Content uploaded by Kirstie M. Farrar
All content in this area was uploaded by Kirstie M. Farrar on Feb 26, 2014
Content may be subject to copyright.
Retaliatory Aggression and the Effects
of Point of View and Blood in
Violent Video Games
Department of Communication
Wake Forest University
Department of Communication Studies
University of Connecticut
In this study, an experimental design was utilized to test, first, the effect of a
violent game versus a no game control on physical and verbal aggression and
retaliatory aggression against a confederate. In addition, the effects of two
internal video game manipulations were explored. Overall, those in the violent
game condition were more verbally and physically aggressive than those in the
no game condition. In terms of internal game features, third-person play with
the blood on, especially when combined with aggressive cognitions and to
some extent, hostile affect, encouraged more aggressive outcomes.
Marina Krcmar (Ph.D., University of Wisconsin-Madison, 1995) is Associate Professor in
the Communication Department at Wake forest University. Her research focuses on children,
adolescents, and the media.
Kirstie Farrar (Ph.D., University of California-Santa Barbara, 2001) is Assistant Professor
in the Department of Communication Studies at the University of Connecticut. Her research
interests include effects of the mass media on individuals, specifically the effects of the mass
media on adolescent socialization.
Correspondence should be addressed to Marina Krcmar, Department of Communication,
Wake Forest University, 1834 Wake Forest Road, Winston-Salem, NC 27106. E-mail:
Mass Communication and Society, 12:115–138
Copyright #Mass Communication & Society Division
of the Association for Education in Journalism and Mass Communication
ISSN: 1520-5436 print=1532-7825 online
There is now substantial evidence that, like exposure to television violence,
exposure to violent video games increases aggression (see Anderson &
Bushman, 2001, for a meta-analysis). However, unlike television, game tech-
nology allows for internal game manipulations (e.g., playing in first or third
person; manipulating the presence of blood and gore). Not much research to
date (see Eastin, 2007, and Farrar, Krcmar, & Nowak, 2006, for examples)
has examined the effect of contextual game features on aggressive outcomes.
In our study, an experimental design is utilized, manipulating two internal
game features—point of view (POV; first vs. third person) and blood (on
or off)—to explore first the mechanisms by which video game exposure
effects aggression and second the role that internal game features may play
in moderating effects.
Game Play and Aggressive Outcomes
Compared to research on television violence, effects research on video game
violence is still in its infancy and has only been accumulating since the mid-
1980s. However, most of the research that has been done is consistent with
the notion that video game violence does affect players in numerous ways. It
should be noted a recent meta-analysis found the effect of video game play
on aggression to be smaller than that typically reported for television
(Anderson & Bushman, 2001; Sherry, 2001), suggesting that perhaps the
process of effects from game play may be somewhat different from the pro-
cess involved in television exposure and effects. Nevertheless, video game
play does seem to result in increases in aggression, and there are several
important theoretical reasons that this may be the case. First, when playing
a violent video game, the individual assumes an active role in the storyline.
Instead of simply watching a television or movie character act violently,
video game players actually choose to aggress themselves and carry out
these actions. Anderson and Dill (2000), in their general aggression model
(GAM), argue that this combination of choice and action in the video game
environment may lead to the construction of a more complete aggressive
script than would be formed by passively viewing television.
Second, violent video games present a more complete learning environ-
ment for aggression than does television. In violent video games, players
are rewarded for acting out aggressive scripts by earning more points or
advancing to new levels. In this way, aggressive players receive direct rein-
forcement, which is arguably a stronger and more vivid reward than
the types of vicarious rewards one might get from watching a television
character rewarded for behaving in an aggressive manner.
116 KRCMAR AND FARRAR
Finally, violent video games present increased chances for identification
with the aggressor. Identification with a media character is known to
enhance the potential for media effects (e.g., Leyens & Picus, 1973). When
playing a game, participants assume the role of the ‘‘hero,’’ and they control
the actions of that character as they navigate the gaming environment. In
addition, players can now choose to play many games with characters that
resemble them on many key features such as gender, race, style of cloth-
ing, and physical appearance. This should enhance identification with the
game character. Identification with the hero of the game is expected to be
particularly enhanced in the types of games know as ‘‘first-person shooters’’
(Anderson & Dill, 2000). In these games, the player literally sees the game
world through the character’s eyes.
In part because of these reasons, substantial evidence now exists that vio-
lent game play is associated with aggressive effects. For example, violent
game play can influence aggressive cognitions, or thoughts (Anderson &
Dill, 2000; Tamborini et al., 2004); can influence aggressive affect, leading
to feelings of hostility (Anderson & Dill, 2000; Ballard & Wiest, 1996;
Farrar et al., 2006, Tamborini et al., 2004); and has been found in survey
research to be associated with aggressive delinquent behavior, even after
controlling for aggressive personality (Anderson & Dill, 2000). Experimen-
tal work using more up-to-date games has also found exposure to violent
video games to be linked to aggressive behavior (Anderson & Dill, 2000;
Cicchirillo & Chory-Assad, 2005; Farrar et al., 2006). Meta-analyses con-
ducted on the research on violent video games to date have also supported
an effect of game play on aggression (Anderson, 2004; Anderson & Bushman,
2004; Sherry, 2001).
The GAM and Violent Games
The GAM, which comprehensively integrates central elements from several
earlier aggression theories including social cognitive theory (Bandura, 1994),
script theory (Huesmann, 1986), cognitive-neoassociation theory (Berkowitz,
1989), Geen’s (1990) affective aggression model, and Zillmann’s (1983) excita-
tion transfer model, attempts to integrate existing knowledge and research on
the learning, development, instigation, and expression ofaggression (Carnagey
& Anderson, 2003). The GAM seeks to explain the development of aggression
across many contexts. Because it simultaneously considers personalogical
factors and the environmental cues that may contribute to the learning of
aggression, it can appropriately be applied to the study of video game violence.
The GAM is based on the premise that aggression results from knowl-
edge structures that develop out of personal experience and can influence
perception from basic visual patterns to complex behavioral sequences. Over
VIDEO GAMES 117
time and with repeated use, these knowledge structures can become
automatized. Furthermore, they become linked to affective states,
behavioral programs, and beliefs such that a particular affect, perhaps
frustration, may become associated with aggression over time, thus guiding
individuals’ behavioral responses to their social environment. Consider the
game play environment: Players actively engage in game play; receive points
for acting aggressively; attempt and learn various aggressive roles, actions,
and strategies; and through repeated play may learn to play the game quite
skillfully. Through this exposure to the stimulus, repeated interaction with
the violence, and repeated game play, it is likely that players can establish
similarly aggressive knowledge structures or scripts, enhance them through
practice, and activate them when faced with a real, albeit tamer potentially
aggressive situation. Thus, playing violent games, both in the short term
and over time, can lead to aggressive behavior.
GAM focuses on the person in the situation, or the ‘‘episode.’’ A single
social interaction, or episode, contains the person and situation inputs (such
as environmental cues); cognitive, affective, and arousal routes through
which these input variables are interpreted and have their impact; and out-
comes of the underlying appraisal and decision processes. In sum, any given
situation includes personalogical variables such as aggressive disposition;
the cognitive, affective, or arousal factors of the person in the situation;
and the decision to act. Specifically, then, GAM proposes that aggression
results from the person in the situation; however, the interpretation of that
situation and the ultimate outcome of that situation is mediated by cognitive
appraisal, affective response, and arousal resulting from the situation.
In the short term, video game violence is likely to affect aggression
because it may temporarily influence cognitive responses, affective out-
comes, and arousal, making the decision to behave aggressively, more likely.
In the long term, each exposure to violent stimuli (whether game violence or
aggressive interactions) is like another learning trial. Over time, these
aggressive scripts that are activated become more ingrained and more read-
ily accessible. These scripts then color the person’s expectations and inten-
tions involving social behavior and their perceptions of the actions of
others in their environment. The creation of these aggressive knowledge
structures can change the individual’s personality, making them more
aggressive and hostile in outlook (Anderson & Dill, 2000).
Arousal Routes to Aggression
Although other theories (e.g., Excitation Transfer; Zillmann, 1983) have
emphasized the importance of arousal as a mechanism in aggressive beha-
vior, GAM cites arousal as only one of three primary routes. Whether
118 KRCMAR AND FARRAR
because of a particular arousing situation or a repeated tendency of an
individual to respond with arousal to a frustrating or upsetting situation,
aggression itself is mediated by it. According to GAM, with heightened
arousal, the likelihood to respond with aggression is increased.
Cognitive Routes to Aggression
The GAM would predict that the effects of violent video games can be
mediated in a second way: cognitive responses. Research on television
violence has found that merely seeing a picture of a gun or a weapon can
prime aggressive thoughts (e.g., Anderson, Benjamin, & Bartholow, 1998).
Furthermore, research on video games suggests that exposure to violent
video games can increase the accessibility of aggressive thoughts (Anderson
& Bushman, 2001). The GAM argues that these aggressive cognitions or
thoughts are the mechanism by which exposure to media violence impacts
aggressive outcomes and, therefore, cognition mediates aggression. Therefore,
for our study, the following is hypothesized:
H1: Participants who play a violent video game will become more aggressive
compared to those in the control group; however, this effect will be
mediated by aggressive cognitions.
Despite the findings that video game violence can increase aggression,
and theory that predicts that cognition can mediate aggressive outcomes,
it is unclear how various game features (such as POV) may affect outcomes.
In fact, little published research has manipulated any of the contextual
features commonly found in today’s games to assess their impact on aggres-
sive outcomes. Why might game features make a difference? Consider the
research on television violence. Overall, the context of a television portrayal
of violence (e.g., showing pain cues, or rewarding the perpetrator) plays a
very important part in the effects process (Comstock & Paik, 1991; Gunter,
1994; Wilson, Linz, & Randall, 1990). Historically, most video games were
designed so that the player viewed the environment from the more tradi-
tional third person POV (the player sees their entire character on the screen).
More recently, however, games designed in the first person POV (the player
sees the environment as if looking through the eyes of the character) have
become increasingly popular, particularly a genre of violent games known
as first-person shooters (e.g., Doom, Halo). In fact, recent video game sales
data places Bioshock, a first-person shooter game, in the top three bestsel-
lers for video games in August 2007 (‘‘‘Madden’ Fuels,’’ 2007), and the
release of Halo 3, another first-person shooter, led to sales of more than
$300 million in just over 1 week (Rooney, 2007). Also, in many video games,
VIDEO GAMES 119
players can choose to view the environment from the more traditional
third-person POV or can select the first-person POV. Clearly, many of
today’s gamers are exposed to a first-person mode of play. It is possible that
players identify more strongly with an aggressive character and feel more
involved in the action when they are playing in first-person mode (Tambor-
ini et al., 2004). However, Farrar et al. (2006) found that playing in third
person, not in first person, tended to increase a sense of involvement. There-
fore, it remains somewhat unclear how POV can affect involvement in video
Furthermore, Farrar et al. (2006) found that there was no main effect for
POV on aggressive outcomes. However, in their study, players were not
provoked after playing the game. Previous research on television violence
has found that provocation coupled with exposure to violent media, rather
than either variable alone, is more likely to result in aggressive outcomes
(Zillmann, 1983). Therefore, it is still unclear how POV might affect aggres-
sive thoughts and the subsequent aggressive outcomes associated with
In addition to switching the POV, many modern games allow the user to
deactivate the blood and gore present in a violent game or to change the color
of the blood from red to green, for example. This is a feature that is very
appealing to parents who are concerned with the level of gore in the games
their kids play. However, traditional research based on television and film
stimuli would suggest that the presence of blood would serve to lessen the
probability that the player would become more aggressive as a result of expo-
sure (Wilson et al., 1997). On the other hand, it may be the case that in a
video game environment, blood could serve as a reward, showing the player
that she or he did well, or succeeded in achieving a goal. Blood may serve as
reinforcement, thus strengthening the risk of aggressive thoughts and conse-
quently, aggressive behavior on the part of the player (Smith, Lachlan, &
Tamborini, 2003). It is also possible that the presence of blood in repeated
aggressive encounters could serve to desensitize the player and thus lead to
increased aggression. In fact, in Farrar et al.’s (2006) study, those who played
the game in the blood on condition had more physically aggressive inten-
tions. Ballard and Wiest (1996) found that participants were significantly
more hostile and exhibited more systolic blood pressure reactivity after play-
ing Mortal Kombat with the blood and gore present compared to those par-
ticipants who played the same game without the gore. These findings lend
initial support to the notion that blood in a video game environment may
operate differently from blood in a television environment.
Therefore, contextual features of the game itself may well be linked to the
outcome of game. However, sex of the player has also been demonstrated to
play a role in video game play. Specifically, although previous research has
120 KRCMAR AND FARRAR
not consistently found any gender differences in aggression after playing
violent video games (see the meta-analysis by Anderson & Bushman,
2001), differences have been found in the way that men and women play
video games. For example, playing in first person appears to reduce a sense
of presence and involvement for women but not for men (Farrar et al., 2006;
Tamborini et al., 2001). Therefore, given that POV is an important variable
in this study and that men and women may, in fact, have very different
mental models for video games (Farrar et al., 2006), gender differences
are examined in this research.
In the GAM, aggressive cognitions are thought to mediate aggressive
outcomes. However, when the focus is on asking what outcomes might
be expected under various game features, the nature of the relationship
differs. It seems unlikely that the effect of POV or blood on aggression,
for example, would be mediated by aggressive cognitions. After all, aggres-
sion is present in both conditions (blood on vs. off and first vs. third
person). Instead, it seems more likely that when the blood cue acts as a
reinforcement, or when POV influences aggression, cognitions would mod-
erate the relationship. That is, if a cue generates violent thoughts, aggression
may occur. In the absence of violent thoughts, no aggression would occur.
Although the GAM stresses the importance of mediating factors between
exposure and aggression, it is likely that moderators are important as well.
Game features that increase the likelihood of aggressive cognitions, for
example, by definition increase the likelihood of that cognitive function ulti-
mately acting as a mediator. According to GAM, the mediator is a necessary
condition for aggression. Therefore, our first research question asks the
RQ1: Will the impact of the (a) blood and gore manipulation and (b) POV mani-
pulation on participants’ aggression be moderated by aggressive cognition?
Affective Routes to Aggression
The second route through which violent video games can influence aggres-
sion according to the GAM is affective. Playing a violent video game can
increase feelings of hostility in some situations (Anderson & Bushman,
2001). According to the GAM, these feelings of hostility might also lead
to aggressive outcomes. Specifically, the experience of hostility as a result
of game play is another mechanism by which aggressive outcomes occur.
Therefore, our second hypothesis states the following:
H2: Hostile affect will mediate the effect of the violent video game on
VIDEO GAMES 121
It is also not known how different contextual features of video games will
impact player’s levels of aggressive affect and how this will, in turn, impact
aggression. Similar to the argument made earlier, although GAM predicts
that cognition and affect mediate the relationship between media exposure
and aggression, it is unclear how various game features affect aggressive
outcomes. It seems likely that if a particular game feature (e.g., first-person
POV) does result in hostile affect, then aggressive outcomes may occur.
Once again, it is important to understand the moderating role of game
features to ultimately predict aggressive outcomes. Therefore, the following
research question is posed:
RQ2: Will the impact of the (a) blood and gore manipulation and (b) POV
manipulation on participants’ aggression be moderated by hostile
This study uses a 2 (violent game vs. no game) 2 (gender) 2 (first vs.
third person) 2 (blood on vs. blood off) nested experimental design. Parti-
cipants were first randomly assigned to play a violent game or to a no game
control group. Those in the violent game condition were further randomly
assigned to play the game for 12 min in first or third person and with the
blood feature on or off. Finally, all participants were insulted by an experi-
mental confederate prior to responding to a questionnaire.
Participants in this study were 186 undergraduate students (97 female, 89
male) enrolled in an introductory communication course at a large East
Coast university. Although it is possible that communication students
may have some exposure to the subject of video games and violence in their
communication courses, these data were collected at the very beginning of
the semester long before any lectures on media effects, and as this course
is a prerequisite for upper division courses in the major the students would
not have taken any other courses on this topic. In addition, the vast majority
of students in this course were taking it to meet general education require-
ments and were not even communication majors. Students received course
credit for participating. Participants mean age was 19.67 (range ¼18–22).
The racial composition of the sample was 88% White, 2% Asian, 4% African
American, 3% Latino=a, and 3% Other.
122 KRCMAR AND FARRAR
The violent game used for this study was Hitman II, Silent Assassin. This
game was selected for its violent content as well as the ability to manipulate
the two internal game features of interest (POV, third vs. first person, and
blood on or off). This game received an ‘‘M’’ rating for Mature according
to the Entertainment Software Rating Board. Content descriptors for
Hitman include blood, sexual content, and violence.
For the purposes of this research, violence is defined using the operational
definition from the National Television Violence Study: ‘‘Any overt depiction
of a credible threat of physical force or the actual use of such force intended
to physically harm an animate being or group of beings’’ (Smith et al., 1998,
p. 30). The purpose of Hitman II is to maneuver a hitman through several
missions, assassinating enemies, while attempting to rescue a friend. This
clearly meets the definition of violence for the purposes of this study.
Male and female undergraduates were randomly assigned to the no game
control group or to one of the four violent game conditions. Prior to playing
the assigned game, those in the game conditions reviewed an instruction
sheet detailing how to play and were instructed to play the game for
12 min. Games were played by each individual in a separate cubicle so that
players could not see each other. Participants were also instructed to wear
headphones during game play so that players could not hear anything other
than their own game. All games were played on a Sony PlayStation II
gaming console hooked up to a 13-in. color television monitor.
Experimental conditions were created by randomly assigning players to
the violent game or no game control group. The violent game players were
further assigned to one of four violent game conditions, varying their per-
spective (first-person POV vs. third-person POV) and blood (present or
absent). In the first-person POV, the player literally looks through the eyes
of the character experiencing the game environment from the hitman’s per-
spective. He or she sees nothing of their character other than his arm
extended outward clutching a gun. In contrast, in the third-person POV
the game player sees the entire body of the character on the screen, as if
he or she were standing a few feet behind the hitman character. To ensure
that all participants playing the violent game would have the same experi-
ence regardless of their skill level, game play was set to ‘‘god mode’’ so that
players could not be killed during play. Participants also filled out a posttest
with several measures regarding the game they just played and an additional
‘‘current events’’ questionnaire (designed with very difficult current events
questions with the assumption that most students would answer at least
VIDEO GAMES 123
several of them incorrectly), immediately after playing. Those in the no
game condition were asked to simply fill out the instrument with the game
related items removed.
After filling out the questionnaire, an experimental confederate, whom
they had not yet encountered, immediately led the participants (one by
one) to another room down the hall. The student was a senior-level drama
student to capitalize on her acting experience and to help ensure that none
of the study participants were likely to be familiar with her. They were told
that they had ‘‘one last thing to do before we’re all done.’’ En route, the con-
federate casually glanced at their current events answers and insulted their
performance by commenting that their performance on the task was really
terrible compared to other students. After this, the confederate did not
engage in any discussion with the participants until, upon reaching their des-
tination, the confederate told participants that they would be filling out a
form that evaluated her as well as the research assistant who helped them
with the video game. She concluded by telling participants that their answers
would remain confidential but would help determine if she would get fund-
ing in the following year. This was done to provoke participants, a protocol
used in many studies (e.g., Berkowitz & Powers, 1979) to provide them with
a target for any aggression that might result from the stimulus of interest
(e.g., video game play). However, because all participants are provoked in
the same way, any resulting differences in aggression are likely due to the
manipulation and not to this constant. Once this was done, the confederate
returned after several minutes, handed participants a debriefing form, and
thanked them for their participation.
Video game ratings. After playing the violent video game, participants
completed a nine-item scale rating how easy the game was to play, how
enjoyable the game was, how frustrating they found it to play, how violent
the game content was, how violent the graphics were, how much blood and
gore was present in the game, whether the game was fast or slow in pace,
how engaging the game was, and how well it maintained their attention.
Demographic variables. Participants indicated their gender, age, year
in school, and race.
Video game exposure. Participants were asked to first indicate whether
they had ever played the video game (17% of the participants had played
Hitman prior to exposure). It is possible that participants with prior
124 KRCMAR AND FARRAR
exposure to the game would react differently to the stimulus. However, ran-
dom assignment to conditions should alleviate this concern. Next, they were
asked to indicate how often they play a range of different types of video
games: action, adventure, three-dimensional shooter games, arcade type
games, role-playing or interactive fantasy games, simulation or strategic
planning games, sports games, or massive multiplayer online role-playing
games. Each game type contains an example game title from that category
and participants indicated on a 7-point scale from never to frequently how
often they played each type of game. These reliably (a¼.85) measured fre-
quency of game play and were therefore averaged (M¼2.56, SD ¼1.91).
Dependent variables. Aggressive behavioral intentions and aggressive
behavior were measured in two different ways. First, aggressive behavioral
intentions were measured by using a modified version of the Buss–Perry
aggression questionnaire (Buss & Perry, 1992). As originally conceived this
measure taps stable, trait aggression. However, because using a trait, rather
than a state, measure of aggression attenuates the effects of short-term expo-
sure to video game violence (Farrar & Krcmar, 2006), a slightly reworded
version of the scale reflecting state rather than trait aggression was utilized.
This instrument has been validated in previous research (see Farrar &
Krcmar, 2006). This construct corresponds to the well-known construct of
behavioral intention in the attitude literature (Ajzen & Fishbein, 1980).
The revised measure is identical to the original with two minor modifica-
tions. First, before responding to the aggression items, participants read an
ambiguous story that states, ‘‘Imagine that you leave this building when
you’re done completing this survey. Someone bumps into you, spilling your
drink and the contents of your backpack. They then step towards you.’’ Sec-
ond, participants respond to the Buss–Perry items (Buss & Perry, 1992),
however, instead of being worded as traits (e.g., ‘‘Given enough provoca-
tion, I would hit someone’’) they are worded to reflect state aggression
(e.g., ‘‘Given enough provocation, I would hit this person’’). Responses
range from 0 (extremely uncharacteristic of me)to6(extremely characteristic
of me). The four dimensions of the scale: physically aggressive intentions
(M¼2.69, SD ¼1.41), verbally aggressive intentions (M¼2.69, SD ¼1.61),
temper (M¼2.01, SD ¼1.47), and resentment (M¼2.72, SD ¼1.44) were
each reliable (a>.80). An examination of the face validity of the items led
us to use physically aggressive intentions and verbal aggression as indicators
of behavioral aggression.
Behavioral aggression was measured through participants’ rating of a
confederate research assistant. Participants in fact rated the two research
assistants, one who was an actual, nonconfederate assistant and one who
was also an experimental confederate. Comparison between the two ratings
VIDEO GAMES 125
was used as a manipulation check of the provocation by the confederate.
The confederate insulted the participants during the study because previous
research has found that aggressive outcomes are likely when media violence
is present and when the participant has a target against whom to aggress
(Zillmann, 1983). The scale was made up of two types of questions. The
first five questions asked participants to rate each assistant’s efficiency,
organization, and kindness on a 10-point scale (a¼.93). The insulting con-
federate was rated generally lower (M¼7.62, SD ¼2.02) than the noncon-
federate assistant (M¼8.92, SD ¼1.51), indicating that the manipulation
was effective (t¼5.63, n¼220, p<.05). The next two questions were asked
to give participants a chance to aggress against the confederate. Specifically,
they were asked to judge if the assistant should be funded in the following
year and if she should be given full-time funding. Responses ranged from
0(definitely not)to10(definitely yes). These two items were strongly corre-
lated (r¼.95) so they were collapsed into a single score with higher numbers
indicating higher funding recommendations and therefore less harm.
Moderators=mediators. Accessibility of aggressive cognitions was
measured using the ‘‘speed-of-association test’’ which has been previously
validated as a measure of aggressive cognitions (Bushman, 1998). Partici-
pants are told they are completing a test to see how quickly they can think.
The sheet of paper has 50 words, each with a blank space next to it. In the
space after each word, participants are told to write—as quickly as they
can—the first word or phrase that comes to mind. In total, 25 of the words
were homonyms with a potentially aggressive interpretation (such as boot).
These could elicit either neutral words (i.e., shoe) or aggressive terms (i.e.,
kick). The remaining 25 were neutral words with no aggressive meaning
(e.g., note). Upon completion, each participant had generated 50 words.
Responses were then coded as an aggressive word (e.g., hit) or a nonaggres-
sive one. Participants with more readily accessible aggressive thoughts are
expected to list more aggressive terms.
Aggressive affect was measured using the 15-item State Hostility scale
(Anderson, Deuser, & DeNeve, 1995). This 4-point scale ranges from not
at all to very much so. The scale proved reliable (a¼.91) and included items
such as ‘‘I feel angry,’’ ‘‘I am burned up,’’ ‘‘I feel aggravated,’’ and ‘‘I am
annoyed.’’ The overall mean for this scale was 1.25 (SD ¼.35).
By utilizing the GAM, two different kinds of relationships would be
expected. First, the effect of the violent game itself should be mediated by
126 KRCMAR AND FARRAR
both cognitive and affective processes. Second, it is expected that any effects
of the internal game features (such as point of view) would be moderated by
aggressive cognitions and affective hostility because one game feature (such
as first person) may result in increased hostility and subsequent aggression,
whereas another feature (such as third person) may not increase hostility
and subsequent aggression. Therefore, in our analyses, when examining
the effect of the violent game versus no game, all participants were included
in the analyses (N¼186) and tested for aggressive effects that were mediated
by cognitive and affective processes using strategies developed by Baron and
Kenny (1986). When looking further at the impact of the two manipulations
(blood and POV), only those participants in the violent game conditions
were included (n¼148) and tested for aggressive effects that were moderated
by cognitive and affective processes. To test moderation four product terms
were created. Each product term was made up of one manipulation (i.e.,
POV or blood) multiplied by one moderator (i.e., aggressive cognition or
hostile affect). For each analysis, the relevant product term was used (e.g.,
the POV manipulation times hostile affect) as the independent variable. Ver-
bal and physically aggressive intentions, as well as funding recommenda-
tions for the insulting research assistant confederate were tested as the
dependent variables in each case. Individual group means in these analyses
were tested using post hoc procedures because of the lack of prior research
that would help us form concrete hypotheses about the nature of specific
Cognitive Routes to Aggression
H1. The first hypothesis predicted that those in the violent game
condition would be more aggressive than those assigned to the control
group and that this effect would be mediated by aggressive cognition. To
understand the main effect of condition on verbal and physically aggressive
intentions and on ratings of the confederate, an analysis of variance was
utilized. There was a main effect of condition on verbally aggressive inten-
tions, F(1, 185) ¼10.72, p<.05. Those in the violent game condition were
significantly more verbally aggressive than those in the no game condition.
There was also a main effect for condition on physically aggressive inten-
tions, F(1, 185) ¼10.96, p<.05. Those in the violent game condition were
significantly more physically aggressive than those in the no game condition.
Last, there was a main effect on funding recommendations for the confed-
erate, F(1, 184) ¼19.92, p<.05. Those in the no game condition gave the
confederate significantly lower funding scores than those in the violent game
condition (see Table 1).
VIDEO GAMES 127
Next, possible mediation effects were examined. To do this, only those in
the violent game condition and those in the no game condition were
included in the analyses (n¼186). Consistent with multiple regression pro-
cedures for testing mediation, verbally aggressive intentions was first used
as the criterion and game play (violent vs. no game) as the predictor.
This regression was significant (R¼.24), F(1, 183) ¼10.72, p<.05. Next,
a second regression was run, using aggressive cognitions (the mediator) as
the criterion variable and game play as the predictor. This regression was
not significant (R¼.12), F(1, 183) ¼2.34, p>.05. Therefore, aggressive cog-
nitions do not mediate the relationship between game play and verbally
To test if aggressive cognition mediates the link between game play and
physically aggressive intentions, the same procedure was used to test for
mediation. First, physically aggressive intention was used as the criterion and
game play as the predictor. This regression was significant (R¼.24), F(1,
184) ¼10.96, p<.05. Next, using aggressive cognition (the mediator) as
the criterion and game play as the predictor, the equation was not signifi-
cant. Therefore, H1 is not supported because aggressive cognition does
not mediate the relationship between game play and verbal or physically
aggressive intentions. Rather, there is a direct effect of game play on the
RQ1a and 1b. To address RQ1, which asked if the effect of (a) the POV
manipulation and (b) the gore manipulation on verbally aggressive intentions,
physically aggressive intentions, and the confederate’s funding score was
moderated by aggressive cognitions, only those in the violent game conditions
were examined. Gender was controlled for on the first step, on the second step
each of the individual variables of interest was entered. On the third step the
product terms of those variables of interest were entered (i.e., POV multiplied
by aggressive cognition; gore multiplied by aggressive cognition; Aiken &
Main Effect of Game Condition on Verbal and Physical Aggression and on Retaliation
Verbal aggression Physical aggression Retaliation
M SD M SD M SD
Violent game 2.56 1.39 2.59 1.55 6.30 2.60
No game 1.79 0.81 1.72 0.93 7.98 1.91
Note.N¼217. Higher numbers indicate more aggression and are significant at p<.05 for
differences between the game and no game conditions. For retaliation, higher numbers indicate
a more positive evaluation.
128 KRCMAR AND FARRAR
West, 1991). Three regressions were run for the (a) POV interaction: one for
verbally aggressive intentions, one for physically aggressive intentions, and
one for funding recommendations. In addition, three regressions were run
for the gore manipulation. In total, six analyses were run to test RQ1.
To test RQ1a, asking about the possible moderating effect of POV, the
first POV regression was run using verbally aggressive intentions as the
dependent variable. This revealed that there was no effect of gender on
the first step. On the second step, there was no significant change in R
for the POV manipulation or the aggressive cognition variable. On the third
step, the interaction between POV and aggressive cognition did have a sig-
nificant effect on verbally aggressive intentions (see Table 2). Those who
played the violent game in third person and who had more violent cogni-
tions were significantly more verbally aggressive than those in the other
three conditions (see Table 3 for means and standard deviation).
The second POV regression was run using physically aggressive intentions
as the dependent variable. There was an effect of gender on the first step. On
the second step, there was no significant change in R
for the POV manipula-
tion or the aggressive cognition variable. On the third step, the interaction
between POV and aggressive cognition did have a significant effect on physi-
cally aggressive intentions(see Table 2). Post hoc contrasts revealed that those
who played the violent game in third person and who had more violent cogni-
tions had significantly more physically aggressive intentions than those in the
other three conditions. (See Table 3 for means and standard deviations.)
Interaction Effects for Aggressive Cognitions and Game Features (POV and Blood) on
Verbal, Physical, and Retaliatory Aggression
Verbal agg. Physical agg. Retaliation
POV Blood POV Blood POV Blood
Step 1 (Controls)
Gender (0 ¼male; 1 ¼female) .00 .00 .126.96.36.199
Step 2 (Main effects)
POV, aggressive cognition .01 — .00 — .00 —
Blood, aggressive cognition — .01 — .02 — .01
Step 3 (interactions)
Interaction between POV and cognition or .05— .02 .05—
Interaction between blood and cognition .04.01 — .02
Note.N¼186. Each column represents a single regression analysis, which indicates that the
variables were not used in this equation. POV ¼point of view; agg ¼aggression.
Significant (p<.05) change at that step.
VIDEO GAMES 129
The third POV regression was run using the confederate’s funding
recommendation as the dependent variable. There was an effect of gender
on Step 1. On Step 2, there was no significant change in R
for the POV manip-
ulation or the aggressive cognition variable. On Step 3, the interaction between
POV and aggressive cognition had a significant effect on funding recommen-
dations for the confederate. Those who played in third person and who experi-
enced fewer aggressive cognitions provided the confederate with significantly
higher ratings than those in the other three conditions. (See Table 3 for means
and standard deviation.) Therefore, significantly less aggression occurred
among those players in third person with fewer aggressive cognitions.
To test RQ1b, we tested the possible moderating effect of the presence of
blood and gore. For verbally aggressive intentions, there was no effect of gen-
der on step one. On Step 2, there was no significant main effect either the
blood manipulation or for aggressive cognitions. On Step 3, the change in
was significant (see Table 2). That is, there was a significant interaction
between the blood manipulation and aggressive cognitions. Post hoc con-
trasts showed that those in the blood on condition with higher aggressive
cognitions were more verbally aggressive (M¼2.81, SD ¼1.41) than those
with the blood on and low aggressive cognitions (M¼2.28, SD ¼1.52), those
with blood off and high aggressive cognitions (M¼2.50, SD ¼1.36) and
those with blood off and low aggressive cognitions (M¼2.45, SD ¼1.27).
With physically aggressive intentions as the dependent variable, there
was an effect for gender (see Table 2) but no main effects on Step 2 or inter-
action effects on Step 3 for the blood manipulation and aggressive cogni-
tions. For recommendations for the confederate there was an effect for
gender but no main or interaction effects for the blood manipulation and
aggressive cognitions. Therefore, for verbally aggressive intentions and phy-
sically aggressive intentions, aggressive cognitions moderate the effect of the
Interaction Effect of Aggressive Cognitions and Game Features
(First vs. Third Person) on Aggression
Verbal aggression Physical aggression Retaliation
Fewer violent cognitions 2.58 (1.36)
More violent cognitions 2.36 (1.326)
Fewer violent cognitions 2.29 (1.43)
More violent cognitions 3.19 (1.26)
Note.N¼186. Higher numbers for the confederate indicate higher recommendations.
Range ¼1–10. Means with different subscripts significantly different at p<.05.
130 KRCMAR AND FARRAR
blood manipulation. In this case, those who experienced aggressive
cognitions after playing in the blood on condition were most verbally
Affective Routes to Aggression
H2. H2 predicted that hostile affect would mediate the effect of the
violent video game on aggressive outcomes. To test this, only those in the
violent game condition and those in the no game condition were analyzed
(n¼186). Consistent with multiple regression procedures for testing media-
tion, verbally aggressive intentions was used as the criterion and game play
(violent vs. no game) as the predictor. This regression was significant
(R¼.24), F(1, 183) ¼10.72, p<.05. The second regression, using hostile
affect (the mediator) as the criterion variable and game play as the predictor
was also significant (R¼.20), F(1, 183) ¼7.28, p>.05.Third, with hostile
affect as the criterion and verbally aggressive intentions as the outcome, this
regression was not significant (R¼.01), F(1, 183) ¼13, p>.05. Therefore, in
this study hostility does not mediate the link between violent game play and
verbally aggressive intentions.
To test the effect on physically aggressive intentions, physically aggressive
intentions was used as the criterion and game play (violent vs. no game) as
the predictor. This regression was significant (R¼.23), F(1, 183) ¼10.97,
p<.05. Next, using hostile affect (the mediator) as the criterion variable and
game play as the predictor, this regression was also significant (R¼.20), F(1,
183) ¼7.28, p<.05. Next, we examined if the mediator affects the outcome.
This regression was not significant. Therefore, in our study, hostility does not
mediatethe link between violent game play and physically aggressive intentions.
Last, funding for the confederate was used as the dependent variable. The
first regression was significant (R¼.18), F(1, 183) ¼6.37, p<.05. The second
regression, using hostility, demonstrated that it also affects funding recom-
mendations (R¼.19), F(1, 183) ¼6.43, p<.05. Last, the mediator was con-
trolled for the effect of game play on funding recommendations was tested.
This regression was significant at the final step (R¼.24), F(3, 180) ¼4.22,
p<.05, and the beta for game play was still significant (B¼–.15, p<.05) after
controlling for hostility. When hostility, the potential mediator, was not used
as a control variable, the effect of game play was somewhat larger (B¼–.19,
p<.05). Therefore, hostility partially mediates the effect of game play on the
funding recommendations. H2 is partially supported.
RQ2. RQ2 asked if the effects of POV manipulation and the blood
and gore manipulation on aggression would be moderated by hostile affect.
To test this only those in the violent game conditions were included in the
VIDEO GAMES 131
analysis and we controlled for gender on the first step. On Step 2 each of the
individual variables of interest was entered. On the third step the product
term of the two variables of interest was used (Aiken & West, 1991). Three
regressions were run for each of the internal game manipulations (POV and
gore), one for verbally aggressive intentions, one for physically aggressive
intentions, and one for funding recommendations. First, the results for
POV are presented followed by the results for the blood=gore manipulation.
For POV with verbally aggressive intentions as the dependent variable,
there was no effect for gender. On Step 2, there was no effect of either hos-
tility or POV on verbally aggressive intentions, and there was not an inter-
action between hostility and POV. Therefore, hostility does not moderate
the effect of POV on verbally aggressive intentions. With physically aggres-
sive intentions as the dependent variable, gender had a significant effect
(R¼.42), F(1, 145) ¼31.36, p<.05, but neither the main effects nor the
interaction effects were significant. Last, the confederate recommendations
were examined. Again, there was a main effect for gender (R¼.27), F(1,
145) ¼11.41, p<.05, but there were no main or interaction effects for
hostility and POV.
Next, the effect of the blood manipulation was examined. With verbally
aggressive intentions as the dependent variable, there was no main effect for
gender, there was no main effect for hostility or blood, but there was a signifi-
cant interaction (R¼.21), F(1, 142) ¼3.92, p<.05. Post hoc contrasts
revealed that those who were high in hostility and played the game with the
blood feature turned on (M¼3.38, SD ¼1.35) were significantly more verb-
ally aggressive than those high in hostility with no blood (M¼2.88,
SD ¼1.59) or low in hostility with blood on (M¼2.82, SD ¼1.29). Those
who were low in hostility and played the game with no blood or gore
(M¼2.34, SD ¼1.45) were significantly less verbally aggressive than those
in the other groups. For physically aggressive intentions, gender had a signifi-
cant effect on Step 1, Steps 2 and 3 were not significant. For funding recom-
mendations, there was a main effect for gender but no main or interaction
effects. Overall, for the blood manipulation, only verbally aggressive intention
was affected by the manipulation and that effect was moderated by hostility.
Summary of Findings
Overall, those in the violent game condition were more verbally and physi-
cally aggressive than those in the no game condition. In addition, aggressive
cognitions and affective hostility did not mediate the effects of game play as
predicted by the GAM.
132 KRCMAR AND FARRAR
In terms of the internal game manipulations, those who played the
violent game in third person and who had more violent cognitions had
significantly more verbally aggressive intentions, had more physically
aggressive intentions and, to some extent, were more punitive toward the
confederate than other participants. Furthermore, those in the blood=gore
on condition with more aggressive cognitions were more verbally aggressive.
Therefore, aggressive cognitions do moderate the effects of internal game
manipulations. For hostile affect, those who were high in hostility and
played the game with blood on were significantly more aggressive than other
players. Therefore, hostility has some moderating effect but only for the
blood manipulation, whereas aggressive cognitions moderated the effect
of both manipulations. Overall, it can be said that aggressive cognitions
had a more consistent moderating effect than aggressive affect. In addition,
third-person play with the blood on, especially when combined with aggres-
sive cognitions and to a lesser extent, hostile affect, encouraged the more
Role of Internal Game Features
One increasingly prevalent game feature is the ability to play in the more
traditional third-person point of view or to play in first person. It has been
suggested that playing in the first person, which involves looking through
the characters’ eyes as if they were your own, could enhance identification
with the game character. Theoretically, increased identification should lead
to increased aggression. This study, however, found the most aggressive out-
comes occurred in third-person play. Although this finding may seem some-
what counterintuitive, it is consistent with the work of Farrar et al. (2006). It
is possible that to identify with the video game character, it is necessary to see
that character. Recall that in first-person play, gamers see an outstretched
arm but no character. Perhaps, rather than encouraging identification, first
person feels more akin to ‘‘no-person’’ play. Counter to their expectations,
Eastin and Griffiths (2006) found that participants experienced more pre-
sence when playing standard console video games as opposed to virtual rea-
lity. They suggested that comfort level with the interface may have played a
role in the level of presence experienced. Therefore, to further illuminate our
findings, future research might pit experienced players against novice players
to see if perhaps first person play becomes more comfortable, and thus
engaging, as players gain experience in the environment.
As previously mentioned, another common game feature is the ability to
turn ‘‘off’’ the blood and gore that may result from violence in the game.
Unlike television research where the presence of blood lessens the likelihood
of imitation, this study found that the presence of blood within the game
VIDEO GAMES 133
increased verbally aggressive intentions when it was accompanied by an
increase in aggressive cognitions. This offers some support for GAM, which
argues that it is through aggressive cognitions, initially generated by violent
depictions, that aggressive outcomes occur. In a video game environment,
blood seems to act as a reward. These results are consistent with both past
television research and with social cognitive theory.
This study offers partial support for the generalized affective model of
aggression and those theories (e.g., cognitive priming and, to a lesser extent,
the affective aggression model) that are incorporated into it. First, GAM
was supported by the findings that suggested that cognitions moderate the
effects of the internal game features. Recall that GAM proposes that there
are two routes to aggression—affective and cognitive—and that these two
routes can work either separately or in consort to increase aggression. How-
ever, in our study, cognition moderated outcomes more consistently than
affect did. Why might this be the case? Consider the findings of a recent
meta-analysis conducted by Sherry (2007). Sherry found that aggressive out-
comes resulting from video game play seemed to occur after short-term play
(e.g., 10 min) but not after longer term play. In our study, we also found
negative effects after short-term play, especially as they related to cognition.
It is possible that cognitive priming occurs in the short term but that the
effects wear off relatively quickly. If this is the case, then short-term video
game play and experimental designs that test for immediate outcomes would
favor cognitive explanations such as those offered by priming theories and
GAM, which incorporates notions of priming. Longer term effects, such
as affective hostility, may take longer to occur and may in fact have longer
term effects, making a research design such as the one used in this study,
unable to detect them. Therefore, our design supports cognitive aspects of
GAM to a greater extent because cognition and affect may have different
response times that can not both be detected at the same time.
Second, it is important to note that for the main game manipulation,
neither aggressive cognitions nor hostile affect were found to mediate
effects. Rather, the effects for the main manipulation were direct. Therefore,
it appears that the effects of game manipulations can not be explained by
GAM, necessarily. Instead, game features seem to effect outcomes in much
the same way that contextual features of television violence affect aggres-
sion: rewards (i.e., the appearance of blood in a video game) increase aggres-
sive outcomes and seeing the perpetrator (i.e., playing in third person in a
video game) increase aggressive outcomes. From a theoretical perspective,
social cognitive theory (Bandura, 1994) can explain these findings because
134 KRCMAR AND FARRAR
it argues that learning and imitation will most likely occur if (a) there is a
close identification between the observer and the model as might occur
in third-person play when the game character can be seen clearly, and (b)
if the character is rewarded for his or her behavior, as is the case when blood
the blood feature is activated in the game. Although aggressive cognitions
might occur when these features occur in the game, aggressive cognitions
may not be necessary for aggressive outcomes to ensue.
Ultimately, from the data presented here, it may be that GAM offers a solid
and broad-based explanation for aggressive outcomes. However,it may in fact
be too broad an explanation to deal with the details and intricacies of game
features. When Sherry (2007) stated that ‘‘theories designed to explain and
predict the social influences of television are not adequate to account for video
game effects’’ (p. 257), he may well offer the clearest explanations for the cur-
rent findings. If different theories (i.e., GAM, social cognitive theory and
priming) must be called upon to explain different types of game effects (e.g.,
main effects of game play, interaction effects for internal features) then clearly
we do not have a coherent and cogent theory. In fact, additional research must
be conducted to develop likely theoretical explanations that parsimoniously
explain all aspects and levels of game play effects or perhaps to incorporate
the media type (e.g., television, video game) into the theory.
Limitations and Future Research
Although the findings of this study offer insight into the role of internal
game features in the effects process, there are some limitations that should
be taken into consideration. In this study the presence of blood and POV
was manipulated, but feelings of reward in the former instance and identifi-
cation with the violent character in the latter were not measured. It is there-
fore not entirely clear why blood on and, more interesting, third person
encouraged aggressive outcomes. Although it is speculated here that third-
person play offers more opportunity to see, and thus identify with, the char-
acter, more research is needed to specify the mechanism at work here. As
previously mentioned, it is possible that identification with the character
occurs somewhat differently for players depending on their level of skill.
For example, third-person play may be easier and less confusing, and there-
fore more absorbing for novice players, whereas first person play feels more
realistic and natural to expert players. It may also be relevant that the main
character in the game was male. Recent research by Eastin (2006) suggests
that for female game players, matching gender of the main character
increases both presence and aggression. Therefore, aggressive outcomes
for female game players in our study may have been attenuated by the male
gender of the main game character.
VIDEO GAMES 135
Another possible limitation concerns the female gender of the confeder-
ate. Research suggests that research participants are more likely to aggress
against males than females (Eagly & Steffen, 1986). It is possible that parti-
cipants would have been more likely to aggress against a male confederate in
this situation, and this may have affected our results. Also, we found that
participants in the no game condition gave the confederate significantly
lower ratings than those in the violent condition. A very likely explanation
for this finding is that participants in the no game condition were irritated
that they did not get to play the game and thus took out their irritation
on the confederate as she seemed to be closely associated with the running
of the study.
Finally, it should be recognized that although setting the game to ‘‘god
mode’’ ensured that all players of varying skill levels would be able to play
for the allotted time without being killed, it may not have achieved the goal of
ensuring that all players experienced the game in the same way. It is possible
that this setting actually removed the game play element from the experience
and this may have affected participants’ cognitions toward the game. This issue
in particular may account for the relatively small effects sizes: Without the
external validity of real, competitive play with the possibility to lose the game,
effects may have been attenuated. In addition, this issue may partly explain the
somewhat low levels of hostility and aggression overall. Again, without a sense
of real competition, both aggression levels and effects sizes may have been smal-
ler than they might be if true competition had occurred.
Overall, future research should recognize that because of the unique char-
acteristics of video game play, direct parallels between television and video
games might not be possible. For example, blood as a result of seeing a film
character injured appears to operate differently from blood that occurs after
a successful ‘‘hit’’ in a video game. This finding has been similar in several
studies (see Farrar et al., 2006) and seems to suggest that blood in a video
game is a reward, and not a consequence, as it is in many television pro-
grams and films. Therefore, although an understanding of the overall effects
of game play in general continues to grow (Anderson & Bushman, 2001) as
video games become ever more sophisticated, with more game features and
options for play, more research is needed to understand how these features
function to affect aggressive outcomes.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions;
Thousand Oaks, CA: Sage.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior.
Englewood Cliffs, NJ: Prentice-Hall.
136 KRCMAR AND FARRAR
Anderson, C. A. (2004). An update on the effects of playing violent video games. Journal of
Anderson, C. A., Benjamin, A. J., & Bartholow, B. D. (1998). Does the gun pull the trigger?
Automatic priming effects of weapon pictures and weapon names. Psychological Science,
Anderson, C. A., & Bushman, B. J. (2001). Effects of violent video games on aggressive beha-
vior, aggressive cognition, aggressive affect, physiological arousal, and prosocial behavior: A
meta-analytic review of the scientific literature. Psychological Science,12(5), 353–359.
Anderson, C. A., Deuser, W. E., & DeNeve, K. (1995). Hot tempers, hostile affect, hostile
cognition and arousal: Tests of a general model of affective aggression. Personality and Social
Psychology Bulletin,21, 434–448.
Anderson, C. A., & Dill, K. E. (2000). Video games and aggressive thoughts, feelings, and
behavior in the laboratory and in life. Journal of Personality and Social Psychology,78(4),
Ballard, M. E., & Wiest, J. R. (1996). Mortal Kombat
: The effects of violent videogame play
on males’ hostility and cardiovascular responding. Journal of Applied Social Psychology,
Bandura, A. (1994). Social cognitive theory of mass communication. In J. Bryant & D.
Zillman (Eds.), Media effects: Advances in theory and research (pp. 61–90). Hillsdale, NJ:
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic and statistical considerations. Journal of
Personality and Social Psychology,51, 1173–1182.
Berkowitz, L. (1989). Frustration-aggression hypothesis: Examination and reformulation.
Psychological Bulletin,106, 59–73.
Berkowitz, L., & Powers, P. C. (1979). Effects of timing and justification of witnessed
aggression on observers’ punitiveness. Journal of Research in Personality,13, 71–80.
Bushman, B. J. (1998). Priming effects of media violence on the accessibility of aggressive
constructs in memory. Personality and Social Psychology Bulletin,24(5), 537–545.
Buss, A. H., & Perry, M. (1992). The Aggression Questionnaire. Journal of Personality and
Social Psychology,63, 452–459.
Carnagey, N. L., & Anderson, C. A. (2003). The role of theory in the study of media violence:
The general aggression model. In D. Gentile (Ed.), Media violence and children (pp. 87–106).
Westport, CT: Praeger.
Cicchirillo, V., & Chory-Assad, R. M. (2005). Effects of affective orientation and video game
play on aggressive thoughts and behaviors. Journal of Broadcasting & Electronic Media,
Comstock, G., & Paik, H. (1991). Television and the American child. New York: Academic
Eagly, A. H., & Steffen, V. (1986). Gender and helping behavior: A meta-analytic review of the
social psychological literature. Psychological Bulletin,100, 309–330.
Eastin, M. S. (2006). Video game violence and the female game player: Self-and opponent
gender effects on presence and aggressive thoughts. Human Communication Research,32,
Eastin, M. S. (2007). The influence of competitive and cooperative group game play on state
hostility. Human Communication Research,33, 450–466.
Eastin, M. S., & Griffiths, R. P. (2006). Beyond the shooter game. Examining presence and
hostile outcomes among male game players. Communication Research,33(6), 448–466.
Farrar, K. M., & Krcmar, M. (2006). Measuring state and trait aggression: A short, cautionary
tale. Media Psychology,8(2), 127–138.
VIDEO GAMES 137
Farrar, K., Krcmar, M., & Nowak, K. L. (2006). Contextual features of violent video games,
mental models and aggression. Journal of Communication,56(2), 387–405.
Geen, R. G. (1990). Human aggression. Pacific Grove, CA: Brooks=Cole.
Gunter, G. (1994). The question of media violence. In J. Bryant & D. Zillman (Eds.), Media
effects (pp. 163–211). Hillsdale, NJ: Erlbaum.
Huesmann, L. R. (1986). Psychological processes promoting the relation between exposure to
media violence and aggressive behavior by the viewer. Journal of Social Issues,42, 125–139.
Leyens, J. P., & Picus, S. (1973). Identification with the winner of a fight and name mediation:
Their differential effects upon subsequent aggressive behavior. British Journal of Social and
Clinical Psychology,12, 374–377.
‘Madden’ fuels video game sales in August. (2007, September 14). FOXNews Business and
Money [Online]. Retrieved from http://www.foxnews.com/story/0,2933,296821,00.html
Rooney, B. (2007, October 25). Microsoft’s ‘Halo’ effect. CNNMoney.com. Retrieved from
Sherry, J. L. (2001). The effects of violent video games on aggression. A meta-analysis. Human
Communication Research,27(2), 409–431.
Sherry, J. L. (2007). Violent video games and aggression: Why can’t we find effects? In R. W.
Preiss, B. M. Gayle, M. Burrell, M. Allen, & J. Bryant (Eds.), Mass media effects research:
Advances through meta-analysis (pp. 245–262). Mahwah, NJ: Erlbaum.
Smith, S.L., Lachlan, K., & Tamborini, R. (2003). Popular video games: Quantifying the
presentation of violence and its context. Journal of Broadcasting & Electronic Media,47(1),
Smith, S. L., Wilson, B. J., Kunkel, D., Linz, D., Potter, W. J., Colvin, C. M. et al, (1998).
Violence in television programming overall: University of California, Santa Barbara study.
In National Television Violence Study (Vol. 3, pp. 5–220). Newbury Park, CA: Sage.
Tamborini, R., Eastin, M. S., Lachlan, K., Skalski, P., Fediuk, T., & Brady, R. (2001, May).
Hostile thoughts, presence and violent virtual video games. Paper presented at the 51st annual
convention of the International Communication Association, Washington, DC.
Tamborini, R., Eastin, M. S., Skalski, P., Lachlan, K., Fediuk, T. A., & Brady, R. (2004).
Violent virtual video games and hostile thoughts. Journal of Broadcasting & Electronic Media,
Wilson, B. J., Linz, D., & Randall, B. (1990). Applying social science research to film ratings:
A shift from offensiveness to harmful effects. Journal of Broadcasting & Electronic Media,
Wilson, B. J., Kunkel, D., Linz, D., Potter, J., Donnerstein, E., Smith, S., Blumenthal, E., &
Gray, T. (1997). Television violence and its context. In Mediascope (Ed.), National Television
Violence Study (Vol. 1, pp. 7–10). Newbury Park, CA: Sage.
Zillmann, D. (1983). Arousal and aggression. In R. Geen & E. Donnerstein (Eds.), Aggression:
Theoretical and empirical reviews (pp. 75–101). New York: Academic.
138 KRCMAR AND FARRAR