (p = .09), state-self control did not significantly
moderate the effects of perceived discomfort on aggressive behavioral intentions as
hypothesized. State self-control did not significantly moderate the effects of perceived
discomfort on aggressive behavioral intentions for low trait aggression individuals (p >
.05).

Arousal
A multiple linear regression was calculated to determine if state self-control significantly
moderated the effects of arousal on aggressive behavioral intentions.

A significant

regression equation was found (F(5,426) = 86.46, p < .001), with a multiple R of .71 and
2

an R of .50. The state self-control-by-arousal variable did not emerge as a significant
predictor of aggressive behavioral intentions after controlling for sex, trait aggression,
arousal and state self-control indicating that state self-control did not significantly
moderate the effects of arousal on aggressive behavioral intentions. State self-control
2

2

2

(R = .30), trait aggression (R = .01), and arousal (R = .01) significantly predicted
aggressive behavioral intentions when controlling for the interaction between state selfcontrol and arousal and sex, while sex did not. State self-control did not significantly

86

moderate the effects of arousal on aggressive behavioral intentions for low trait
aggression individuals (t = -.43, p = .67).

2

Hypotheses 4a - 4c
A factorial analysis of variance was performed to test the final set of hypotheses
which proposed that those exposed to aggressive behavioral cues should have a) the most
hostile thoughts and b) the most state hostility, followed by those who were not exposed
to a behavioral stimulus, and finally by those who were exposed to the anti-aggressive
stimulus, and that trait self-control should moderate the effects of the behavioral stimulus
on hostile thoughts and state hostility, such that there should be a positive relationship
between aggressive stimulus exposure and c) hostile thoughts and d) state hostility among
those with low trait self-control, but no relationship among those with high trait-selfcontrol.
To assess treatment effects, a 3 (treatment condition) x 2 (trait self-control)
between subjects factorial analysis of covariance was calculated comparing the hostile
thoughts of participants who were exposed to the aggressive behavioral stimulus, to the
anti-aggressive stimulus or to no stimulus and who were classified as maintaining high or
low trait self-control. A dichotomous self-control variable was created by splitting the
construct at its median (2.92), and sex and trait aggression were included as a common
covariates. The initial analysis did not support the hypothesis. The main effect for
treatment group was not significant (F(2, 424) = 4.41, p = .26), nor was the main effect
2

Post hoc analyses controlling for Facebook account ownership were performed on all
regression equations to determine whether experience with Facebook impacted the
reported results. Having a Facebook account was not a significant predictor in any of the
equations, nor did it modify any of the reported results.
87

for sex (F(1,424) = .01, p = .91). However, significant main effects for trait self-control
(F(1,424) = 4.41, p = .04) and trait aggression (F(1,424) = 132.05, p < .001) were found.
Finally, the interaction between trait self-control and treatment group was not significant
(F(2,424) = .57, p = .57). Thus, it appeared that neither treatment group nor sex had any
significant impact on hostile thoughts, while trait self-control did.
A 3 (treatment condition) x 2 (trait self-control) between subjects factorial
analysis of covariance was calculated comparing the state hostility of participants who
were exposed to the aggressive behavioral stimulus, to the anti-aggressive stimulus or to
no stimulus, and who were classified as maintaining high or low trait self-control. Trait
aggression and sex were included as covariates. The main effect for treatment group was
significant (F(2,424) = 3.05, p = .04), as were the main effects for trait self-control
(F(1,424) = 9.40, p = .002), trait aggression (F(1,424) = 43.64, p < .001) and sex
(F(1,424) = 12.73, p < .001). Low trait self-control individuals were higher in state
hostility (M = 3.14, SD = 1.0) than high trait self-control individuals (M = 2.73, SD =
.93, Cohen’s d = .42) and females were higher in state hostility (M = 3.06, SD = .98) than
males (M = 2.77, SD = .96, Cohen’s d = .30). Individuals exposed to the aggressive
treatment condition were highest in state hostility (M = 3.09, SD = .96), followed by
those exposed to the anti-aggressive treatment condition (M = 2.92, SD = .97), and
finally by those not exposed to an aggressive stimuli (M = 2.72, SD = .99). Contrary to
the hypothesis, individuals in the control condition who were not exposed to a behavioral
stimulus exhibited the lowest amount of state hostility. However, as predicted, those who
were exposed to the anti-aggressive treatment condition expressed less state hostility than
those exposed to the aggressive treatment condition. An a priori contrast test revealed a

88

significant difference between the aggressive and anti-aggressive treatment groups
(F(2,424) = 3.05, p = .04), where individuals exposed to the anti-aggressive treatment
condition exhibited significantly lower levels of state hostility than those exposed to the
aggressive treatment condition (p = .03, Cohen’s d = .18). In other words, the antiaggressive treatment condition suppressed state hostility more than the angry treatment
condition.
Finally, the interaction between trait self-control and treatment group was not
significant (F(2,424) = .164), p > .05). When the data set was split between low and high
reactive trait aggression individuals, the moderating influence of trait self-control on the
effects of behavioral stimuli exposure on state hostility approached significance for low
trait aggression females (F(2,310) = 2.58, p = .07); however, it fell short of reaching the
.05 significance level. Figure 7 displays the significant relationships revealed by the
present data analysis.

89

Figure 9 Significant Tested Relationships

Trait Self-Control

Behavioral Stimuli Exposure

State Hostility
Hostile Thoughts
Comfort
Arousal
State SelfControl
Behavioral
Intention
All Individuals,
Low Trait Aggression Individuals

Online vs. Offline Aggression

Research Question 1
A constant comparative preliminary analysis was conducted to identify commonalities
between respondents. The first research question asked if participants would do anything
beyond the replies that they had already given to respond to their aggressor. Of the 53%
of participants who provided answers, 56.5% said ‘yes’ and 43.5% said ‘no’. Of those
who said ‘yes’, many respondents suggested that they would engage in a non-aggressive

90

confrontation (40.5%); however, others stated that they would respond aggressively
online (20.61%) or in person (27.48%). Some of the participants who suggested that they
would behave aggressively in person further admitted that they would confront their
perpetrator verbally (15.27%) or physically (12.21%) in the offline world.

Other

responses included intentions to tell an authority figure such as a parent about the
incident (3.1%), to remove the perpetrator from their list of friends (.04%), to take the
post down (.01%) and to ignore the perpetrator (17.67%).
Multiple cases demonstrated that when online aggression does not occur
anonymously, that its repercussions can affect and initiate offline aggressive responses. A
few suggested that online attacks are shallow and unintimidating. These respondents
expressed pity for aggressors who felt the need to instigate others through an online
platform and suggested that aggressors who act face-to-face have courage if nothing else.
However, several participants expressed desires to find their aggressor offline and to
engage in forms of verbal aggression, such as yelling, or in forms of physical aggression,
such as fighting, kicking, hitting or beating (12.21%).

Research Question 2
A constant comparative preliminary analysis was also performed to evaluate the
second open-ended research question, which asked participants if they would have
reacted differently if their classmate had harassed them in a similar manner in person.
Out of all of the participants, 59% of the sample responded to the second research
question. Of those respondents, 65% said ‘yes’ and 34% said ‘no’. Of those who said
‘yes’, some respondents expressed the desire to engage in a non-aggressive confrontation

91

(7.2%). Although many participants expressed the desire to react aggressively when
harassed through an online setting, this desire seemed to escalate greatly when presented
with the idea of an offline aggressor. Nearly 22% of those who responded suggested that
they would react in a verbally aggressive manner by insulting, verbally harassing,
screaming or yelling at their aggressor, or in a physically aggressive manner (22.75%) by
punching, hitting, fighting, slapping or physically attacking their aggressor.

They

excused this behavior through multiple avenues. Nearly 12% of participants suggested
that the nature of public aggression is much more intimidating in offline encounters
because the presence of an active audience demands one’s response. Respondents also
suggested that public offline provocation produces more anger, frustration and/or
embarrassment (11%). Other responses included intentions to talk to an authority figure
such as a parent about the incident (1.8%), or to ignore the perpetrator (10.55%).
When probed about their online aggressor, many individuals (25.19%)
expressed a desire to know why their classmate harassed them or to understand the source
of their classmate’s anger before reacting. This delayed response was barely evident in
responses to the second open-ended research question (.03%), which asked participants if
they would have reacted differently had their classmate harassed them in a similar
manner in person. The final chapter discusses the implication of this study’s results. It
also addresses limitations of this research and outlines potential avenues of future
research related to the present findings.

92

Chapter 6
Discussion
This study sought to understand and inhibit online aggression through the General
Aggression Model’s theoretical framework. The study’s main purpose was to examine
the interaction between aggressive and anti-aggressive behavioral stimuli and self-control
on cognition, affect and aggressive behavior. Its secondary purposes were to consider
trait and state self-control as additions to the GAM, and to explore the differences
between off- and online aggression. Many studies have focused on the former; few have
empirically investigated the latter. The original contribution of the present study was to
examine online aggression through a theoretical framework, to present trait and state selfcontrol as valuable additions to the GAM, and to ascertain the value of exposure to antiaggressive behavioral stimuli in restraining online aggression. Trait and state self-control
emerged as significant moderating variables, behavioral stimuli exposure influenced
affect, but not cognition, and thought-provoking connections between off- and online
aggression surfaced.

Trait and State Self-Control as Moderating Variables
This study’s major contribution to online aggression and to the universe of
aggression research in general is that self-control, whether as a trait, or as a function of
state, is an important antecedent to aggressive behaviors.

Consistent with the

correlational relationships proposed by the GAM and previous research (Bushman &
Geen, 1990; Lindsay & Anderson, 2000; Anderson 1999), hostile thoughts led to
aggressive behavioral intentions.

As expected, trait aggression was consistently a

93

significant predictor of aggressive behavioral intentions. Both trait self-control and state
self-control negatively related to aggressive behavioral intentions, such that higher levels
of self-control were associated with fewer aggressive behavioral intentions.
However, the data presented only a weak significant correlation between trait selfcontrol and aggressive behavioral intentions, which likely accounted for its lack of
predictive utility when controlling for other variables throughout many of the regression
equations. Some adolescents may maintain inflated perceptions of their self-regulatory
capacities. The gap between an individual’s perceived self-regulatory capacity and their
actual self-regulatory capacity could partially explain the weak effect for trait self-control
on behavioral intentions. On the other hand, self-regulatory domain specificity may also
partially explain the gap. Self-control is said to be domain specific, such that one’s
ability to self-regulate one behavior is not necessarily indicative of one’s ability to selfregulate another behavior. Some participants who maintain generally high levels of trait
self-control in most areas, and who thus reported high levels of trait self-control, may
experience self-regulatory failure in the face of retaliatory online aggression.
Alternatively, self-control may not be fully developed even among high school-aged
adolescents, and personality traits are not always good behavioral predictors (Mischel,
1968).
Contrary to expectations, trait self-control was not found to moderate the effects
of hostile thoughts on aggressive behavior. However, trait self-control closely approached
significance in moderating the effects of hostile thoughts on aggressive behavior for low
trait aggression individuals, and state self-control did significantly moderate the effects of
hostile thoughts on aggressive behavior for low trait aggression individuals.

94

Logically, individuals who think fewer hostile thoughts (e.g. I want to get this
person back and I want to get revenge) are less likely to engage in acts of retaliatory
aggression. Hostile thoughts express a cognitive desire to behave aggressively. In their
absence, an individual may feel anger, but not enough anger to actually aggressively
retaliate. For those who develop highly hostile thoughts, what can be done to inhibit
incidents of aggression in instances when aggressive retaliatory cognition clouds
judgment?
According to the results of this study, self-control can significantly inhibit those
thoughts for some individuals.

By applying the presumed utility of self-control to

relationships that already existed within the GAM, this study revealed that self-control
can moderate the relationship between hostile thoughts and aggressive behavioral
intentions for low trait aggression individuals. Those with highly hostile thoughts were
more likely to refrain from aggressive behavioral intentions, despite exhibiting thoughts
of revenge, if they had high levels of self-control. Self regulatory literature suggests that
thought control is one of the primary dimensions of self-control (Baumeister et al., 2004;
Finkenauer et al., 2006). These results empirically bolster the notion that self-control can
play a significant role in cognition.
However, this effect was not evident for high trait aggression individuals.
Descriptive statistics ascertained after splitting the data set between low and high trait
aggression individuals revealed that in general, more low trait aggression individuals
were high in trait self-control (64.3%) than high trait aggression individuals (35.8%).
Previous studies have found that high trait aggression may derail influential effects of
other variables, thus the lack of effect for high trait aggression individuals is clearly

95

conceivable. For example, previous studies have found that exposure to aggressive
stimuli may not affect high trait aggression individuals (Anderson, 1997). In this case,
neither trait nor state self-control were significant moderators for high trait aggression
individuals, suggesting that inhibitory factors may also experience greater difficulty in
impeding aggressive behavior for those with highly hostile personalities.
General aggressiveness is also related to an attribution bias in people (Lindsay &
Anderson, 2000). The appraisal processes of individuals who maintain highly hostile
personalities tend to be somewhat skewed, such that these individuals often justify
aggressive actions by attributing blame to the other person. In this case, high trait
aggression may have reduced the mediating effects of the appraisal process, thereby
eliminating the need or desire to engage in self-control. During the appraisal stage, they
could have justified their retaliatory aggressive behavior by attributing blame to the
provoker. They may not have perceived the act of retaliatory aggression as a negative or
socially undesirable action, but as a necessary, warranted and justifiable or uncontrollable
reaction to which they were entitled.
A similar pattern between state hostility and aggressive behavior emerged.
Consistent with the correlational relationships proposed by the GAM and previous
research (Bushman & Geen, 1990; Lindsay & Anderson, 2000; Anderson 1999), a
positive relationship between state hostility and aggressive behavioral intentions
suggested that individuals experiencing more anger are more likely to form aggressive
behavioral intentions. Individuals experiencing highly hostile states were less likely to
admit to aggressive behavioral intentions if they were higher in state self-control. Low
trait aggression individuals experiencing highly hostile states were also less likely to

96

admit to aggressive behavioral intentions if they were higher in trait self-control. Once
again, the data presented a clear division between low and high trait aggression
individuals for trait self-control. However, state self-control emerged as a significant
moderator of the effects of state hostility on aggressive behavioral intentions for both low
and high trait aggression adolescents.
This study’s results are consistent with previous studies which have found that
factors such as pain and uncomfortable temperatures result in increased aggression
(Anderson & Bushman, 2002). Aggression research manipulating temperature effects
often include perceived comfort as a manipulation check (Anderson et al., 1996) to
determine whether inductions produced adequate degrees of discomfort. These studies
have manipulated comfort levels by immersing an individual’s hand in ice cold water or
modifying room temperatures to make individuals uncomfortable. This study did not
involve a comfort manipulation; however, perceived discomfort was similarly associated
with aggressive behavioral intentions.

A positive relationship between perceived

discomfort and aggressive behavioral intentions emerged, indicating that those who felt
comfortable, rested and soothed, were less likely to exhibit aggressive behavioral
intentions.

Participants who felt uncomfortable in the current study did not feel

uncomfortable because they were subjected to painful stimuli, but level of comfort still
functioned as expected.
Once again, trait self-control emerged as a significant inhibitory variable,
demonstrating that its importance extends beyond cognition and hostile affect. Those
who were more likely to express aggressive behavioral intentions because of low comfort
levels were significantly less likely to express aggressive behavioral intentions if they

97

maintained higher trait self-control. Individuals are more receptive to logical and rational
thought processes when they feel rested and comfortable (two items used to measure
perceived discomfort) than when they are grappling with uncomfortable and unpleasant
feelings. Individuals with rested mindsets through which they can rationally consider
potential consequences likely have more control over their self-regulatory capabilities.
Self-monitoring, one of the necessary components of self-control (Baumeister, 2002),
operates more efficiently within comfortable internal states.

Contrary to these

expectations, however, state self-control was not a significant moderator of the effects of
perceived discomfort on aggressive behavioral intentions. Interestingly, the confirmatory
factor analysis suggested that certain items (e.g. I thought through all of the alternatives
before I reacted and I used a lot of self discipline in this situation) be removed from the
state self-control scale, although their equivalents were represented in the trait selfcontrol scale. The strongest indicators of the self-monitoring component were therefore
absent from the measure of state self-control which could partially explain why trait selfcontrol emerged as a significant moderator but why state self-control did not.
Consistent with the GAM propositions, arousal was positively correlated with
aggressive behavioral intentions, suggesting that increased arousal leads to increased
aggressive behavioral intentions. Trait self-control, but not state self-control, moderated
the relationship between arousal and aggressive behavioral intentions.

This study

measured arousal through self-report items (e.g. I feel active and I feel energetic).
Anderson and colleagues (1995) suggest that arousal can be measured through either
physiological indicators such as blood pressure or through self-report items, yet most
previous studies that have connected arousal to aggressive behaviors relied on the former.

98

Previous GAM studies have measured arousal through self-report (Chory-Asaad, 2004),
however, physiological indicators may more accurately measure the effects of
provocation or other stimuli on the individual, while the self-report measures used in this
study seem somewhat disconnected from the bodily affects of arousal that one would
expect from aggressive provocation. Perhaps the moderating influence of the effects of
arousal on aggressive behavioral intentions would have been more evident if arousal was
measured through physiological indicators. Another possibility is that as an automatic
process, arousal is not as subject to state-level control, which requires thoughtful
consideration. Automatic processes such as arousal present a potential challenge to the
utility of self-control because they occur outside of awareness in a largely involuntary
manner.
In general, self-control’s influence on the present internal state factors can
potentially be significant. Interactions between self-control and aggressive thoughts,
state hostility, comfort and arousal are particularly important in light of recent selfregulatory research which suggests that an individual’s self-regulatory capacity closely
relates to a muscle that can be trained and expanded (Baumeister, 2002). Individuals who
have participated in self-regulatory regimens have successfully increased their selfregulatory capabilities by exerting control over mundane, but habitually routine activities.
Activities included in previous self-regulatory regimens include brushing one’s teeth with
the non-dominant hand and refraining from using sentences that begin with the word “I”
(Finkel et al., 2009). In a similar fashion, adolescents who consistently exert conscious
control over activities that require active self-monitoring of even non-related behaviors
can increase their capacity to do so in future situations.

99

Similarly, self-regulatory

exercises should allow adolescents who were previously unable to overcome hostile
cognition and affect to exert greater levels of control over their future behaviors.

The Relationship between Trait and State Self-Control
Trait self-control was an important predictor of state self-control; however, the
relationship was weak. The data presented only a weak relationship between the two
constructs. Adolescents who reported possessing a capacity for resisting temptation and
the ability to avoid engaging in behaviors that they knew were wrong at the beginning of
the study, did not necessarily resist temptation or avoid engaging in risky behavior after
provocation at the end of the study.
Motivation likely mediates the relationship between trait self-control and state
self-control. This research demonstrated that an individual may actually elect or not elect
to engage in self-control regardless of their ability to control their impulsive inclinations.
Perhaps some high trait self-control individuals were not motivated to inhibit their desire
to behave aggressively because they believed that responding in a similar manner as their
provoker would be good for them rather than bad for them. Aggressive behavior is
generally considered socially undesirable, but the motivation to pursue more personally
important goals may override consideration of unattractive behavior. To illustrate, the
individual may believe that he must behave aggressively to prevent further acts of
aggression by the provoker. Thus, he would be motivated to behave aggressively in order
to create good future results. Another possibility is that some of these individuals had
depleted self-regulatory resources and were not supplied with significant enough
motivation to restrain their behavior. This alternative is consistent with previous research

100

(Muraven & Slessareva, 2003) which demonstrated that individuals can overcome
depleted self-regulatory states when supplied with a meaningful motivation. Finally, the
weak relationship between the trait and state measures may have been because trait selfcontrol measured a more general form of self-control while state self-control was a
domain specific indicator. It also provides support for the argument that personality traits
are not always adequate behavioral indicators because people sometimes behave quite
differently depending on the situation (Mischel, 1968; Fleeson, 2007). In any case, the
significant role of trait self-control is intriguing through its influence on state self-control.
The relationship reveals a somewhat predictable interaction; however, it presents another
avenue through which one’s trait self-control can positively counteract aggressive
behaviors.

Behavioral Stimuli
Exposure to the aggressive behavioral cue functioned partially as hypothesized,
such that the anti-aggressive behavioral stimulus inhibited state hostility significantly
more than the aggressive behavioral stimulus, but the treatment conditions did not
significantly impact hostile thoughts. Previous GAM studies have found that aggressive
stimuli act primarily through the cognitive channel to affect aggressive behavior. The
GAM proposes that aggressive stimuli can impact aggressive behaviors through both
cognition and affect, yet multiple studies have discovered only partial support for this
proposition, finding that aggressive cues influence aggressive behavior primarily through
cognition and not through affect.

The mere presence of guns increases aggressive

behavior of participants (Berkowitz & LePage, 1967) and the presence of guns or other

101

aggressive cues increases the accessibility of hostile thoughts (Anderson, Benjamin &
Bartholow, 1998; Bartholow, Anderson, Carnagey & Benjamin, 2005; Meier, Robinson
& Wilkowski, 2007).
In line with previous studies that have demonstrated the utility of pro-social
behavioral stimuli to pro-social behavior (Huston-Stein and Friedrich, 1973;
Greitemeyer, 2009; Greitemeyer & Osswald, 2010), the current results indicate that
behavioral stimuli can impact aggressive behavioral intentions through affect. Exposure
to the anti-aggressive behavioral stimulus decreased state hostility such that individuals
who were exposed to the happy face reported feeling less angry, less bitter, less
aggravated and less resentful than those who were exposed to the angry facial expression.
Contrary to expectations, however, participants in the anti-aggressive treatment
condition actually appeared angrier than the control condition participants who did not
view a behavioral stimulus, although the two groups were not statistically different from
each other in state hostility. Despite the pretest consensus on the behavioral stimulus’s
meaning, when paired with the anger-inducing scenario, the happy facial expression may
have acquired a mocking connotation. For instance, the perception that not only is this
individual harassing you, but s/he thinks that it is funny may have been interpreted.

The

facial expression stimulus may have been perceived as an additional message from the
online provoker rather than as a non-affiliated visual cue. The stimulus facial expression
was displayed on the right side of the comment box, away from the words; however, the
absence of other visual cues from the website may have made the image appear as if it
came from the provoker. Previous computer mediated communication research suggests
that emoticons are used to convey emotions that are otherwise not strongly conveyed in a

102

medium that lacks non-verbal cues, and that an emoticon can impact a message’s
interpretation when it is perceived as belonging to the message. If this was the case in the
present study, the results are consistent with previous research (Thompson & Foulger,
1996) in that positive facial expressions can increase, rather than decrease anger when the
message is perceived as extremely negative. The emoticons were not images of actual
people; however they potentially personalized the attack by providing a face for the
provoker. In any case, the results suggested that mere exposure to certain behavioral
stimuli can affect aggressive behavior through an individual’s internal state. With further
research, online social networks or other online websites can potentially curtail acts of
aggression on their sites by displaying, or not displaying, the cues that affect cognition
and affect.
Consistent with previous research (Berkowitz & LePage, 1967; Anderson,
Benjamin & Bartholow, 1998; Bartholow, Anderson, Carnagey & Benjamin, 2005;
Meier, Robinson & Wilkowski, 2007) hostile thoughts influenced aggressive behavior.
However, the aggressive stimulus’ impact on aggressive behavior through hostile
thoughts was curiously absent. Perhaps the provocation of participants in the present
study can account for some of the discrepancies between current and prior findings.
Many previous studies have attempted to invoke hostile thoughts, state hostility, and
aggressive behavior solely through aggressive cues (Berkowitz & Lepage, 1967;
Anderson, Benjamin, & Bartholow, 1998) or provocation, while the present study
combined the two stimuli.

Exposure to the anger-inducing scenario undoubtedly

influenced these variables. The scenario was designed to impact cognition by increasing
hostile thoughts and affect through escalating state hostility. Its direct impact could not

103

be assessed because it was not experimentally manipulated across participants; however,
it was designed to engage the cognition by igniting vengeful thoughts against the
attacker. Results suggested that the anger-inducing scenario sufficiently invoked hostile
thoughts, without the aggressive cue, for those who subsequently intended to engage in
aggressive behavior.
The divergence between the channel affected in previous studies and the current
one may also be explained through the nature of the behavioral stimulus. Most prior
studies have used a gun or another such weapon as an aggressive cue to impact the
present internal state (Berkowitz & Lepage, 1967). For multiple reasons, the present
study cued behavior through emoticons relating to anger or happiness rather than through
more commonly used weapons. Weapons may more readily impact cognition than state
hostility because they provide a tool through which one can enact revenge. Hostile
thoughts are measured by sentiments such as ‘I want to get revenge’ and ‘I’ll show this
person.’ Exposure to a weapon likely activates an automatic and vengeful mental script.
Angry facial expressions, on the other hand, do not provide a tool through which
someone can engage in retaliatory aggression and they likely activate a different type of
mental script by increasing an already present angry or bitter emotion. Trait self-control
did not moderate the relationship between behavioral stimuli exposure on cognition or
affect. However, low self-control led to high state hostility and highly hostile thoughts.

Online Aggression and Gender
Aggression is generally considered a male dominant trait and most aggression
research supports the notion that males tend to exhibit more aggressive behavior than

104

females (Bandura et al., 1961; Kowalski & Limber, 2007). Preliminary cyberbullying
studies suggest that aggressive tendencies may function differently online. Results from
this study’s exploration on gender roles in online aggression are consistent both with
previous research that suggests that males and females are equally likely to engage in
online aggression (Werner, Bumpus & Rock, 2009; Ybarra & Mitchell, 2004b) and with
those that contend that females are more likely to engage in online aggression than males
(Smith, Madavi, Carvalho, Tippet, 2006; Kowalksi, Limber & Agatston, 2008). While
61.4% of females admitted to having used a technology such as text messages, chat room
or email to aggress, only 35.6% of males made the same confession.
Further information is needed to identify the specific technologies that females
used to aggress in the past.

Text messages and email generally involve private

interactions, while communication in chat rooms frequently occurs anonymously between
individuals who have not met each other in the offline world. These descriptive statistics
support the notion that females are more likely to behave aggressively in private or
anonymous settings. Consistent with gender role literature contending that males are
socialized to openly exhibit more aggression than females (Bandura et al., 1961;
Kowalski & Limber, 2007) this research lends credence to the proposition that females
may feel more comfortable with aggression when it is performed through a private or
anonymous electronic barrier.

The significantly smaller percentage of males who

reported prior cyberbullying behavior suggested that when the desire to behave
aggressively arises, males may prefer to express their aggression through more
conventional channels.
Results indicated that females were significantly higher in state hostility than

105

males, however, sex consistently proved to be an insignificant predictor of aggressive
behavioral intentions. A chi square test of independence was calculated comparing the
frequency of online aggressive behavioral intentions for males and females. Only 24% of
females indicated that they were at least somewhat likely to use the Internet to
aggressively respond to their provoker within the context of this study.

A slightly

2

smaller, but not significantly different (X (1) = .720 p > .05) percentage of males
(20.5%) also intended to behave aggressively online.

Online aggression within the

current context was neither anonymous nor private. If anonymity and privacy promote
electronic aggression for females as descriptive statistics suggested, the public and
identifiable nature of this study’s behavior of interest should restrain aggressive online
conduct. When stripped of anonymity and dimensions of privacy, females may view
online aggression as increasingly similar to offline aggression. In any case, the high
percentage of females who admitted to some form of cyberbullying illustrates a
potentially important difference between off- and online aggression.

Online vs. Offline Aggression
Through qualitative analysis, this study identified thought-provoking similarities,
differences and connections between off- and online aggression. After provocation,
many respondents stated that they would respond aggressively online or in person. This
finding supports previous research that has identified Internet victimization as an
antecedent for future online aggression (Ybarra & Mitchell, 2004b; Kowalski & Limber,
2007; Wolak et al., 2007; Werner, Bumpus & Rock, 2009) and offline aggression as an
antecedent for future offline aggression (Barboza et al., 2009). However, the current

106

results suggested that online aggression can also serve as an impetus for offline
aggression. Most of the current horror stories that depict the offline consequences of
online aggression reflect self-inflicted violence on the part of the victims. However,
results from the present study strongly implied that victims of online aggression may
direct serious violent reactions towards others rather than towards themselves. Multiple
respondents expressed the desire to fight with, kick, hit or beat the classmate who
provoked them on the Internet. An adolescent who bullies someone online might receive
a verbal attack or even a punch in the face the next day at the school yard if the so-called
victim was sufficiently angered. Many respondents referred to online attacks as shallow
and unintimidating, yet some of these same individuals expressed a desire to aggressively
engage their attacker in person if not in an online environment.
Despite the potential for off- and online cyclical violence, qualitative results also
suggested that the effects of online aggression are generally less severe than their offline
counterparts.

A much larger percentage of individuals stated that they would have

behaved in a verbally or physically aggressive manner if they were provoked offline
rather than online. Many suggested that face-to-face encounters make them angrier, more
embarrassed, and create more pressure than do online encounters. These comments
aligned closely with the GAM’s propositions. Increased anger should lead to increased
aggressive behavior. If an online encounter does not sufficiently arouse or anger the
individual in the way that an offline encounter would, it will not result in aggressive
behavior. Likewise, when the feelings produced by online attacks involve pity for the
attacker (as suggested by some participants) rather than anger or frustration for the
victim, they will unlikely result in retaliatory aggression.

107

A few individuals suggested that an online encounter would be worse than an
offline encounter because the public nature of the Internet allows people to read the
comment repeatedly and to share the comment with other people, whereas the comment
exists only in memory in face-to-face encounters. Furthermore, within the context of this
study, the individual would have to depend on an outside source to remove the comment
because it was made on their friend’s profile. However, it is possible that many of the
respondents who did not belong to Facebook did not know how the online social network
functioned, including the implications of the anonymity afforded or not afforded by the
website. Despite the few who feared the potential shame produced from lasting online
comments, participants overwhelmingly agreed that the nature of public aggression is
much more intimidating in offline encounters because the presence of the active audience
demands one’s response.

Many participants seemed to disregard the impact of the

audience in the online environment suggesting that either they did not consider the online
audience, or that an online audience generates far less pressure than an offline audience.
Finally, many participants expressed a desire to understand why their classmate
harassed them before reacting when they were provoked online. The desire to understand
the source of the classmate’s anger before reacting was not expressed when individuals
were probed about how they would react if they were provoked offline.

Online

provocations may result in more thoughtful reactions because they remove many of the
pressures generated from offline provocations. Online provocations more than offline
provocations may thereby present opportunities for adolescents to engage in more
successful self-control efforts because online encounters grant them more time and fewer
pressures. Baumeister (2002) suggested that the resources needed to engage in self-

108

control include personal standards, an active monitoring process and the capacity to alter
one’s behavior. While removed pressures and increased time cannot directly increase an
individual’s self-regulatory capacity, they can assist one’s monitoring process and help
them to remember their personal standards.

Limitations
As with all research projects, this study contained multiple limitations. Attempts
were made to make the online environment in which the high school students participated
in the study as authentic as possible. However, most institutions of secondary education
block online social networking sites, including Facebook from their Internet servers.
Thus, participating in the project from the classroom environment created a degree of
artificiality.

While participation in the survey was completely anonymous so that

students would feel comfortable with acting normally, adolescents who engage in online
aggression most likely do so from their own homes or from another private location. The
anger-inducing provocation scenario likely felt somewhat unauthentic to participants
because it asked them to imagine a hypothetical situation. An actual anger-inducing
situation would expectedly produce an even greater degree of hostile cognition and
affect.
This study was limited by a lack of a manipulation check for behavioral stimuli.
The color, size and location of the faces embedded within the anger-inducing scenario
were likely difficult to ignore. However, a specific question was not included within the
questionnaire to determine whether participants saw the behavioral stimuli, making
treatment condition conclusions more difficult to ascertain. Furthermore, participants

109

answered questions relating to trait aggression and attitude towards aggression before
exposure to stimulus materials which could have possibly primed aggression-related
internal processes.
Behavioral intentions and not behavior comprised the study’s major outcome
variable.

Many scholars argue that behavior should be clearly distinguished from

behavioral intentions because intentions do not always accurately predict an individual’s
actions (Budd, 1987; Weinstein, 2007). However, others contend that behavioral
intentions are the best known predictor of behavior (Ajzen, 2002), and multiple metaanalytic reviews have demonstrated at least moderate links between the two constructs
(Milne, Sheeran & Orbell, 2000; Armitage & Connor, 2001; Webb & Sheeran, 2006).
Some of the sample was inexperienced with Facebook.

Only 41% of the

participants had Facebook accounts, which was well below the national average of 73%
of teenagers with social networking accounts (Pew, 2009). Inexperience with Facebook
directly may have adversely affected their responses. However, regression results were
still consistent with those reported after controlling for Facebook account ownership,
suggesting that inexperience with Facebook was not problematic in this sample. Nearly
half claimed to have been past victims of cyberbullying, nearly a third confessed to
having been prior cyberbullies and nearly 70% knew someone who had been bullied
through technology in the past.
The red tape surrounding research involving minors greatly complicated this
project.

Gaining access to the desired population was very difficult and certain

concessions for what would be considered appropriate and allowable to show minors had
to be considered. It is possible that experimental inductions involving aggressive stimuli

110

may have revealed more significant findings or may have demonstrated stronger effect
sizes if the stimuli could have been more blatantly aggressive.
For example, prior GAM research using aggressive cues has found that exposure
to images of guns or other weapons has significantly increased hostile thoughts in adults
(Bartholow et al., 2005). In light of increased levels of fear about school violence and
massacres, gaining access into high schools with a research project involving aggression
that proposed showing adolescents images of guns or similarly violent content would
have made an already difficult task nearly impossible. The fact that the community in
which the study took place has experienced devastatingly extreme school violence in the
past amplified the potential difficulty of showing guns or other weapons in the schools.

Future Research
Future research involving online aggression and aggression in general should
continue to test the roles of trait self-control and state self-control in aggressive behavior.
Results from the current data set suggest that trait-self-control and state self-control can
inhibit many of the factors that contribute to online aggression; however, they yielded
small effect sizes. This study specifically examined public, identifiable retaliatory online
aggression, but it would be interesting to explore whether high trait self-control can
successfully inhibit private and/or anonymous online aggression as well.

Perceived

anonymity afforded through electronic media may potentially encourage some unlikely
traditional bullies to become cyberbullies, especially in instances of retaliation.
Furthermore, the Internet characteristically provides asynchronous communication and a
significant physical barrier. The former removes time pressure by granting an unlikely

111

bully the opportunity to carefully formulate words and craft an attack, while the latter, in
combination with anonymity, removes the threat of immediate retaliation.
Self-control characteristically involves avoiding behaviors that may be bad for the
individual in the long-term, but perceived consequences of electronic aggression may
differ substantially with private and anonymous communication. If no one can identify
the perpetrator or if privacy removes normative expectations, this particular dimension of
self-control may be less useful. At the same time, trait self-control also characteristically
assists the individual in controlling thoughts and anger: two avenues that lead to
aggressive behavior from the perspective of the GAM. Self-control should similarly
affect hostile thoughts and state hostility regardless of whether the behavior occurs
anonymously or openly.
Through descriptive statistics which suggested that females are nearly twice as
likely to have used a technology such as text messages, chat rooms or email to aggress,
the present research provided preliminary support for the notion that females are more
likely than males to use forms of electronic communication for aggression. It also
supported the notion that females are just as likely to aggress online as males when
aggression is enacted publicly and the perpetrator is identifiable. These findings are
mainly supported through descriptive statistics, but either notion cannot be definitively
considered until more rigorous tests on the subject matter are conducted. Future research
should address whether electronic mediums truly enable bullying for females in a way
that offline environments do not.
Future research should also explore whether motivation significantly mediates the
relationship between trait self-control and state self-control. This study demonstrated that

112

the relationship between trait self-control and state self-control is significant, but weak.
What factors keep an individual who maintains high levels of trait self-control (at least
through self-report) from actively utilizing that resource when facing aggressive stimuli?
The strongest indicators of the self-monitoring component were absent from the present
study’s measure of state self-control. Refinement of the state self-control scale may lead
future studies to find a stronger link between the two variables.
In combination with many other studies, this research emphasizes the difficulty of
inhibiting aggressive behavior for high trait aggression individuals. Trait self-control
failed to inhibit aggressive behavioral intentions for low trait aggression individuals with
high levels of state hostility.

Additionally, the provocation scenario without any

aggressive imagery produced enough state hostility for high trait aggression individuals
to behave aggressively without the added aggressive stimulus. Methods promoting the
inhibition of aggressive behaviors may be most successful for individuals who are not
chronically primed to exhibit aggressive behaviors; yet high trait aggression individuals
are the ones who need behavioral interventions the most.
Rather than looking at all individuals, future research should more specifically
identify high trait aggression individuals and examine how to inhibit their aggressive
behaviors.

Understanding how to inhibit aggressive behaviors for non aggressive

adolescents can decrease instances of aggression, but aggressive individuals
disproportionately contribute to overall societal aggression. Impeding aggressive
tendencies is important for everyone; however, it is particularly important for high trait
aggression individuals. They are the most likely candidates for destructive and habitual
aggressive behavior. This study provided promising results in that high trait self-control

113

successfully inhibited aggressive behavioral intentions for those high in hostile cognition
and for those low in comfort and high state control successfully inhibited aggressive
behavioral intentions for those high in state hostility regardless of level of trait
aggression. Counteracting retaliatory contemplations for high trait aggression individuals
is particularly important.

Behavioral Stimuli
One of this study’s primary goals was to consider the effects of an anti-aggressive
behavioral stimulus on impeding hostile cognition and affect. The goal was not achieved
for cognition; however, future research should consider other anti-aggressive or prosocial behavioral stimuli. Previous research has demonstrated the effectiveness of prosocial video games and pro-social song lyrics in promoting pro-social behavior
(Greitemeyer, 2009; Greitemeyer & Osswald, 2010). However, the behaviors in question
in the previous studies dealt with helping other people rather than with regulating one’s
own behavior. Pro-social behavioral stimuli such as the peace symbol were removed
from this study after the pretest because they generated convoluted and dissimilar
meanings among pretest participants. However, perhaps the right pro-social or antiaggressive behavioral stimulus can effectively curtail hostile cognition.
Future research can also explore whether positive facial expressions can more
strongly counter the effects of provocation scenarios when the visual stimulus is clearly
separated from the content of the message. Facial expression symbols were placed
outside of the comment box in the current study’s stimulus materials. They were not
intended to be interpreted as messages from the provoker; however, the placement of the

114

behavioral stimulus within the stimulus material may have confused participants. Future
studies may also choose to directly test the differences between symbols interpreted as
coming from the provoker and symbols interpreted as coming from an outside source.

Online vs. Offline Aggression
Preliminary cyberbullying studies rely on assumptions about electronic medium
characteristics to draw inferences about differences between off- and online aggression.
While these assumptions may be well grounded, more empirical research is needed to
support the suppositions. This study looked specifically at online aggression perpetrated
in a public environment by a known perpetrator, yet online aggression can also occur
privately and anonymously. Many cyberbullying assumptions likely depend largely on
which type of online aggression occurs. Future studies should focus on the effects of
these types of aggressive behaviors because many of the paths to and consequences of
aggressive retaliatory behavior would likely change.
Depending on which electronic platform a cyber-bully uses to express aggression
towards another individual, the bully can be afforded complete anonymity. Over half of
the cyberbully victims in Kowalski and Limber’s (2007) study did not know the identity
of the person who bullied them. This element of cyber-bullying differs substantially from
traditional bullying in that a perpetrator does not necessarily have to know his/her victim,
and victims are blind to who harasses them. In one of the first studies involving hostile
behavior in computer mediated communication, uninhibited remarks such as namecalling, swearing and insults appeared more frequently in the anonymous computer
mediated group than in the identifiable computer mediated communication group and the

115

face-to-face group (Siegel, Durbrovsky, Kiesler, & McGuire, 1986). Aggressive behavior
through online social interactions may therefore have little perceived consequences and
less perceived social risk (Caplan, 2005; Walther, 1996), thereby reinforcing the
behavior. Anonymity also easily enables a victim of traditional bullying by providing a
medium through which s/he can easily retaliate.

The literature surrounding this

possibility is inconsistent and incomplete.
The impact of normative influences on cyber-bullying is also likely domain
specific.

Because cyber-bullying frequently occurs in private settings (e.g. text

messaging and e-mail) or is enacted anonymously where the perpetrator is not subjected
to public judgment, the values and beliefs of reference groups towards the behavior are
less likely to impact the individual as strongly as they would in a more public domain.
Anonymity diffuses the overall utility of consideration of future consequences as it
pertains to social norms that reject cyber-bullying. Bagozzi et al. (2000) suggested that
normative influences are likely moderated by the extent to which a behavior is enacted
publicly or privately, where perceived social norms would likely exert little influence on
behavior in completely private settings where actions are likely to be unknown by others.
Future online aggression scholars can enjoy the benefits of grounding their
research in a long line of prior aggression research, but should also explore new
components of aggression as modified by electronic characteristics.

The behavior

presents many clear similarities to traditional bullying, but it also presents some potential
and intriguing divergences. Descriptive statistics presented by preliminary studies are
helpful in guiding prospective online aggression research. However, upcoming work on
the behavior can remove many of the speculative assumptions about online aggression by

116

moving beyond descriptive statistics to systematically and scientifically focus on the
manners through which cyberbullying modifies existing conceptualizations of aggression.

Conclusion
This study contributed to online aggression research and aggression research in
general by examining the effects of behavioral stimuli on cognition and affect, by
presenting new important variables for the resident theory in media aggression, by
demonstrating manners through which online aggression can be curtailed, and by
identifying differences and similarities between on- and offline aggression. Contrary to
most previous research, the behavioral stimuli used in this study impacted the affective,
rather than the cognitive route to aggressive behaviors. Future research should explore
whether, as suggested, it is the type of behavioral stimulus that makes the difference
between which route to aggressive behavior is taken.
The study presented trait self-control and state self-control as new variables for
the General Aggression Model. Trait self-control was shown to significantly moderate
the negative effects of cognition, affect and comfort on aggressive behavioral intentions
while state self-control was shown to significantly moderate the negative effects of
cognition and affect. Adolescents with highly hostile thoughts and low comfort levels
were less likely to engage in aggressive behaviors if they also maintained high trait selfcontrol. The study demonstrated that trait and state self-control are negatively related to
aggressive behavioral intentions and that a positive relationship between trait and state
self-control exists.

The combined impact of trait and state self-control present a

potentially powerful force through which future aggressive behaviors can be curtailed.

117

Current self-control and self-regulation research demonstrates that individuals can train
themselves to engage in effective self-control and improve their capacity for trait selfcontrol (Baumeister, 2002) which can be used to inhibit future incidents of aggressive
behaviors. Steps towards building one’s self-regulatory resource are incredibly important
for high trait aggression individuals who emerged as the least likely candidates for high
self-control and the most likely candidates for aggressive behavior.
Finally, this study explored differences between off- and online aggression. The
implications of anonymous and private online aggression were not explored because
participants were provoked through a public environment by a known perpetrator.
Qualitative results suggested that retaliatory aggression produced from offline
provocation is much more severe than aggression produced from online provocation in
that it produces more anger, embarrassment and pressure for victims and many
individuals expressed the desire to retaliate not only verbally, but physically as well.
Online aggression presented an important characteristic that was reasonably
absent from offline aggression. Separation from the provoker in a physical sense affords
the potential for delayed response in which individuals have time to think about taking
the best course of action rather than being consumed by the immediate pressure of
responding.

The online environment, which characteristically involves physical

separation and asynchronous communication, may provide a more effective platform for
self-control than the offline environment because it allows people to stop and think on
their own before acting. Future research should empirically test this proposition and it
should explore the question of how much time people wait before responding to insults
online.

118

At last, one of the most important, if not troublesome findings from the study was
the potential for online aggression to result in offline aggression. Many individuals
expressed a desire not only to retaliate online, but to find their provoker in person and to
engage in verbal or physically aggressive behavior. Especially when the perpetrator is
known, online aggression can result in a potentially vicious cycle that is not necessarily
confined to the boundaries of electronic mediums. There is still much to learn about the
idiosyncrasies of online aggression; however, this study provided a valuable platform for
understanding how it functions, how it can be inhibited, and how it should be explored in
the future.

119

APPENDICES

120

APPENDIX A
Figure 10 Stimulus Materials for Aggressive Condition
Imagine yourself in the following situation:
You sign into your Facebook account and you see that a classmate from school has
posted a humiliating comment about you on your friend’s wall for everyone to see. The
classmate harasses you in the message. Imagine that your name is on the blank line.
This is what you see:

Wall

Info

Photos

+

What’s on your mind?

Share
Settings

Just Friends
______________________________________ is such an UGLY
LOSER!!! I wouldn’t touch that freak with a ten-foot pole and
neither would anybody else! ugh
14 minutes ago Comment Like

*For interpretation of the references to color in this and all other figures, the reader is
referred to the electronic version of this dissertation.

121

APPENDIX B
Figure 11 Stimulus Materials for Anti-Aggressive Condition
Imagine yourself in the following situation:
You sign into your Facebook account and you see that a classmate from school has
posted a humiliating comment about you on your friend’s wall for everyone to see. The
classmate harasses you in the message. Imagine that your name is on the blank line.
This is what you see:

Wall

Info

Photos

+

What’s on your mind?

Share
Just Friends

Settings

______________________________________ is such an UGLY
LOSER!!! I wouldn’t touch that freak with a ten-foot pole and
neither would anybody else! ugh
14 minutes ago Comment Like

122

APPENDIX C
Figure 12 Stimulus Materials for Control Condition
Imagine yourself in the following situation:
You sign into your Facebook account and you see that a classmate from school has
posted a humiliating comment about you on your friend’s wall for everyone to see. The
classmate harasses you in the message. Imagine that your name is on the blank line.
This is what you see:

Wall

Info

Photos

+

What’s on your mind?

Share

Just Friends

Settings

______________________________________ is such an UGLY
LOSER!!! I wouldn’t touch that freak with a ten-foot pole and
neither would anybody else! ugh
14 minutes ago Comment Like

123

APPENDIX D
Measurement Items
A. Trait Aggression
1. I yell at others when they annoy me.
2. I react angrily when provoked by
others.
3. I get angry when frustrated
4. I have temper tantrums.
5. I have damaged things because I felt
mad.
6. I have hurt others to win a game.
7. I become angry or mad when I don’t
get my way.
8. I get angry when I lose a game.
9. I get angry when others threaten me.
10. I feel better after yelling at or
hitting someone when they are mean
to me.
11. I hit others to defend myself.
12. I get angry or mad or hit others
when teased.

C. Attitude
1. Suppose a boy says something bad to
another boy, John.
Do you think it’ ok for John to
scream at him?
2. Do you think it’s OK for John to hit
him?
3. Suppose a boy says something bad
to a girl.
Do you think it’s wrong for the girl
to scream at him?
4. Do you think it’s wrong for the girl
to hit him?
5. Suppose a girl says something bad
to another girl, Mary.
Do you think it’s OK for Mary to
scream at her?
6. Do you think it’s OK for Mary to hit
her?
7. Suppose a girl says something bad
to a boy. Do you think it’s wrong
for the boy to scream at her?
8. Do you think it’s wrong for the boy
to hit her?
9. Suppose a boy hits another boy,
John. Do you think
it’s wrong for John to hit him
back?
10. Suppose a boy hits a girl. Do you
think it’s OK for the girl to hit him
back?
11. Suppose a girl hits another girl,
Mary. Do you think it’s wrong for
Mary to hit her back?

B. Trait self-control
1. I am good at resisting temptation.
2. I have a hard time breaking
habits.*
3. I am lazy.
4. I say inappropriate things.*
5. I do certain things that are bad for
me, if they are fun.*
6. I refuse things that are bad for me.
7. I wish I had more self-discipline.*
8. People would say that I have iron
self-discipline.
9. Pleasure and fun sometimes keep
me from getting my work done.*
10. I have trouble concentrating.*
11. I am able to work effectively
toward long-term goals.
12. Sometimes I can’t stop myself
from doing something even if I
know it’s wrong.*
13. I often act without thinking
through all the alternatives.

D. State Hostility
1. I feel angry
2. I feel peeved.
3. I feel aggravated.
4. I feel grovely.
5. I feel annoyed.
6. I feel resentful.
124

7. I feel bitter.
8. I feel furious.
9. I feel polite.*
10. I feel kind.*

5. I thought that I should humiliate my
classmate even if it was bad for me
in the long run.
6. I used a lot of self-discipline in the
situation.
7. I refused to humiliate my classmate
because I thought that it would be
bad for me.
8. It was hard to concentrate when I
was deciding how to react to my
classmate.
9. People would say that I had iron selfdiscipline in this situation.
10. I could not stop myself from treating
my classmate the same way they
treated me even if I thought that it
was wrong.

E. Hostile thoughts
1. I want to get revenge
2. I just want to hurt this person as
bad as s/he hurt me.
3. I want to treat this person like s/he
treated me.
4. When someone attacks me like this
person did, I attack them back.
5. I’ll show this person!
6. I should do something to this
person.
7. I have to get this person back.
8. I want to get back at this person.
9. This person needs to be taught a
lesson.

I. Behavior
1. Send an angry private message back
to the classmate who wrote on my
friend’s wall.
2. Post a humiliating comment about
my classmate on my friend’s
Facebook wall.
3. Embarrass my classmate in front of
other people on Facebook.
4. Curse back at my classmate in
person.
5. Ask my friend to remove the
insulting comment from their wall.
6. Use Facebook to treat my classmate
the same way that I was treated.
7. Threaten my classmate on
Facebook.
8. Humiliate my classmate in front of
other people while at school.
9. Insult my classmate to their face.
10. Curse back at my classmate online.
11. Humiliate my classmate in front of
other people while at school.

F. Comfort
1. I feel comfortable
2. I feel pleasant
3. I feel cozy
4. I feel full of pep
5. I feel cheerful
6. I feel rested
7. I feel snug
8. I feel soothed
9. I feel uncomfortable*
G. Arousal
1. I feel active
2. I feel energetic
3. I feel lively
H. State self-control
1. I could not resist the temptation to
treat my classmate the same way I
was treated.
2. I could not stop myself from reacting
angrily towards my classmate.
3. I had a hard time breaking the habit
of reacting in a negative way.
4. I wanted to say inappropriate things.

125

REFERENCES

126

REFERENCES

Agatston, P.W., Kowalski, R. & Limber, S. (2007). Youth violence and electronic
media: similar behaviors, different venues? Journal of Adolescent Health, 41(6),
S59-S60.
Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl &
J. Beckmann (Eds.), Action control: From cognition to behavior. Berlin,
Heidelber, New York: Springer-Verlag.
Ajzen, I. (2002). Residual effects of past on later behavior: Habituation and reasoned
action perspectives. Personality and Social Psychology Review, 6, 107-122.
Anderson, C.A., Deuser, W.E. & DeNeve. (1995). Hot temperatures, hostile affect,
hostile cognition, and arousal: tests of a general model of affective aggression.
Society for Personality Psychology, 21(5), 434-448.
Anderson,C.A. (1997). Effects of violent movies and trait aggression on hostile feelings
and hostile thoughts. Aggressive Behavior, 23, 161-178.
Anderson, C.A., Anderson, K.B. & Dueser, W.E. (1996). Examining an affective
aggression framework weapon and temperature effects on hostile thoughts, affect,
and attitudes. Personality and Social Psychology Bulletin, 22(4), 366-376.
Anderson, CA, Benjamin, A.J., & Bartholow, B.D. (1998). Does the gun pull the trigger?
Automatic priming effects of weapon pictures and weapon names. Psychological
Science, 9(4), 308-314.
Anderson, C.A., Berkowitz, L., Donnerstein, E., Huesmann, R., Johnson, J.D., Linz, D.,
Malamuth, N.M. & Wartella, E. (2003). The influence of media violence on
youth. Psychological Science in the Public Interest, 4: 81-110.
Anderson, C.A. & Bushman, B.J. (2002). Human Aggression. Annual Review of
Psychology, 53, 27-51.
Anderson, C.A. & Carnagey, N.L. & Eubanks, J. (2003). Exposure to violent media: the
effects of songs with violent lyrics on hostile thoughts and feelings. Journal of
Personality and Social Psychology, 84(5), 960-971.
Anderson, C.A. & Dill, K.E. (2000). Video games and hostile thoughts, feelings, and
behavior in the laboratory and in life. Journal of Personality and Social
Psychology, 78, 772–90
Archer, J. & Southall, N. (2009). Does cost-benefit anayalsis or self-control predict
involvement in bullying by male prisoners? Aggressive Behavior, 35(31), 31-40.
127

Armitage, C.J. & Conner, M. (2001). Efficacy of the theory of planned behavior: A
meta-analytic review. British Journal of Social Psychology, 40, 471-499.
Arneklev, B.J., Grasmick, H.G., Tittle, C.R., Bursik, R.J. (1993). Low self-control and
imprudent behavior. Journal of Quantitative Criminology, 9(3); 1573-7799.
Austin, W.E. (1993). Exploring the effects of active parental mediation of television
content. Journal of Broadcasting & Electronic Media, 37(2), 147-158.
Austin, E.W., Pinkleton, B.E., Fujioka, Y. (2000). The role of interpretation processes
and parental discussion in the media’s effects on adolescents’ use of alcohol.
Pediatrics, 105(2), 343-349.
Avakame, E.F. (1998). Intergenerational Transmission of Violence, Self-Control, and
Conjugal Violence: A Comparative Analysis of Physical Violence and
Psychological Aggression. Violence and Victims, 13(3).
Bagozzi, R.P., Wong, N., Abe, S. & Bergami, M. (2000). Cultural and situational
contingencies and the theory of reasoned action: application to fast food restaurant
consumption. Journal of Consumer Psychology, 9(2), 97-106.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice Hall.
Bandura, A. (1991). Social cognitive theory of self-regulation. Organizational Behavior
and Human Decision Processes, 50; 248-287.
Bandura, A., Ross, D., & Ross, S. A. (1961). Transmission of aggression through
imitation of aggressive models. The Journal of Abnormal and Social Psychology,
63, 575-582.
Barboza, G.E., Schiamberg, L.B., Oehmke, J., Korzeniewski, S.J., Post, L.A. & Heraux,
C.G. (2009). Individual characteristics and the multiple contexts of adolescent
bullying: an ecological perspective. Journal of Youth and Adolescence, 38, 101121.
Bargh, J.A., Lambardi, W.J., & Higgins, E.T. (1998). Automaticity of person X situation
effects on impression formation: It’s just a matter of time. Journal of Personality
and Social Psychology, 55 (4), 599-605.
Bartholow, B.D., Anderson, C.A., Carnagey, N.L. & Benjamin, A.J. (2005). Interactive
effects of life experience and situational cues on aggression: the weapons priming
effect in hunters and non hunters. Journal of Experimental Social Psychology, 41,
48-60.

128

Baumeister, R.F. (2002). Reflections and review. Yielding to temptation: self-control
failure, impulsive purchasing, and consumer behavior. Journal of Consumer
Research, 28.
Baumesiter, R.F., DeWall, Ciarocco, N.J. & Twenge, J.M. (2005). Attitudes and social
cognition social exclusion impairs self-regulation. Journal of Personality and
Social Psychology, 88(4), 589-604.
Baumeister, R.F. & Exline, J.J. (2000). Self-control, morality, and human strength.
Journal of Social and Clinical Psychology; 19(1), 29-42.
Baumeister, R.F., Gailliot, M., DeWall, N. & Oaten, M. (2006). Increase regulatory
success, and how depletion moderates the effects of traits on behavior. Journal of
Personality, 74(6).
Baumeister, R.F., Heatherton, T.F., & Tice, D.M. (1994). Losing control. How and why
people fail at self-regulation. San Diego, CA: Academic Press.
Baumeister, R.F. & Vohs, K.D. (2003). Self-regulation and the executive function of the
self. Leary, M.R. & Tangney, J.P. (Eds.) In Handbook of Self and Identity.
Guilford Press: New York, NY.
Beran, T. & Li, Q. (2005). Cyber-harassment: A study of a new method for an old
behavior. Journal of Educational Computing Research, 32(3), 265-277.
Berkowitz L, LePage A. 1967. Weapons as aggression-eliciting stimuli. Journal of
Personality and Social Psychology, 7, 202–207.
Boulton, M., Bucci, E., & Hawker, D. (1999). Swedish and English secondary school
pupils’ attitudes towards, and conceptions of, bullying: Concurrent links with
bully/victim involvement . Scandinavian Journal of Psychology, 40, 277–284.
Brame, B. (2001). Developmental trajectories of physical aggression from school entry
to late adolescence. Journal of Child Psychology and Psychiatry, 42(4), 503-512.
Brewer, M.B. (1988). A dual process model of impression formation. In T.K. Scrull &
R.S. Wyer, Jr. (Eds.), Advances in Social Cognition (Vol. 1, pp. 1-36). Hillsdale,
NJ: Earlybaum. Fiske & Neuberg, 1990
Budd, R.J. (1987). Response bias and the theory of reasoned action. Social Cognition,
5, 95-107.
Bushman, B.J. (1998). Priming effects of media violence on the accessibility of
aggressive constructs in memory. Personality and Social Psychology, 24(5), 537545.

129

Bushman, B.J. & Anderson, C.A. (2001). Media violence and the American public:
Scientific facts versus media misinformation, American Psychologist, 56(6/7),
477-489.
Bushman, B.J. & Geen, R.G. (1990). Role of cognitive-emotional mediators and
individual differences in the effects of media violence on aggression. Journal of
Personality and Social Psychology, 58(1), 156-163.
Bushman, B.J. & Huesmann, L.R. (2000). Effects of televised violence on aggression.
In Singer, D.G. & Singer, J.L., Handbook of Children and the Media. Sage
Publications, Inc.; Thousand Oaks, CA.
Bushman, B.J., Huesmann, R. (2006). Short-term and long-term effects of violent media
on aggression in children and adults. Archives of Pediatric and Adolescent
Medicine, 160, 348-352.
Bushman, B.J., Phillips, C.M. & Baumeister, R.F. (2000). Do people aggress to improve
their mood? Catharsis beliefs, affect regulation opportunity, and aggressive
responding. Journal of Personality and Social Psychology, 81(1), 17-32.
Cantor, J. & Wilson, B.J. (2003). Media and violence: Intervention strategies for
reducing aggression. Media Psychology, 5; 363-403.
Caplan, S. (2005). A social skill account of problematic Internet use. Journal of
Communication, 55 (4), 721–736.
Carver, C.S., & Scheir, M.F. (1998). On the Self-Regulation of Behavior. New York:
Cambridge University Press.
Children Now. (2000). Girls and gaming: A console video game content analysis.
Oakland, CA. Retrieved from http://www.childrennow.org/media/video-games
girls.pdf.
Chory-Asaad, R.M. (2004). Effects of television sitcom exposure on the accessibility of
verbally aggressive thoughts. Western Journal of Communication, 68(4), 431453.
Colder, C.R. & Stice, E. (1998). A longitudinal study of the interactive effects of
impulsivity and anger on adolescent problem behavior. Journal of Youth
Adolescence (77), 255-274.
Coleman, R. (2003). Race and ethical reasoning: The importance of race to journalistic
decision-making. Journalism & Mass Communication Quarterly, 80 (2), 295310.
Craig, W.M., Pepler, D. & Atlas, R. (2000). Observations of bullying in the playground
130

and in the Classroom. School Psychology International, 21(1), 22-36.
Deffenbacher, J.L., Petrilli, R.T., Lynch, R.S., Oetting, E.R. & Swan, R.C. (2003).
Thoughts Questionnaire: A measure of Angry Cognitions When Driving.
Cognitive Therapy Research, 27(4).
th

DeFleur, M. & Ball-Rokeach, S. (1989). Theories of Mass Communication. 5 Ed.
White Plains, NY: Longman.
Del Signore, J. (2010, March, 25). Cops investigate cyberbully role in teen’s suicide.
Gothamist. Retrieved July 15, 2009 from
http://gothamist.com/2010/03/25/cops_investigate_cyberbully_role_in.php.
Denson, T.F., Aviles, F.E., Pollock,m V.E., Earleywine, M., Vasquez, E.A. & Miller, N.
(2007). The effects of alcohol and the salience of aggressive cues on triggered
displaced aggression. Aggressive Behavior, 34(1), 25-33.
Derks. D. & Grumbkow, J. (2007). Emoticons and social interaction on the internet: the
importance of social context. Computers in Human Behavior, 23(1), 842-849.
DeWall, C.N., Baumeister, R.F., Stillman, T.F. & Gaiilliot, M.T. (2007). Violence
restrained: effects of self-regulation and its depletion on aggression. Journal of
Experimental Social Psychology, 43(1), 62-76.
Devine, P.G. (1989). Stereotypes and prejudice: Their automatic and controlled
components. Journal of Personality and Social Psychology, 56, 5-18.
Dibbell, J. (1993). A rape in cyberspace. The Village Voice, 38(51); 36.
Dixon, T. & Linz, D. (2000). Overrepresentation and under representation of Africa
Americans and Latinos as lawbreakers on television news. Journal of
Communication, 51(2): 134-154.
Duckworth, A.L. & Seligman, M.E.P (2006). Self-discipline gives girls the edge: gender
in self-discipline, grades and achievement test scores. Journal of Educational
Psychology, 98(1), 198-208.
Eagly, A.H., & Chaiken, S. (1993). The Psychology of Attitudes. Fort Worth, TX:
Harcourt, Brace, Jovanovich.
Eastin, M.S., Greenberg, B.S. & Hofschire, L. (2006). Parenting the Internet. Journal of
Communication, 56(3); 486-504.
Farrar, K. & Krcmar, M. (2006). Measuring state and trait aggression: A short,
131

cautionary tale. Media Psychology; 8(2), 127-138.
Ferguson, C.J. (2009). Media violence effects: confirmed truth or just another x-file?
Journal of Forensic Psychology Practice, 9(2), 103-126.
Ferguson, C.J., San Miguel, C., Kilburn, J.C. & Sanchez, P. (2007). The effectiveness of
school-based anti-bullying programs: a meta-analytic review. Criminal Justice
Review, 32; 401.
Finkel, E.J. & Campbell, W.K. (2001). Self-control and accommodation in close
relationships; an interdependence analysis. Journal of Personality and Social
Psychology, 81(2), 263-277.
Finkel, E.J., DeWall, C.N., Slotter, E.B., Oaten, M. & Foshee, V.A. (2009). Selfregulatory failure and intimate partner violence perpetration. Journal of
Personality and Social Psychology, 97(3), 483-499.
Finkenauer, C., Engels, R.C.M.E. & Baumeister, R.F. (2005). Parenting behavior and
adolescent behavioural and emotional problems: The role of self-control.
International Journal of Behavioral Development, 29(1), 58-69.
Fleeson, W. (2007). Situation-based contingencies underlying trait-content
manifestation in behavior. Journal of Personality, 7(4), 825-862.
Friedrich, L.K. & Huston-Stein, A. (1973). Aggressive and prosocial television programs
and the natural behavior of preschool children. Monographs of the Society for
Research in Child Development, 38(4).
Freedman, J.L. (1988). Television violence and aggression: What the evidence shows.
In S. Oskamp (Ed.), Applied social psychology annual: Television as a social
issue (Vol. 8, pp. 144-162). Newbury Park, CA: Sage.
Fujioka, Y. (1999). Television portrayals and African-American stereotypes.
Journalism Quarterly, 76(1): 52-75.
Gebhart, B. (2009, December 12). Raising awareness about cyberbullying. KCBS.
Gentile, D.A., Lynch, P.J., Linder, J.R. & Walsh, D.A. (2004). The effects of violent
game habits on adolescent hostility, aggressive behaviors, and school
performance. Journal of Adolescence, (27), 5-22.
Gerbner, G., & Gross, L. (1976). Living with television: The violence profile. Journal of
Communication, 26(2), 172-199.
Giumetti, G.W. & Markey, P.M. (2007). Violent video games and anger as predictors of
132

aggression. Journal of Research in Personality, 41(6), 1234-1243.
Grandey, A.A., Dickter, D.N. & Sin, H. (2004). The customer is not always right:
customer aggression and emotion regulation of service employees. Journal of
Organizational Behavior, 25, 1-22.
Greenberg, B.S. The effects of language intensity modification on perceived verbal
aggressiveness. Communication Monographs,43(2), 130-139.
Greenberg, B.S. (1980). Life on television : Content Analyses of U.S. TV Drama.
Norwood, NJ: Ablex.
Greenberg, B. & Salwen (1996). Mass communication theory and research: concepts
and models. In Salwen, B & Stacks, D.W. An Integrated Approach to
Communication Research and Theory. Lawrence Erlbaum Associates, Inc;
Mahwah, NJ.
Greitemeyer, T. (2009). Effects of songs with prosocial lyrics on prosocial behavior:
further evidence and a mediating mechanism. Personality and Social Psychology
Bulletin, 35(11), 1500-1511.
Greitemeyer, T. & Osswald, S. (2010). Effects of prosocial video games on prosocial
behavior. Journal of Personality and Social Psychology, 98(2), 211-221.

Gross, J.J. & Levenson, R.W. (1997). Hiding feelings: the acute effects of inhibiting
negative and positive emotion. Journal of Abnormal Psychology, 106, 95-103.
Hawker, D. S. J., & Boulton, M. J. (2000). Twenty years research on peer victimization
and psychosocial maladjustment: A meta-analytic review of cross-sectional
studies. Journal of Child Psychiatry and Psychiatry, 41, 441–455.
Heatherton, T.F. (1996). Self-regulation failure: An overview. Psychological Inquiry,
7(1); 1-15.
Hennessy, D.A. & Schwartz, S. (2007). Personal predictors of spectator aggression at
little league baseball games. Violence and Victims, 22, 205-215
Hinduja, S. & Patchin, J.W. (2008). Cyber-bullying: An exploratory analysis of factors
related to offending and victimization. Deviant Behavior, 29(2), 129-156.
Hoffmann, W., Friese, M. & Strack, F. Impulse and self control from a dual-systems
perspective. Perspectives on Psychological Science, 4(2), 162-176.
Joireman, J., Anderson, J. & Strathman, A. (2003). The aggression paradox:
Understanding links among aggression, sensation seeking, and the consideration
133

of future consequences. Journal of Personality and Social Psychology, 84(6),
1287-1302.
Kawakami, K., Spears, R., & Dovidio, J. (2002). Disinhibition of stereotyping: Context,
prejudice and target characteristics. European Journal of Social Psychology, 32,
517-530.
Kayany, J.M. (1998). Contexts of uninhibited online behavior: flaming in social
newsgroups on Usenet. Journal of the American Society for Information Science,
49(19): 1135-1141.
Kiesler, S., Siegel, J. & McGuire, T.W. (1984). Social psychological aspects of
computer-mediated communication. American Psychologist, 39(10), 1123-1134.
Kim, V. (2009, December 14). First Amendment protects kids’ right to be nasty online.
Los Angeles Times.
Kinney, T.A. (1994). An inductively derived typology of verbal aggression and its
association to distress. Human Communication Research, 21(2), p. 183.
Kirsh, S.J. (2006). Children, Adolescents and Media Violence: A Critical Look at the
Research. Sage Publications, Inc. Thousand Oaks, CA.
Kirsh, S.J. (2003). The effects of violent video games on adolescents. The overlooked
influence of development. Aggression and Violent Behavior, 8(4), 377-389.
Kowalski, R.M, Limber, S.P., Agatston, P.W. (2008). Cyberbullying: Bullying in the
Digital Age. Blackwell Publishing.
Kowalski, R.M., & Limber, S.P. (2007). Electronic bullying
among middle school students. The Journal of Adolescent Health, 41, S22-S30.
Kramer, U.M., Jansma, H., Tempelmann, C. & Munte, T.F. (2007). Tit-for-tat: the
neural basis of reactive aggression. Neurolmage, 38(1), 203-211.
Kunda, Z. & Spencer, S.J. (2003). When do stereotypes come to mind and when do they
color judgment? A goal-based theoretical framework for stereotype activation
and application. Psychological Bulletin, 129 (4), 522-544.
Lavine, H., Sweeny, D. and Wagner, S. (1999). Depicting women as sex objects in
television advertising: effects on body dissatisfaction. Personality and Social
Psychology Bulletin, 25, 1049-1058.
Lea, M., O’Shea, T., Fung, P., & Spears, R. (1992). “Flaming” in computer-mediated
communication: Observations, explanations, implications. In Lea, M. (Ed.),
Contexts of Computer Mediated Communication, 89-112. New York: Harvester134

Wheatsheaf.
Lee, K.C., Perry, S.D. (2004). Student instant message use in a ubiquitous computing
environment: effects of deficient self-regulation. Journal of Broadcasting &
Electronic Media, 48(3); 399-420.
Lee, L. (2010). Cyber bullying through blast sites. My Fox Mepmphis.com. Retrieved
July 18, 2010 from http://www.myfoxmemphis.com/dpp/news/local/071410cyber-bullying-through-blasts-sites.
Leifer, A.D. & Roberts, D.F. (1972). Children’s responses to television violence. In:
Murray, J.P., Rubinstein, E.A. & Comstock, G.A. (Eds.). Television and Social
Behavior, Vol. 2: Television and Social Learning. Rockville, MD: National
Institute of Mental Health. P. 43-180.
Liebert, R.M. & Schwartzberg, N.S. (1977). Effects of Mass Media. Annual Review of
Psychology, 28; 141-173.
Lindsay, J.J. & Anderson, C.A. (2000). From antecedent conditions to violent actions:
A General aggression model. Personality and SocialPscyhology Bulletin, 26,
533.
Livingstone, S. & Helsper, E.J. (2008). Parental mediation and children’s Internet use.
Journal of Broadcasting and Electronic Media, 52(4), 581.
Maag, C. (2007, December 16). When the bullies turned faceless. The New York Times.
Meier, B.P., Robinson, M.D. & Wilkowski, B.M. (2007). Aggressive primes activate
hostile information in memory: who is most susceptible? Basic and Applied
Social Psychology,
29(1), 23-34.
McGuire, W.J. (1986). The myth of massive media impact: Savagings and salvaging. In
G. Comstock (Ed.), Public Communication and Behavior (Vol. 1, pp. 173-257).
San Diego, CA: Academic Press.
McNair, Lorr & Droppleman, L.F. (1992). The Profile of Mood States. San Diego:
Educational & Industrial Testing Service.
Miller, C. (2006). Cyber Harassment: Its forms and perpetrators. Law Enforcement
Technology, 33(4), 26-30.
Milne, S., Sheeran, P., Orbell, S. (2000). Prediction and intervention in health related
behaviour: A meta-analytic review of Protection-Motivation Theory. Journal of
Applied Social Psychology, 30, 106-143.

135

Mischel, W. (1968). Personality and Assessment. New York: John Wiley.
Moller, I. & Krahe, B. (2009). Exposure to violent video games and aggression in
German adolescents: a longitudinal analysis. Aggressive Behavior, 35, 75-89.
Muraven, M. & Baumeister, R.F. (2000). Self-regulation and depletion of limited
resources: does self-control resemble a muscle. Psychological Bulletin, 126(2);
247-259.
Muraven, M., Tice, D.M. & Baumeister, R.F. (1998). Self-control as a limited resource:
regulatory depletion patterns. Journal of Personality and Social Psychology, 74,
774-789.
Muraven, M., Baumeister, R.F., & Tice, D.M. (1999). Longitudinal improvement of
self-regulation through practice: Building self-control strength through repeated
exercise. The Journal of Social Psychology, 139(4); 446.
Muraven, M. & Slessareva, E. (2003). Mechanisms of self-control failure: motivation
and limited resources. Personality and Social Psychology Bulletin,29, 894-906.
Nathanson, A.I. (1999). Identifying and explaining the relationship between parental
mediation and children's aggression. Communication Research, 26; 124.
Olweus, D. (1999). Sweden. In Smith, P. K., Morita, Y., Junger- Tas, J., Olweus, D.,
Catalano, R. & Slee, P. (Eds) The nature of school bullying: A cross-national
perspective, 7–27 . London & New York: Routledge.
O’Connor, D.B., Archer, J. & Wu, F. (2001). Measuring aggression: self reports, partner
reports, and responses to provoking scenarios. Aggressive Behavior, 27, 70-101.
O’Moore M., & Kirkham, C. (2001). Self-esteem and its relationship to bullying
behaviour. Aggressive Behavior, 27, 269–283.
O'Sullivan, P.B. & Flanagin, A.J. (2003). Reconceptualizing 'flaming' and other
problematic messages. New Media and Society, 5; 69.
Paik, H. & Comstock, G. (1994). The effects of television violence on antisocial
behavior: A meta-analysis. Communication Research, 21(4), 516-546.
Parks, M.R., Floyd, K. Making friends in cyberspace. Journal of Computer-Mediated
Communication.
Pearce, J.B. & Thompson, A.E. (1998). Practical approaches to reduce the impact of
bullying. British Medical Journal; 79, 528-531.
Pechmann, C., Levine, L., Loughlin, S. & Leslie, F. (2005). Impulsive and self136

conscious adolescents’ vulnerability to advertising and promotion. Journal of
Public Policy and Marketing, 24(2), 202-221.
Pellegrini, A.D., Bartini, M. & Brooks, F. (1999). School bullies, victims, and
aggressive victims: factors relating to group affiliation and victimization in early
adolescence. Journal of Educational Psychology, 91(2), 216-224.
Pew (2009). Pew Internet and American Life Project, September 2009 Tracking Survey.
Philip, T.A. (1993). A social influence model of flaming in computer-mediated
communication. Annual Meeting of the Western States Communication
Association.
Piquero, A.R. & Rosay, A.B. (1998). The reliability and validity of Grasmick et al.’s
self-control scale: A comment on Longshore et al. Criminology, 36(1), 157-174.
Reinig, B.A., Briggs, R.O., Nunamaker, J.F. (1997). Flaming in the electronic classroom.
Journal of Management Information Systems, 14(3); 45.
Rahman, A.K.M.R. & Huq, M.M. (2005). Aggression in adolescent boys and girls as
related to socio-economic status and residential background. Journal of Earth
Science,1(1), 5-9.
Raine, A., Dodge, K., Loeber, R., Gatzke-Kopp, L., Lynam, D., Reynolds, C.,
Stouthamer-Loeber, M., Liu, J. (2006). The reactive-proactive aggression
questionnaire: differential correlates of reactive and proactive aggression in
adolescent boys. Aggressive Behavior; 32(2), 159-171.
Randall, P. (1995). A factor study on the attitudes of children to bullying . AEP
(Association of Educational Psychologists) Journal, 11, 22–26 .
Raskauskas, J. & Stoltz, A.D. (2007). Traditional and electronic bullying among
adolescents. Developmental Psychology, 43(3), 564-575.
Rigby, K., & Cox, I. (1996). The contribution of bullying at school and low self-esteem
to acts of delinquency among Australian teenagers. Personality and Individual
Differences, 21, 609–612.
Rigby, K., Smith, P.K. & Pepler, D. (2004). Working to prevent school bullying: key
issues. In Smith, P.K., Pepler, D. & Rigby, K. (Eds.) Bullying in Schools: How
Successful Can Intervention Be? Cambridge University Press. Cambridge, United
Kingdom.
Roberto, A.J., Meyer, G., Boster, F.J., Roberto, H.L. (2003). Adolescents’ decisions
about verbal and physical aggression. An Application of the Theory of reasoned
137

action. Human Communication Research, 29(1), 135-147.
Rosen, L.D., Cheever, N.A. & Carrier, L.M. (2008). The association of parenting style
and child age with parental limit setting and adolescent MySpace behavior.
Journal of Applied Developmental Psychology, 29, 459-471.
Roskos-Ewoldsen, D.R., Roskos-Ewoldsen, B. & Carpentier, F.D. (1994). Media
Priming. In Bryant, J. & Oliver, M.B. (Eds.) Media Effects Advances in Theory
rd
and Research, 3 Edition, Erlbaum: New York, New York, Chapter 5.
Ruys, K.I. & Stapel, D.A. (2008). Emotion elicitor or emotion messenger? Subliminal
priming reveals two faces of facial expression. Psychological Science, 19(6),
593-600.
Salmivalli, C. & Voeten, M. (2004). Connections between attitudes, group norms, and
behaviour in bullying situations. International Journal of Behavioral
Development, 28(3), 246-258.
Schank, R.C. & Abelson, R.P. (1977). Scripts, Plans, Goals, and Understanding: An
Inquiry into Human Knowledge Structures. Lawrence Erlbaum; Oxford, England.
Scheithauer, H. (2002). Unprototypical forms of aggression among children: A metath
analysis. Presented at 7 Workshop on Aggression, Vienna, Austria, November.
Seay, A.F., Kraut, RE. (2007). Project massive: self-regulation and problematic use of
online gaming. Conference on Human Factors in Computing Systems.
Siegel, J., Dubrovsky, V., Kiesler, S. & McGuire, T.W. (1986). Group processes in
computer-mediated communication. Organizational Behavior and Human
Decision Processes; 37(2), 157-187.
Siever, L.J. (2008). Neurobiology of aggression and violence. The American Journal of
Psychiatry, 165(4), 429.
Smith, I. (1994). Types of bullying behaviour and their correlates. Aggressive
Behaviour, 20, 359-368.
Smith, P.K. (2004). Bullying: recent developments. Child and Adolescent Mental
Health, 9(3), 98–103.
Smith, P.K, Mahdavi, J., Carvalho, M., Fisher, S., Russell, S., Goldsmiths, N.T. (2006).
Cyberbullying: Its nature and impact on secondary school pupils. Journal of
Child Psychology and Psychiatry; 49(4), 376-385.
Sproule, J.M. (1989). Progressive propaganda critics and the magic bullet myth. Critical
Studies in Media Communication, 6(3), 225-246.
138

SPSS, Inc. (2007). Statistical package for the social sciences (Version 17.0 for
Windows). Chicago.
Stucke, T.S. & Baumeister, R.F. (2006). Ego depletion and aggressive behavior: Is the
inhibition of aggression a limited resource? European Journal of Social
Psychology, 36(1), 1-13.
Swearer, S.M. & Doll, B. (2001). Bullying in schools: an ecological framework. Journal
of Emotional Abuse, 2(2 &3); 7-23.
Tangney, J.P., Baumeister, R.F., Boone, A.L. High self-control predicts good
adjustment, less pathology, better grades and interpersonal success. (2004).
Journal of Personality, 72(2); 271-324.
Thompson, P.A. & Foulger, D.A. (1996). Effects of pictographs and quoting on flaming
in electronic mail. Computers in Human Behavior, 12, 225-243.
Tice, D.M., Baumeister, R.F., Shmueli, D. & Muraven, M. (2007). Restoring the self:
positive affect helps improve self-regulation following ego depletion. Journal of
Experimental Social Psychology, 43, 379-384.
Tice, D.M. & Bratslavsky, E. (2000). Giving to feel good: The place of emotion
regulation in the context of general self-control. Psychological Inquiry, 11(3);
149-159.
Tomeo, T. (2007). Noise: How our Media Saturated Culture Ruins Dominates and
Dismantles Families. Ascension Press, LLC; Westchester, PA.
Topalli, V. & O’Neal, E.C. (2003). Retaliatory motivation enhances attributions of
hostility when people process ambiguous social stimuli. Aggressive Behavior,
29(2), 155-172.
Valois, R.F., MacDonald, J.M., Bretous, L., Fischer, M.A., Drane, J.W. (2002).
American Journal of Health Behaviors, 26(6), 454-464.
Vasquez, E.A., Denson, T.F., Pedersen, W.C., Stenstrom, D.M. & Miller, N. (2005).
The moderating effect of trigger intensity on triggered displaced aggression.
Journal of Experimental Social Psychology, 41, 61-67.
Vohs, K.D., Baumeister, R.F. & Ciarocco, N.J. (2005). Self-regulation and selfpresentation: regulatory resource depletion impairs impression management and
effortful self-presentation. Journal of Personality and Social Psychology, 88(4),
632-657.
Vohs, K.D., Baumeister, R.F., Schmeichel, B.J., Twenge, J.M., Nelson, N.M. & Tice,
139

D.M. (2008). Making choices impairs subsequent self-control: a limited resource
account of decision making, self-regulation, and active initiative. Journal of
Personality and Social Psychology, 94(5), 883-898.
Wang, J., Iannotti, R.J. & Nansel, T.R. (1999). School bullying among adolescents in the
United States: physical, verbal, relational and cyber. Journal of Adolescent
Health, 45(4), 368-375.
Walther, J.B. (1992). Interpersonal effects in computer-mediated interaction: a relational
perspective. Communication Research, 19(52).
Walther, J.B. (1996). Computer-mediated communication. Communication Research;
23(1), 3-43.
Walther, J.B. & A’ddario, K.P. (2001). The Impacts of Emoticons on Message
Interpretation in Computer-Mediated Communication. Social Science Computer
Review, 19, 324-347.
Walther, J.B. & Burgoon, J.K. (1992). Relational communication in computer-mediated
interaction. Human Communication Research, 19, 50-88.
Webb & Sheeran, (2006). Does changing behavioral intentions engender behavior
change? A meta-analysis of the experimental evidence. Psychological Bulletin,
132(2), 249-268.
Weinstein, N.D. (2007). Misleading tests of health behavior theories. Annals of
Behavioral Medicine, 33, 1-10.
Werner, N.E., Bumpus, M.F., & Rock, D. (2009). Involvement in Internet Aggression
During Early Adolescence. Journal of Youth and Adolescence, 39(6), 607-619.
Wolak, J., Mitchell, K.J., & Finklehor, D. (2007). Does online harassment constitute
bullying? An exploration of online harassment by known peers and online-only
contacts. Journal of Aolescent Health, 41, S51-S58.
Wood, W., Wong, F.Y. & Chachere, J.G. (1991). Effects of media violence on viewers’
aggression in unconstrained social interaction. Psychological Bulletin, 109, 371383.
Yagil, D. (2001). Interpersonal antecedents of drivers’ aggression. Transportation
Research Part F: Traffic Psychology and Behaviour, 4(2), 119-131.
Ybarra, M.L., Diener-West, M. & Leaf, P.J. (2007). Examining the overlap in Internet
harassment and school bullying: implications for school intervention. Journal of
Adolescent Health, 41, S42-S50.

140

Ybarra, M.L. & Mitchell, K.J. (2004a). Online aggressor/targets, aggresssors, and
targets: a comparison of associated youth characteristics. Journal of Child
Psychology and Psychiatry, 45(7), 1308-1316.
Ybarra, M.L. & Mitchell, K.J. (2004b). Youth engaging in online harassment:
associations with caregiver-child relationships, Internet use, and personal
characteristics. Journal of Adolescence, 27(3), 319-336.
Ybarra, M.L., Mitchell, K.J., Finkelhor, D. & Wolak, J. (2007). Internet prevention
messages: targeting the right online behaviors. Archives of Pediatric Adolescent
Medicine, 161(101), 138-145.

141