Content uploaded by Bruce D Bartholow
Author content
All content in this area was uploaded by Bruce D Bartholow on Jul 28, 2022
Content may be subject to copyright.
Violent and nonviolent video games produce opposing effects on aggressive and
prosocial outcomes
Marc A. Sestir
a,
⁎, Bruce D. Bartholow
b
a
Hobart and William Smith Colleges, Geneva, NY, USA
b
University of Missouri-Columbia, Columbia, MO, USA
abstractarticle info
Article history:
Received 3 February 2010
Revised 13 May 2010
Available online 25 June 2010
Keywords:
Media effects
Aggression
Prosocial behavior
Experimental studies routinely show that participants who play a violent game are more aggressive
immediately following game play than participants who play a nonviolent game. The underlying assumption
is that nonviolent games have no effect on aggression, whereas violent games increase it. The current studies
demonstrate that, although violent game exposure increases aggression, nonviolent video game exposure
decreases aggressive thoughts and feelings (Exp 1) and aggressive behavior (Exp 2). When participants
assessed after a delay were compared to those measured immediately following game play, violent game
players showed decreased aggressive thoughts, feelings and behavior, whereas nonviolent game players
showed increases in these outcomes. Experiment 3 extended these findings by showing that exposure to
nonviolent puzzle-solving games with no expressly prosocial content increases prosocial thoughts, relative
to both violent game exposure and, on some measures, a no-game control condition. Implications of these
findings for models of media effects are discussed.
© 2010 Elsevier Inc. All rights reserved.
A major development in mass media over the last 25 years has been
the advent and rapid growth of the video game industry. From the
earliest arcade-based console games, video games have been immedi-
ately and immensely popular, particularly among young people and
their subsequent introduction to the home market only served to
further elevate their prevalence (Gentile, 2009). Given their popularity,
social scientists have been concerned with the potential effects of video
games on those who play them, focusing particularly on games with
violent content. While a large percentage of games have always
involved the destruction of enemies, recent advances in technology
have enabled games to become steadily more realistic. Coupled with an
increase in the number of adult players, these advances have enabled
the development of games involving more and more graphic violence.
Over the past several years, the majority of best-selling games have
involved frequent and explicit acts of violence as a central gameplay
theme (Smith, Lachlan, & Tamborini, 2003).
A video game is essentially a simulated experience. Virtually every
major theory of human aggression, including social learning theory
(Bandura, 1977), cognitive neoassociation theory (Berkowitz, 1990),
script theory (Huesmann, 1998), and the General Aggression Model
(GAM; Anderson & Bushman, 2002), predicts that repeated simulation
of antisocial behavior will produce an increase in antisocial behavior
(e.g., aggression) and a decrease in prosocial behavior (e.g., helping)
outside the simulated environment (i.e., in “real life”). In addition, an
increase in the perceived realism of thesimulation is posited to increase
the strength of negative effects (Gentile & Anderson, 2003). Meta-
analyses of existing empirical studies generally support these predic-
tions, in that exposure to violent video games, relative to nonviolent
game exposure, causes increases in aggressive behavior, cognition, and
affect, and decreases in prosocial behaviors and attitudes, (e.g.,
Anderson, 2004; Anderson & Bushman, 2001).
Although the effects of violent video game exposure on aggression
are generally accepted, a common criticism of experimental research on
this topic is that the negative effects of exposure to violent games are
likely short-lived (see Freedman, 2001). In the experimental literature,
researchers have uniformly measured aggression immediately orwithin
a few minutes after video game play. Thus, little is known about the
temporal stability ofacute exposure effects. Testing the duration of such
effects has important implications for understanding their nature. For
example,if violent game effectson aggression are due largely to priming
(e.g., Huesmann, 1998), the effects should decay rapidly in the absence
of an opportunity to aggress, as aggressive cognitions and behavioral
scripts recede from temporarily heightened accessibility (c.f., Higgins,
1996). In contrast, if exposure to a violent game creates a heightened
state of aggression-related goal activation (e.g., Bargh, Gollwitzer, Lee-
Chai, Barndollar & Troetschel, 2001; see also, Forster, Liberman &
Higgins, 2005), the goal should remain active, and the likelihood of
aggression higher, until an opportunity to aggress is presented,
presumably for a longer period of time than if effects were due primarily
to heightened accessibility of aggressive cognitions.
In addition to assessment of temporal stability, another important
consideration is the choice of control condition. To establish baseline
Journal of Experimental Social Psychology 46 (2010) 934–942
⁎Corresponding author.
E-mail address: Sestir@hws.edu (M.A. Sestir).
0022-1031/$ –see front matter © 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jesp.2010.06.005
Contents lists available at ScienceDirect
Journal of Experimental Social Psychology
journal homepage: www.elsevier.com/locate/jesp
levels of aggression, studies testing violent game effects typically use
nonviolent games as a control group (e.g. Anderson & Dill, 2000;
Carnagey, Anderson, & Bushman, 2007). Thus, any post-game increase in
aggression-related outcomes in the violent game condition is attributed
solely to the violent game. This strategy is potentially problematic for
several reasons, primarily that nonviolent games are likely not a “pure”
control. While efforts often are made to match games on aggression-
relevant factors such as arousal, frustration and interest (see Anderson &
Dill, 2000), nonviolent games still tend to differ from violent games in
qualities other than the absence of violent content. Thus, it is possible that
violent and nonviolent games simply have divergent effects on outcomes
of interest. In other words, while violent game exposure might increase
aggressiveness, nonviolent game exposure might also decrease aggres-
siveness below some pre-exposure baseline.
This could occur for a number of reasons. Many nonviolent games,
while still fast-paced, tend to emphasize problem-solving as a primary
strategy for success, whereas the majority of violent games, while often
incorporating problem-solvingelements, primarily center on impulsive,
“see-and-shoot”strategies. It is theoretically quite feasible that focusing
on logical problem-solving in nonviolent games could make constructs
associated with impulsive, hostile actions less accessible in memory.
Alternately, or in addition, rehearsal of reasoned problem-solving
strategies could temporarily reinforce tendencies toward more con-
trolled evaluation of initial behavioral inclinations, which, according to
the General Aggression Model (GAM), will decrease the likelihood of
hostile aggression (Anderson & Bushman, 2002).
Finally, recent evidence (Gentile et al., 2009; Mares, Palmer, & Sullivan,
2008; Narvaez, Mattan, MacMichael, & Squillace, 2008) indicates that, just
as violent game play can reduce prosocial tendencies, exposure to certain
nonviolent games may elicit increased accessibility of positive, prosocial
constructs. To date, however, such findings have been restricted to games
with explicitly prosocial themes and content. However, to the extent that
other types of nonviolent games lead to a focus on logical, reasoned
problem-solving, it could be that similar effects would emerge even if a
player's aim is not expressly prosocial, which could contribute to potential
decreases in the accessibility of aggression-related constructs. Thus,
testing game effects on prosocial outcomes could provide information
relevant to understanding effects on aggressive outcomes. A secondary
aim of the current study was to test these possibilities.
In sum, the primary purpose of this research was to test the temporal
stability of video game effects on aggression-related outcomes, and to test
whether violent and nonviolent games have divergent effects on
aggression-related and prosocial-related outcomes. In two experiments,
participants were randomly assigned to play either a violent or nonviolent
video game, or to play no video game, after which aggression-related
outcomes were assessed either immediately or after a brief delay. In a
third experiment, we expanded the design to test for potential effects of
violent and nonviolent game exposure on accessibility of prosocial
constructs. The delay manipulation served two related purposes pertinent
to the aims of this research. First, assessing aggressive outcomes both
immediately and following a brief delay helps to establish the temporal
stability of acute game exposure effects. Second, use of a delay
manipulation provides a means of testing whether effects of violent and
nonviolent games produce divergent outcomes on measures of aggres-
sion. If so, measures of aggression-related outcomes should change
significantly in opposing directions in the two game conditions between
the no-delay and the 15-min delay conditions. In other words, compared
to the no-delay assessments, aggression should decrease following delay
for violent game participants and increase following the delay for
nonviolent game participants.
With these issues in mind, 3 primary hypotheses were advanced for
this study: (1) Playing a violent game will cause increases in aggressive
cognitions, affect and behavior, relative to playing a nonviolent game,
but only when outcomes are assessed shortly following game play;
(2) delaying the assessment of aggression following game play will
decrease aggression in the violent game condition and increase
aggression in the nonviolent game condition; and (3) compared to
playing violent games, playing nonviolent games lacking specifically
prosocial content will increase accessibility of prosocial constructs.
Experiment 1
Method
Participants
One hundred eighty-eight undergraduates (131 women) at the
University of North Carolina at Chapel Hill, recruited using a Psychology
Department website, participated in exchange for course credit.
Materials
Video games. Two violent games (Quake 3 and Unreal Tournament:
Game of the Year Edition) and two nonviolent games (Zuma and The
Next Tetris) were used in this study. Both violent games were first-
person shooter games, where players run through a futuristic
gladiator arena and must kill multiple enemies with a variety of
weapons to succeed. Both nonviolent games were visual puzzle
games, in which players must manipulate objects into various
patterns within a time limit in order to succeed.
Questionnaire measures. Participants completed a number of measures
to control for individual differences that could potentially impact
aggression-related outcomes. These included the Interpersonal
Reactivity Index (IRI, Davis, 1980), designed to measure trait
empathy; the Caprara Irritability Scale (CIS, Caprara, 1982), designed
to measure aggressive impulsiveness; and the Aggression Question-
naire (AQ, Buss & Perry, 1992), designed to measure trait aggression.
Differences in other factors that could contribute to aggression,
including frustration, arousal, and interest levels, were assessed using
Likert-type scales anchored at 1 (Not at all) and 7 (Extremely), and
were administered immediately after video game play.
Aggression-related dependent measures
Order of administration of the two dependent measures was
counterbalanced within each condition.
Aggressive affect. Participants completed a measure of state aggressive
affect consisting of 32 state mood statements (e.g., “Ifeelmad”), rated
on a Likert-type scale anchored at 1 (Strongly disagree)and5(Strongly
agree).This measure has beenused in prior studies(Anderson, Deuser,&
DeNeve, 1995) and has good reliability (α=.93).
Aggressive cognitions. Anderson, Carnagey, and Eubanks (2003) devel-
oped a word completion task designed to measure accessibility of
aggressive cognitions. Participants are given a booklet containing 98 word
fragments, each with one or more missing letters. Participants are given
5min to finish as many items as possible. Completed words are then
coded as nonviolent, violent, ambiguous, or non-words using an included
coding guide (available from Craig A. Anderson), and the percentage of
violent words (ambiguous words are counted ½) compared to the total
number of words constitutes the aggressive cognition score.
Delay task
Participants were asked to spend 15 min drawing from memory
“as complete, detailed, and accurate a map as possible”of their
campus. Similar map-drawing tasks have been used as neutral time
fillers in previous studies (Martin, Ward, Achee, & Wyer, 1993).
Procedure
Prior to each session, the experimenter determined condition assign-
ments by two coin flips: one for video game (violent or nonviolent game),
and one for delay (0 min or 15 min). Upon arrival at the laboratory, each
935M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
participant completed an informed consent form and was escorted i nto an
individual room. Participants then completed all self-report measures,
after which they were introduced to their assigned video g ame through an
overview of controls and objectives and a brief practice level. Participants
were instructed to play the game for 30 min with the door closed.
Participants rated their frustration, arousal, and interest levels following
the gameplay period. After this, participants completed the delay task and
aggression measures, the order of which was determined by condition.
Finally, participants were probed for suspicion, debriefed, and excused.
Results
Data from seven participants were excluded from analyses: 2
began playing a video game before instructions were read, 3
experienced computer failures, 1 left the experiment early, and 1
failed to complete either dependent measure. Excluded participants
were slightly more likely to have been in the violent game conditions
(5 violent, 2 non-violent). Thus, analyses were based on data from 181
participants (129 women).
Preliminary analyses
Video game ratings. Compared to nonviolent games, violent games
were rated as significantly more frustrating, less exciting, and less
interesting, ts(179)N2.83, psb.01 (see Table 1). These effects were
due largely to gender differences: women rated the violent games as
more frustrating (Ms=3.55 and 2.47), t(127) = 3.83, pb.001, less
exciting (Ms=2.84 and 4.32), t(127) = −5.74, pb.001, and less
interesting (Ms=2.37 and 4.10), t(127) = −6.35, pb.001. In contrast,
men rated the games similarly on these qualities, ts(50) b1.3, ps N.20.
Individual difference measures. None of the individual difference
measures (CIS, BPAQ, IRI) correlated significantly with either
dependent measure, and thus will not be discussed further.
Analyses of aggressive outcomes
Aggressive cognition and aggressive affect scores were examined
using separate 2 (Game; violent, nonviolent)× 2 (Delay; 0-min, 15-
min)×2 (Gender; male, female) factorial analyses of variance (ANOVAs).
Aggressivecognition.Mean aggressive cognition scores as a function of
game and delay conditions are given in Table 2. The Game main effect
was not significant, F(1, 177)=2.10, p= .149. However, the predicted
Game×Delay interaction was significant, F(1, 177) = 9.98, p=.002.
Follow-up contrasts comparing the effects of Game separately as a
function of delay showed that, in the 0-min delay condition,
participants who played a violent game used significantly more
aggressive word completions than participants who played a
nonviolent game, F(1, 88)=10.01, p= .002. In contrast, aggressive
thought accessibility did not differ as a function of video game in the
15-min delay condition, F(1, 90) =1.47, p= .226. Additional contrasts
showed that, within the violent game condition, aggressive cognitions
decreased after a 15-min time delay, F(1,83) =5.44, p=.021.
Importantly, within the nonviolent game condition, aggressive
cognitions increased after time delay, F(1,94)=6.78, p= .011.
The Game× Delay× Gender interaction also was significant, F(1,
173)= 5.68, p=.018. Although the patterns of means were similar for
men and women, the effect sizes for the delay manipulation were larger
for men within both the violent (d=.80) and nonviolent game
conditions (d=1.21) than for women (violent game d=.021; nonvi-
olent game d=.38).
Aggressive affect. Mean levels of aggressive affect as a function of game
and delay conditions are given in Table 3. The ANOVA showed a main
effect for Game, F(1, 177) =8.82, p=.003, and a significant
Game×Delay interaction, F(1, 177) = 8.51, p=.004. A planned
contrast of the means in the 0-min delay condition replicated the
typical violent game effect: participants who played a violent game
reported significantly more aggressive affect than those who played a
nonviolent game, F(1, 89)=17.17, pb.001. In contrast, in the 15-min
delay condition hostile affect did not differ as a function of game, F(1,
90)=0.00, p= .970. Additional planned contrasts showed that
aggressive affect decreased with delay in the violent game condition,
F(1, 83)=4.48, p= .036, and increased with delay in the nonviolent
game condition, F(1, 94)=4.79, p= .031. The Game× Delay×Gender
interaction was not significant, F(1, 173) =2.90, p=.091.
Covariates. To ensure results were not affected by differential game
experiences, additional analyses were conducted using ratings of the
video game on frustration, excitement, and interest as covariates (i.e.,
ANCOVAs). In each case, inclusion of the covariates did not significantly
change the patterns of means or p-values associated with effects of
interest.
Discussion
Findingsfrom this experimentprovided initialsupport for two of our
primary hypotheses. When outcomes were assessed immediately
following game play, participants who had played violent games
displayed increased accessibility of aggressive thoughts and increased
hostile affect relative to participants who had played a nonviolent game,
consistent with a large number of previous studies (Anderson & Dill,
2000; Anderson & Bushman, 2001; Carnagey & Anderson, 2005;
Bartholow, Sestir, & Davis, 2005). However, when assessment was
delayed by 15 min, game effects were nonexistent. Moreover, and
consistent with hypothesis 2, delaying assessment led to significant
decreases in aggressive outcomes in the violent game condition and
significant increases in the nonviolent game condition. Since aggression
was not assessed prior to gameplay, it is not possibleto conclusively say
Table 1
Mean ratings of violent and nonviolent games on arousal, frustration, and interest in
Experiment 1 (top section) and Experiment 2 (bottom section).
Experiment 1
Games Arousal Frustration Interest
Nonviolent 4.32
a
2.38
c
4.20
e
Violent 3.46
b
3.04
d
3.04
f
Experiment 2
Games Arousal Frustration Interest
Nonviolent 5.00
a
2.47
c
3.90
d
Violent 3.72
b
2.41
c
4.48
e
Note. For each experiment, means within each column that do not share a subscript are
significantly different, pb05.
Table 2
Mean percentage of aggressive word completions as a function of video game and time
delay: Experiment 1.
0 min Delay 15 min Delay
Violent game 22.2%
a
19.7%
b,c
Nonviolent game 18.5%
b
21.1%
c
Note. Means that do not share a subscript are significantly different, pb05.
Table 3
Mean state hostile affect as a function of video game and time delay: Experiment 1.
0 min delay 15 min delay
Violent game 72.41
a
64.73
c
Nonviolent game 57.73
b
64.60
c
Note. Means that do not share a subscript are significantly different, pb05.
936 M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
that nonviolent game exposure reduces aggression relative to baseline;
however, it is widely considered impossible to accurately assess
aggression multiple times within the same paradigm (see Lindsay &
Anderson, 2000; Bartholow, Anderson, Carnagey, & Benjamin, 2005),
making a true pre–post design unfeasible. Therefore, these data suggest,
but do not confirm, that nonviolent game exposure leads to short-term
decreases in both aggressive affect and aggressive cognition.
These findings further suggest that the typical violent video game
effect reported in numerous experiments (see Anderson, 2004;
Anderson & Bushman, 2001) might represent a combined increase
in aggressiveness following exposure to violent games coupled with a
decrease in aggressiveness following nonviolent game exposure.
These data also are consistent with idea that acute exposure effects
are due to priming of aggressive scripts (see Huesmann, 1998) and/or
hostile affect, rather than activation of aggression-related goals (cf.,
Bargh et al., 2001; Forster et al., 2005).
The effects for aggressive cognitions appear to have been much
larger in men than women.Similar gender differences were reported in
a previous video game study (Bartholow & Anderson, 2002). This
difference may be attributable in part to a sex difference in prior video
game experience: in the current study, many more men (98.1%) than
women (65.9%) reported having at least some recent video game
experience. Additionally, male subjects reported higher levels ofviolent
game exposure (M=19.7) relative to females (M=5.1), F(1,179)=
96.3, pb.001. This difference in prior exposure couldhave affected how
men and women experienced the games in the lab (cf., Bartholow,
Sestir,etal.,2005).
It is possible, however, that the increase in aggressive tendencies in
the nonviolent game condition could have been due to frustration
attributable to the delay task; however, this would not explain the
concomitant decrease in aggressiveness in the violent game condition. If
the delay task was especially frustrating to participants, violent game
players would be predicted to plateau or even increase in aggression after
the delay period. We expect, then, that these effects are not attributable
to an idiosyncratic response to the delay task used. However, Study 2
utilized a different delay task to further address this potential criticism.
Experiment 1 had two additional limitations. First, although hostile
affect and aggressive cognitions both were affected by a delay, most
researchers and policy officials are concerned with these outcomes only
to the extent that they have implications for aggressive behavior, a
relationship that the GAM (Anderson & Bushman, 2002)models
explicitly. Therefore, Experiment 2 utilized an aggressive behavior
measure. A second limitation of Experiment 1 was the lack of a control
condition in which no game was played. Presumably, if nonviolent
games are in fact suppressing aggressive tendencies, then aggressive
tendencies among participants who played no video game should
consistently remain between the means for violent and nonviolent
game players. Without such a condition, it is difficult to know whether
the apparent decrease in aggressive outcomes following nonviolent
game exposure is any different than what would be observed with no
game at all. Experiment 2 included a no-game control condition to
address this issue. We predicted that participants who played no video
game would display intermediate levels of aggression (i.e., in-between
violent and nonviolent game players), and that the delay period would
have no effect on their level of aggression.
Experiment 2
Method and procedures
Participants
Three hundred eighty-nine undergraduate men at the University
of North Carolina participated in the study for credit toward a
course requirement. Participants were recruited through a university
website.
Materials
All video games and questionnaires were identical to those used in
Experiment 1.
Delay task
To test the generality of the time delay effect, a different delay task
was used in Experiment 2. Participants in the time-delay conditions
completed a picture rating task for 15 min. In this task, participants
viewed pictures on a computer screen, each displayed for 1000 ms
and separated by a 1000-ms inter-stimulus interval. They were asked
to rate the “colorfulness”of each picture by pressing one of four keys
on the keyboard as the picture appeared. Each participant viewed 120
pictures. All pictures were taken from the International Affective
Picture System (IAPS; Lang, Bradley, & Cuthbert, 2001), and were
selected based on prior ratings of neutral valence (M= 5.46 on a 9-
point scale) and low arousal (M=3.36 on a 9-point scale).
Aggressive behavior
Aggressive behavior was measured using a version of the
Competitive Reaction Time task (see Taylor, 1967). Participants are
told they are competing against another participant in a reaction time
test, in which they must click their computer mouse button as quickly
as possible following the onset of a visual stimulus. Prior to each of 25
trials, the participant sets the intensity (on a 1–10 scale ranging from
60 to 105 dB, with a 0 dB non-aggressive option) of a noise burst to be
delivered to his opponent if the participant were to win that trial. The
average noise intensity across all 25 trials constituted the measure of
aggressive behavior. The participant also receives noise bursts set by
his ostensible opponent on losing trials (12 out of 25).
In actuality, there was no opponent; a computer program deter-
mined all wins and losses as well as noise levels received by participants
on losing trials. The “opponent”delivered an ambiguous pattern of noise
blasts to the participant; this was chosen because prior research
suggeststhat this pattern makes the CRT taskmore sensitive to effects of
other predictor variables. An ambiguous pattern is one in which the first
trial is of intensity level 5, the final 24 trials contain three blasts each at
intensity levels 2–9, and there is a near-zero correlation between trial
number and intensity. However, as a safeguard against suspicion,
participants also automatically lost any trial in which response time
was longer than 750 ms. This task is a widely used (e.g., see Bartholow
& Anderson, 2002; Bushman, 1995; Dorr & Anderson, 1995)and
externally valid measure of aggressive behavior (Anderson and
Bushman, 1997; Anderson, Lindsay, & Bushman, 1999; Carlson,
Marcus-Newhall, & Miller, 1989; Giancola & Chermack, 1998).
Procedure
The basic procedure for Experiment 2 was identical to that of
Experiment 1, with the exception of a die roll that determined if the
session involved gameplay. For the no-game control condition, the
30-min gameplay period was omitted; all other procedures were
identical.
Results
Forty-three participants (11%) were excluded from analyses. This was
due to equipment failure or failure to follow experimental procedures
correctly (n= 24), showing no variability in responses (choosing the
same noise for each trial; n=11) or suspicion concerning the reaction
time task (e.g., that there was no opponent; n=8). Therefore, data from
347 participants were used for analyses.
Manipulation checks
Post-game questionnaire ratings are given in Table 1. Scores on the
laboratory aggression measure were not correlated with any of the
individual difference measures. Thus, no questionnaire scores were
covaried in the main analyses.
937M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
Aggressive behavior
The noise intensity data were analyzed using a 3 (Condition; violent
game, nonviolent game, no-game)×2 (Delay; 0-min, 15-min) factorial
ANOVA. The predicted Condition×Delay interaction was significant, F(2,
347) =5.52, p= .004 (see Fig. 1). This interaction was investigated using a
series of planned contrasts. In the 0-min delay condition, participantswho
played the violent game were significantly more aggressive (M=6.26,
SD =1.43) than participants who played the nonviolent game (M=5.66,
SD =1.50), F(1, 113)= 7.28, p= .008. Violent game players displayed
marginally more aggression relative to the control group (M= 5.94),
F(1,103) = 3.35 , p= .07; however, aggression levels did not differ
significantly between the nonviolent game and control conditions,
F(1,115)=.89, p= .35. Ag gression levels did not differ sig nificantly
in the 15-min delay condition (Ms =5.8, 6.03, and 5.68 in the violent,
nonviolent, and no-game conditions), F(2, 183) =0.82, p=.442.
Additiona l contrasts examined t he effects of delay within levels of
the Game variable. The 15-min delay, relative to the 0-min delay,
resulted i n significantly less aggressive responses in the violent game
condition, F(1, 101) =6.31, p=.014, and significantly more aggres-
sive responses in the nonviolent game condition, F(1, 132)= 4.90,
p=.029. Delay had no effect on aggression levels in the control
condition, F(1, 108) =0.27, p=.604.
Covariates. To ensure results were not affected by differential game
experiences, ratings of the video game on frustration, excitement, and
interest were again covaried out of each measured comparison. As in
Experiment 1, patterns of means and p-values for effects of interest
were not significantly altered by inclusion of the covariates.
Discussion
The data from Experiment 2 again supported our primary hypotheses,
this time using an aggressive behavior measure. As with measures of
aggressive cognition and hostile affect used in Experiment 1, assessment
of aggressive behavior immediately following game play showed the
typical violent video game effect, with significantly more aggression
displayed by those who played a violent compared to a nonviolent game
(see Anderson & Bushman, 2001; Anderson & Bushman, 2002). When
assessment was delayed 15 min, this time using a picture viewing task,
this difference in aggression disappeared. Most importantly, whereas the
delay significantly changed aggression levels in the predicted directions
for both game-play conditions, as in Experiment 1, aggression levels in
the no-game control condition were not affected by the delay. As in
Experiment 1, with no pre-assessment of aggression, and no significant
difference between the nonviolent and control condition, it is not
possible to conclusively state that nonviolent game exposure is reducing
aggression relative to baseline. However, as in Experiment 1, this pattern
suggests that both types of games–violent and nonviolent–have unique
and opposing effects on aggression-related outcomes.
The fact that responses in the two game-play conditions did not
differ significantly from responses in the control condition further
suggests that interpretation of the violent game effect depends on the
choice of a comparison group. However, it is also important to
consider that the effect of the violent game (compared to the no-game
control) on aggression was not negligible (d= .34; p=.07), suggest-
ing that this effect might have reached statistical significance with a
larger sample.
Findings from the first two experiments converge on the idea that
although violent games appear to increase aggressive tendencies,
nonviolent games seem to actually reduce aggressive outcomes.
However, the mechanism for these seemingly opposing effects is
unclear. Based on predictions derived from the GAM (see Anderson &
Bushman, 2002), it could be that exposure to nonviolent games
decreases accessibility of hostile cognitions, hostile affect and
aggressive behavioral scripts, increases the accessibility of opposing,
prosocial constructs, or both. Although the data from Experiment 1
provided some evidence in support of the former possibility, it is not
clear whether exposure to nonviolent games that lack explicitly
prosocial content have any effects on the accessibility of prosocial
cognitions. Previous research has shown that exposure to explicitly
prosocial video games can increase prosocial script accessibility
(Narvaez et al., 2008) and prosocial behavior (Mares, Palmer, &
Sullivan, 2008). Past research (e.g. Anderson & Bushman, 2001) also
has shown that violent game play not only increases aggression but
also decreases prosocial tendencies. If nonviolent game play essen-
tially counters effects of violent game play, it may serve to increase
activation of prosocial constructs as well.
On the other hand, it could be that whereas exposure to violent
games decreases accessibility of prosocial constructs, nonviolent
games lacking prosocial content do not appreciably affect prosocial
outcomes relative to a control condition. Such a pattern would suggest
that the findings from the first two experiments reflect effects on
aggression-related constructs only, and not a combination of effects
on both aggressive and prosocial processes. Experiment 3 was
designed to test these possibilities.
Experiment 3
Method and procedures
Participants
One hundred and eleven undergraduate students (68 female) at
the University of North Carolina and Gettysburg College participated
in the study for credit toward a course requirement. Participants were
recruited through websites at each University.
Materials
Games and initial questionnaires were identical to those utilized in
Studies 1 and 2.
Dependent measures
Order of administration of the two dependent measures was
counterbalanced within each condition.
Story completion task. In order to assess the accessibility of prosocial and
aggressive cognitions, participants were given three ambiguous
unfinished short stories, each detailing a potentially stressful incident
(a car accident, a disagreement with a friend,not being served quicklyin
a restaurant), and asked to provide 15itemsdetailing “what could have
happened next”for each. This measure has been used in several
previous studies (e.g. Rule, Taylor & Dobbs, 1987; Bushman & Anderson,
2002; Narvaez et al., 2008). Each response was coded for aggressive,
prosocial, or neutral content by two independent judges; the
Fig. 1. Mean aggression levels as a function of game play and measurement delay
conditions, Experiment 2.
938 M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
proportions of prosocial and aggressive responses were the variables of
interest. Prosocialresponses were behaviors, thoughts, or emotions that
supported the well-being of another person (e.g. helping a friend
purchase a stereo, concern for the other driver in an accident).
Aggressive responses were those that involved behaviors, thoughts or
emotions that could do harm to another person (e.g. punching theother
driver, the desire to have a waiter fired). Initialagreement was very high
(ICC= .88), and judges conferred to agree on all discrepant items.
Word completion task. The aggressive cognition task (Anderson,
Carnagey, & Eubanks, 2003) used in Experiment 1 was modified to
include a prosocial subscale. The majority of the scale was identical to
the original, but 30 items were also created where the word fragment
could be completed as a neutral or prosocial word; for example, A_D
could be completed as AND (neutral) or AID (prosocial). The final
measure consisted of 90 items—30 that could be completed as
aggressive or neutral, 30 that could only be completed as neutral, and
30 that could be prosocial or neutral. The first two groups were taken
from the original measure, and the third was developed for this study.
The percentage of completed words that were aggressive and prosocial
was the variable of interest.
Procedure
The basic procedure for Experiment 3 was similar to that of
Experiments 1 and 2, with two exceptions: all participants completed
the dependent measures immediately (i.e., there was no delay
condition), and a die roll determined if the session involved gameplay.
For the no-game control condition, the 30-min gameplay period was
omitted; all other procedures were identical. Once gameplay was
completed, all participants were administered the Story Completion
Task and Word Completion Task in counterbalanced order.
Results
Preliminary analyses indicated that participant gender did not
significantly influence any of the effects of interest. Thus, gender was
not included in the primary analyses. Percentage of aggressive and
prosocial words used in the Word Completion Task and percentage of
aggressive and prosocial story completions generated in the Story
Completion Task were subjected to separate univariate ANOVAs.
Means for both measures as a function of Condition are presented in
Table 4.
Word completion task
There was a significant main effect for Game on both aggressive
words, F(2, 108)= 6.72, pb.01, and prosocial words, F(2,108)= 3.19,
p=.045. Planned contrasts found that violent game players chose
significantly more aggressive words (M=16.9%) than nonviolent
game players (M=13.3%), F(1,66)= 7.38, pb.01. Additionally, violent
game players used significantly more aggressive words than control
participants (M=12.9%),F(1,74) =11.97, pb.001; however, there was
no significant difference between nonviolent and control participants,
F(1,76) = .13, p=.72.
Violent game players also chose significantly fewer prosocial words
(M=11.4%) than nonviolent players (M=13.6%), F(1,66)= 4.21,
p=.044. Nonviolent game players chose significantly more prosocial
words than control participants (M=11.3%), F(1,76)= 5.10, p=.027;
but there was no significant difference between violent game players
and control participants, F(1,76)= .01, p=.92.
Story completion task. There was a significant main effect for Game
on aggressive items, F(2,108)=6.57, pb.01, and prosocial items,
F(2,108)=5.28, pb.01. Violent game players again chose significantly
more aggressive story completions (M=19.7%) than nonviolent game
players (M=12.9%), F(1,66)= 10.3, pb.01. There was a marginally
significant effect when comparing violent game players to control
participants (M=16.0%), F(1,74)= 3.62, p= .061; the nonviolent
condition chose significantly fewer aggressive responses than control,
F(1,76)= 5.26, p=.025.
Violent game players also chose significantly fewer prosocial story
completions (M=12.1%), relative to nonviolent players (M=16.7%),
F(1,66)=9.73, pb.01. Additionally, nonviolent game players gener-
ated significantly more prosocial items than the control participants
(M=14.1%), F(1,76)= 4.62, p= .035, but violent players showed no
significant difference relative to control (1,74)=2.13, p= .149.
Discussion
The results from Experiment 3 largely supported our hypotheses.
Violent game players again showed significantly more aggressive
responses on both dependent measures, relative to both nonviolent
game players and no-game control participants. Nonviolent game
players provided significantly more prosocial responses on both
dependent measures relative to violent game players, and relative to
control participants on the word completion task. Nonviolent game
players also showed fewer aggressive responses than the control
group on the story completion task, although this was not true of the
word completion task. More importantly, nonviolent game players
gave more prosocial responses than either no-game control or violent
game players on both measures (see Table 4).
First, supportive evidence was found for nonviolent games serving
as suppressors of aggressive tendencies. On the story completion
measure, participants who played no game at all generated signifi-
cantly more aggressive responses to the story stems than nonviolent
game players. This would seem to indicate that nonviolent game play
lowered aggressive tendencies. However, this pattern was not seen
for the word completion task, where the nonviolent and control
conditions were nearly identical. It is possible that these conflicting
results are due to the more open-ended nature of the story
completion task—partially completed words may have still elicited
aggressive responses from nonviolent game players given the limited
set of potential completions for each item, but when asked to generate
answers to a more open-ended question, aggressive scripts may have
been less accessible relative to control. However, it is difficult to be
certain as to the cause of these results.
Additionally, this study provided indications that prosocial cogni-
tions can be elicited when participants played games that were not
explicitly prosocial. Nonviolent game players generated significantly
more prosocial responses on both tasks relative to the violent game
and control conditions. This is a novel finding, and is explored further
in General discussion.
General discussion
The current studies had two primary aims: to investigate the
persistence of the well-known violent video game effect, and to
Table 4
Mean proportion of aggressive and prosocial word completions (top) and story
completions (bottom): Experiment 3.
Word completion
Violent game No-game control Nonviolent game
Aggressive words 16.9
a
12.9
b
13.3
b
Prosocial words 11.4
a
11.3
a
13.6
b
Story completion
Violent game No-game control Nonviolent game
Aggressive completions 19.7
a
16.0
a
⁎12.9
b
Prosocial completions 12.1
a
14.1
a
16.7
b
Note. Within each section, means within each row thatdo not share a subscript differ, pb.05.
⁎Aggressive completion means in the Violent gameandNo-gamecontrolconditiondiffered
at p=.06.
939M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
determine whether that effect could be attributable, in part, to
suppression of aggression by nonviolent game play. Addressing the
first of these aims had implications for understanding the theoretical
mechanism(s) responsible for violent game effects. Specifically, the
decrease in aggressive outcomes in the violent game condition
following a delay suggests that acute exposure effects are due
primarily to short-term increases in accessibility of aggression-related
constructs in memory (i.e., a priming effect; see Huesmann, 1998; see
also Higgins, 1996), and probably do not reflect activation of
aggression-related goal structures (Bushman & Anderson, 2001).
Addressing the second aim of this research also has important
implications for understanding video game effects typically
reported in the experimental literature. Whereas delaying assess-
ment by 15 min led to significant decreases in aggression-related
outcomes for participants who played violent games, the delay also
led to significant increases in those outcomes among nonviolent
game participants. Delay did not significantly affect aggressive
behavior in the no-game control condition (Experiment 2). This
pattern indicates not only that acute video game exposure effects
are relatively short-lived, but also that nonviolent games appear to
reduce aggression-related outcomes, and therefore provide a
false baseline for assessing effects of violent games on aggressive
behavior.
One implication of this finding is that researchers interested
in understanding the effects of violent games on aggression-
related outcomes would do well to include both nonviolent
game and no-game control conditions. A broader implication is
that video game exposure effects should be interpreted in terms
of the combined effects of both violent games and nonviolent
games. The current findings show that even when games are
matched on major aggression-relevant variables, it is possible that
nonviolent games may have independent effects on aggressive
tendencies. Thus, it seems inappropriate to attribute differences
between gaming conditions on outcomes of interest solely to effects
of violent games.
The present results also have implications for theoretical models
of aggression. For example, the GAM (Anderson & Bushman, 2002)
does not explicitly outline a mechanism for the reduction of
aggression, but the factors in the model depicted as increasing
aggression (increased aggressive script accessibility, aggressive
affect, hostile schemas) would seem to have nonviolent converses
(explicitly pacifistic script accessibility, non-aggressive affect,
forgiving schemas). If exposure to typical nonviolent games elicited
these responses, then a reduction in state aggression would be
consistent with the GAM. All nonviolent games utilized in this study
focused on puzzle solving within a cartoonish environment. It is
possible that a focus on logical, reasoned thinking–as opposed to the
frenzied, see-and-shoot environment of violent games–might serve
to activate cognitive constructs and affect that could effectively
reduce aggressive inclinations below an individual's usual baseline.
Moreover, just as repeated exposure to violent games is hypothe-
sized to have a lasting impact on aggressive outcomes via
chronically heightened accessibility of aggressive scripts and hostile
affect (Anderson & Dill, 2000; Carnagey & Anderson, 2005)and/or
desensitization (see Bartholow, Sestir, & Davis, 2005; Carnagey,
Anderson, & Bushman, 2007), repeated exposure to nonviolent
games could have the opposite effect. Moreover, from a behavioral
engineering perspective, time spent on nonviolent games will
reduce exposure to violent games, thereby alleviating some
concerns about long-term exposure effects (e.g., see Anderson
et al., 2008).
The findings from Experiment 3 provide clarification as to the
potential mechanisms of the effect of nonviolent game play on
reducing aggression. Specifically, nonviolent games seem to not only
decrease the accessibility of hostile cognitions and affect, but such
games–even those lacking explicitly prosocial content–also appear to
increase the accessibility of prosocial cognitions. Thus, just as violent
and nonviolent games have opposing effects on aggressive outcomes
their effects also oppose one another with respect to prosocial
outcomes. This pattern has implications for understanding both the
findings of previous research testing video game effects on prosocial
outcomes (e.g., Gentile et al., 2009; Narvaez et al., 2008) and the
psychological mechanisms that contribute to video game effects on
aggression. First, with respect to previous studies of video games and
prosocial outcomes, the current data suggest that explicitly prosocial
content in video games is sufficient but not necessary to increase
prosocial responses. Rather, our findings suggest that even games that
focus the player on logical puzzle-solving have similar effects.
Prosocial constructs should, at least, be primed by modeling and
observing instances of similar behavior within a game. However,
actual prosocial content does not appear to be necessary to elicit such
an effect. Future work should compare the effects of explicitly
prosocial games with the type of nonviolent games used here to
determine whether they produce any differences in prosocial
outcomes.
Second, when considered as a set, the findings from all three of
the current studies suggest that the contribution of nonviolent
game exposure to the typical “violent video game effect”reflects
the combined influence of decreases in the accessibility of aggres-
sive constructs and increases in the accessibility of opposing,
prosocial constructs. Experiment 1 demonstrated that nonviolent
game exposure produces short-term decreases in accessibility of
aggressive cognitions and hostile affect, Experiment 2 produced an
analogous set of findings for aggressive behavior, and Experiment
3 showed that playing nonviolent games increases prosocial con-
struct accessibility.
It is important to note that none of the substantive findings of
these studies seem to directly contradict the general implications of
prior violent game research: namely, that violent game play does
increase aggressive outcomes. The current work shows that due to
nonviolent games providing a false baseline, these effects may not be
as large as previously assumed, but the tendency toward increased
aggression was consistently replicated. While prior work directly
comparing violent gameplay to a no-game control is rare, findings
have supported an independent effect of violent play (e.g. Calvert &
Tan, 1994).
Although in theory it would be desirable to specify the nature of
the associations among these outcomes by assessing them all in the
same participants, thereby allowing a statistical test of mediation
(e.g., Baron & Kenny, 1986; MacKinnon, Lockwood, Hoffman, West,
& Sheets, 2002), such tests rely on the assumption that the act of
measuring more proximal variables (e.g., accessibility of aggressive
cognitions) not interfere with or influence more distal variables
(e.g., aggressive behavior). Unfortunately, this assumption often is
violated in the aggression laboratory (see Lindsay & Anderson,
2000), for at least two reasons. First, as underscored by the current
results, effects of manipulated variables on aggression in the lab
tend to be fleeting, and thus effects tend to be more pronounced on
outcomes measured first compared to those measured later (see
Lindsay & Anderson, 2000). Second, measuring proximal variables
(e.g., accessibility of aggressive thoughts) can alter subsequent
behaviors in important ways, such as by inadvertently revealing the
true purpose of the competitive reaction time task (i.e., measuring
aggressive behavior). As discussedindetailelsewhere(see
Bartholow, Anderson, et al., 2005; Bartholow, Sestir, et al., 2005;
Lindsay & Anderson, 2000), this problem is akin to the Heisenberg
uncertainty principle in quantum mechanics, which generally holds
that measuring one observable quantity increases the uncertainty
with which other quantities can be known, because measurement of
one variable influences the values of other related variables.
However, the symmetry in the patterns of outcomes across
experiments strongly suggests that nonviolent game exposure
940 M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
reduces aggressive responding in the short-term via concurrently
suppressing accessibility of aggression-related constructs and
enhancing accessibility of prosocial constructs. Although we have
speculated that these effects could occur because nonviolent games
induce a more reasoned, less impulsive mind-set in players, the
mechanism for these effects remains to be specified in future
research.
A number of shortcomings of the current research could limit the
generalizability of these findings. First, it is possible that the
nonviolent game effect could be restricted to the specificsubsetof
games used here. One caveat to this potential limitation is that many
games that are completely lacking in violent content tend to have
strong puzzle-solving or logical components, and thus these findings
could apply to a broad range of nonviolent games. Future work
should test this hypothesis with different types of nonviolent games.
It could also be the case that the effects seen after the delay
manipulatio n are specific to a 15-min del ay. It is difficult to say where
exactly the “tipping point”might be in terms of the post-delay shift in
means seen in Experiments 1 and 2; use of varying delay manipula-
tions in future studies could help address this. Finally, it is possible
that the specific nature of the delay task could have impacted the
experimental outcomes. Different tasks were used for Experiments 1
and 2 to help address this possibility, but it cannot be conclusively
stated that the delay tasks had no independent effect on the
dependent measures.
In addition, generalizability of the effects across gender groups has
yet to be strongly established. While there were no significant gender
effects in Experiment 3, Experiment 1 suggested that the effects
occurred primarily in men, and Experiment 2 did not include women.
Future research should attempt to reconcile these discrepancies and
arrive at a clearer picture of potential gender differences in gameplay
effects.
In sum, the current findings suggest that, in addition to the well-
documented deleterious effects of violent video games on antisocial
feelings, thoughts, and behaviors, nonviolent video games might also
have positive, prosocial effects. Although the current findings suggest
that these effects stem from both a “damping down”of aggressive
inclinations and the activation of opposing, prosocial constructs,
confirmation of this hypothesis awaits future research. In any case, it
appears clear that media of varying content have the potential to
change patterns of social interaction. To paraphrase an old saying, you
are what you consume; this research indicates that in the case of
media, this can be a blessing as well as a curse.
Appendix A
Prosocial word completion items.
References
Anderson, C. A. (2004). An update on the effects of playing violent video games. Journal
of Adolescence,27, 113−122.
Anderson, C. A., & Bushman, B. J. (1997). External validity of “trivial”experiments: the
case of laboratory aggression. Review of General Psychology,1,19−41.
Anderson, C. A., & Bushman, B. J. (2001). Effects of violent video games on aggressive
behavior, aggressive cognition, aggressive affect, physiological arousal, a nd
prosocial behavior: a meta-analytic review of the scientific literature. Psychological
Science,12, 353−359.
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 aggressive thoughts and feelings. Journal of
Personality and Social Psychology,84, 960−971.
Anderson, C. A., Deuser, W. E., & Deneve, K. M. (1995). Hot temperatures, hostile affect,
hostile cognition, and arousal: tests of a general theory 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,
772−790.
Anderson,C.A.,Lindsay,J.J.,&Bushman,B.J.(1999).Researchinthe
psychological laboratory: truth or triviality? Current Directions in Psychological
Science,8,3−9.
Anderson, C. A., Sakamoto, A., Gentile, D. A., Ihori, N., Shibuya, A., Yukawa, S., Naito, M., &
Kobayashi, K. (2008). Longitudinal effects of violent video games on aggression in
Japan and the United States. Pediatrics,122, e1067−e1072.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall.
Bargh, J. A., Gollwitzer, P. M., Lee-Chai, A., Barndollar, K., & Troetschel, R. (2001). The
automated will: nonconscious activation and pursuit of behavioral goals. Journal of
Personality and Social Psychology,81, 1014−1027.
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.
Bartholow, B. D., & Anderson, C. A. (2002). Examining the effects of violent video games
on aggressive behavior: potential sex differences. Journal of Experimental Social
Psychology,38, 283−290.
Bartholow, B. D., Anderson, C. A., Carnagey, N. L., & Benjamin, A. J., Jr. (2005). Interactive
effects of life experience and situational cues on aggression: the weapons priming
effectin hunters and nonhunters.Journal of Experimental Social Psychology,41,48−60.
Bartholow, B. D., Sestir, M., & Davis, W. (2005). Correlates and consequences of
exposure to videogame violence: hostile personality, empathy, and aggressive
behavior. Personality and Social Psychology Bulletin,31(11), 1573−1586.
Berkowitz, L. (1990). On the formation and regulation of anger and aggression: a
cognitive-neoassociationistic analysis. American Psychologist,45, 494−503.
Bushman, B. J. (1995). Moderating role of trait aggressiveness in the effects of violent
media on aggression. Journal of Personality and Social Psychology,69, 950−960.
Bushman, B. J., & Anderson, C. A. (2001). Media violence and the American public:
scientific facts vs. media misinformation. American Psychologist,56, 477−489.
Bushman, B. J., & Anderson, C. A. (2002). Violent video games and hostile expectations:
a test of the general aggression model. Personality and Social Psychology Bulletin,28,
1679−1686.
Buss, A. H., & Perry, M. P. (1992). The aggression questionnaire. Journal of Personality
and Social Psychology,63, 452−459.
Calvert, S. L., & Tan, S. (1994). Impact of virtual reality on young adults’physiological
arousal and aggressive thoughts. Journal of Applied Developmental Psychology,15,
125−139.
Caprara, G. V. (1982). A comparison of the frustration-aggression and emotional
susceptibility hypothesis. Aggressive Behavior,8, 234−236.
Carlson, M., Marcus-Newhall, A., & Miller, N. (1989). Evidence for a general construct of
aggression. Personality and Social Psychology Bulletin,15, 377−389.
Word Fragment Prosocial Neutral
H _ _ p Help Heap, Harp
S h _ r _ Share Shard, Short
P _ _ c e Peace Place, Piece
S _ _ e Save Some, sore
A _ d Aid Add, and
S _ i l _ Smile Still, Spill
T _ _ c e Truce Twice, Trace
R e _ e _ _ Redeem Revert, reheat
U n _ _ e Unite Untie, Uncle
P r _ _ e _ t Protect Protest, Present
H _ g Hug Hag, Hog
F r _ _ Free Fray, Fret
H _ _ l Heal Hell, Howl
_ i v e Give Live, Hive
(continued)
Word Fragment Prosocial Neutral
A _ _ _ s t Assist Attest, Artist
T _ _ s t Trust Toast, Twist
R _ s c _ _ Rescue Rascal, Rescan
P r o _ _ _ e Promote Produce, Prorate
A g _ _ e Agree Agate, Agape
C o _ _ o r t Comfort Contort, Comport
L _ v _ Love Live, Lava
B e n _ _ _ _ Benefit Bending, Benzene
S u _ _ o r _ Support Suitors, Succors
C o n _ _ _ e Console Convene, Concede
F o r _ _ _ _ Forgive Forests, Forgets
S _ _ t h e Soothe Seethe
P a _ d _ _ Pardon Payday, Pandas
R _ s t _ _ _ Restore Resting, Rusting
_ _ _ t u r e Nurture Picture, Venture
R _ v i _ e Revive Revise, Ravine
Note: List of potential neutral completions is not comprehensive.
Appendix A (continued)
941M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
Carnagey, N. L., & Anderson, C. A. (2005). The effects of reward and punishment in
violent video games on aggressive affect, cognition, and behavior. Psychological
Science,16, 882−889.
Carnagey,N. L., Anderson, C. A., & Bushman, B. J. (2007). The effectof video game violence
on physiological desensitization to real-life violence. Journal of Experimental Social
Psychology,43,489−496.
Davis, M. H. (1980). Measuring individual differences in empathy: evidence for a
multidimensional approach. Journalof Personality and SocialPsychology,44,113−126.
Dorr, N., & Anderson, C.A. (1995). Resolution of the temperature aggression debate. Paper
presented at the Annual Convention of the Midwestern Psychological Association,
Chicago.
Forster, J., Liberman, N., & Higgins, E. T. (2005). Accessibility from active and fulfilled
goals. Journal of Experimental Social Psychology,41, 220−239.
Freedman, J. (2001). Evaluating the research on violent video games. Paper delivered at Playing
By the Rules: The cultural policy challenges of video games. : University of Chicago Available
online at: http://culturalpolicy.uchicago.edu/conf2001/papers/freedman.html
Gentile, D. A. (2009). Pathological video game use among youth 8 to 18: a national
study. Psychological Science,20, 594−602.
Gentile, D. A., & Anderson, C. A. (2003). Violent video games: the newest media violence
hazard. In D. Gentile (Ed.), Media Violence and Children (pp. 131−152). Westport,
CT: Praeger.
Gentile,D.A.,Anderson,C.A.,Yukawa,S.,Ihori,N.,Saleem,M.,Ming,L.K.,Shibuya,
A.,Liau,A.K.,Khoo,A.,&Sakamoto,A.(2009).Theeffectsofprosocialvideo
games on prosocial behaviors: international evidence from correlational, experi-
mental, and longitudinal Studies. Personality and Social Psychology Bulletin,35,
752−763.
Giancola, P. R., & Chermack, S. T. (1998). Construct validity of laboratory aggression
paradigms: a response to Tedeschi and Quigley (1996). Aggression and Violent
Behavior,3, 237−253.
Higgins, E. T. (1996). Knowledge activation: accessibility, applicability, and salience. In
E. T. Higgins, & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic
principles (pp. 133−168). New York: Guilford Press.
Huesmann, L. R. (1998). The role of social information processing and cognitive schema
in the acquisition and maintenance of habitual aggressive behavior. In R. G. Geen, &
E. Donnerstein (Eds.), Human aggression: theories, research, and implications for
social policy (pp. 73−109). San Diego: Academic Press.
Lang, Bradley, & Cuthbert (2001). International affective picture system (IAPS):
Instruction manual and affective ratings. Technical Report A-5, The Center for
Research in Psychophysiology, University of Florida.
Lindsay,J.J.,&Anderson,C.A.(2000).Fromantecedentconditionstoviolentactions:ageneral
affective aggression model. Personality and Social Psychology Bulletin,26, 533−547.
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002). A
comparison of methods to test mediation and other intervening variable effects.
Psychological Methods,7,83−104.
Mares, M., Palmer, E., & Sullivan, T. (2008). Prosocial effects of media exposure. In L.
Robert,P.Jenkins,&T.Gullotta(Eds.),Preventing violence in America
(pp. 133−155). Thousand Oaks: Sage Publications.
Martin, L. L., Ward, D. W., Achee, J. W., & Wyer, R. S., Jr. (1993). Mood as input: people
have to interpret the motivational implications of their moods. Journal of
Personality and Social Psychology,64, 317−326.
Narvaez,D.,Mattan,B.,MacMichael,C.,&Squillace,M.(2008).Killbandits,collectgoldorsave
the dying: the effects of playing a prosocial video game. Media Psychology Review,1(1).
Rule, B. G., Taylor, B. R., & Dobbs, A. R. (1987). Priming effects of heat on aggressive
thoughts. Social Cognition,5, 131−143.
Smith,S.,Lachlan,K.,&Tamborini,R.(2003).Popular video games: quantif ying the presentation
of violence and its context. Journal of Electronic and Broadcasting Media,47(1), 58−76.
Taylor, S. P. (1967). Aggressive behavior and physiological arousal as a function of
provocation and the tendency to inhibit aggression. Journal of Personality,35, 297−310.
942 M.A. Sestir, B.D. Bartholow / Journal of Experimental Social Psychology 46 (2010) 934–942
