the effect of video game competition and violence on aggressive behavior: which characteristic has...

The Effect of Video Game Competition and Violence onAggressive Behavior: Which Characteristic Has the

Greatest Influence?

Paul J. C. Adachi and Teena WilloughbyBrock University, Canada

Objective: This study is the first to our knowledge to isolate the effect of video gameviolence and competitiveness on aggressive behavior. Method: In Pilot Study 1, aviolent and nonviolent video game were matched on competitiveness, difficulty, andpace of action, and the effect of each game on aggressive behavior was then comparedusing an unambiguous measure of aggressive behavior (i.e., the Hot Sauce Paradigm)in Experiment 1. In Pilot Study 2, competitiveness was isolated by matching games ondifficulty and pace of action, and systematically controlling for violence. The effect ofvideo game competition on aggressive behavior was then examined in Experiment 2.Results: We found that video game violence was not sufficient to elevate aggressivebehavior compared with a nonviolent video game, and that more competitive gamesproduced greater levels of aggressive behavior, irrespective of the amount of violencein the games. Conclusion: It appears that competition, not violence, may be the videogame characteristic that has the greatest influence on aggressive behavior. Futureresearch is needed to explore the mechanisms through which video game competitive-ness influences aggressive behavior, as well as whether this relation holds in thelong-term.

Keywords: violent video games, aggressive behavior, competitiveness, violence

The effect of violent video games on aggres-sion, which is defined as behavior that is in-tended to harm another individual (Coie &Dodge, 1998), is a hot topic today as videogames continue to increase in popularity. Forinstance, a large scale study in the United Statesfound that 88% of youth aged 8 to 18 years playvideo games (Gentile, 2009). In terms of fre-quency, youth played 3 or 4 times per week onaverage (median), and the average amount oftime spent playing video games per weekwas 13.2 hours. Although some research hasfailed to find a relation between violent videogame play and aggression (e.g., Ferguson &Reuda, 2010; Ferguson et al., 2008; Williams &

Skoric, 2005), other research has shown thatplaying violent video games produces higherlevels of aggressive behavior, aggressive cogni-tion, aggressive affect, and physiologicalarousal in the short-term than nonviolent videogames (see Anderson, Gentile, & Buckley,2007). However, there are three major limita-tions with the studies that have found an effect.First, to date, no study has matched a violentand nonviolent video game on competitiveness,difficulty, and pace of action simultaneously,and thus, the violent content has not been iso-lated. Consequently, it is unclear whether theviolent content alone is responsible for elevatedlevels of aggression. Second, previous experi-mental studies have tended to use a measure ofaggression that may also measure competitive-ness, leading to questions about whether violentvideo games are related to aggression or com-petitiveness. Third, the effect of video gamecompetition on aggressive behavior has notbeen examined. Hence, the goal of the currentstudy was to examine whether a violent videogame produced greater levels of aggression thana nonviolent video game using an unambiguous

This article was published Online First August 15, 2011.Paul J. C. Adachi and Teena Willoughby, Department of

Psychology, Brock University, Ontario, Canada.Teena Willoughby received funding from the Social Sci-

ences and Humanities Research Council of Canada and theOwl Children’s Trust.

Correspondence concerning this article should be ad-dressed to Paul J. C. Adachi, Department of Psychology,Brock University, St. Catharines, Ontario, Canada, L2S3A1. E-mail: [email protected]

Psychology of Violence © 2011 American Psychological Association2011, Vol. 1, No. 4, 259–274 2152-0828/11/$12.00 DOI: 10.1037/a0024908

259

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

measure of aggressive behavior, when bothgames were equated on competitiveness, diffi-culty, and pace of action. In addition, we testedwhether a competitive video game producedmore aggressive behavior than a less competi-tive video game when matched on violence,difficulty, and pace of action.

The General Aggression Model

The most comprehensive theory of the asso-ciation between violent video games and ag-gression is Anderson and Bushman’s (2002)General Aggression Model (GAM), which wasadapted from past theories of aggression (seealso Anderson & Carnagey, 2004, for a detaileddescription of the model). The model depicts acyclical relationship between an individual andthe environment, in which person variables suchas trait hostility, as well as situation variablessuch as exposure to real-world or media vio-lence (e.g., violent video games), interact toinfluence an individual’s present internal state.Within an individual’s internal state are cog-nition (aggressive scripts or hostile thoughts),affect (anger and frustration), and arousal (el-evated heart rate or blood pressure). Cogni-tion, affect, and arousal are the hypothesizedmechanisms that interact to then influence anindividual’s aggressive behavior. Accordingto Anderson and Bushman, violent videogames function as a situation variable that canincrease aggressive cognition, affect, andarousal, in turn leading to increased aggres-sive behavior.

Empirical Background

Experimental studies examining the short-termeffect of violent video games on aggression havetypically involved randomly assigning partici-pants to play either a violent or nonviolent videogame, followed by a measure of aggression (e.g.,Bushman & Anderson, 2002). Some of the re-searchers have found that participants in theviolent video game condition have shown moreaggression than participants in the nonviolentcondition for both men and women (see Ander-son et al., 2007, 2010, for a detailed review, aswell as Ferguson & Kilburn, 2010, for a critiqueof this research). However, there are three lim-itations with this research that have yet to be

concurrently addressed in a single study. Eachlimitation will be reviewed in turn.

Differences Between Violent and NonviolentVideo Games Other Than Violence

In general, violent video games tend to bemore competitive than nonviolent video games(Carnagey & Anderson, 2005). Consequently,studies that have found that violent video gamesproduced more aggression than nonviolentvideo games, but failed to equate the games oncompetitiveness, cannot conclude that the vio-lent content alone was responsible for the ele-vated levels of aggression. We propose thatviolence (e.g., fighting, shooting, killing), com-petitiveness (e.g., competing with other playersor computer-controlled opponents), difficulty(e.g., how difficult the game is to successfullycomplete), and pace of action (e.g., rate of speedof action sequences) are four main video gamecharacteristics that may influence aggressivebehavior through the mechanisms (i.e., physio-logical arousal, aggressive cognition, and ag-gressive affect) proposed by the GAM (seeAdachi & Willoughby, 2011, for a detailed ex-planation). For example, Barlett, Branch, Rode-heffer, and Harris, (2009) found that a violentvideo game produced greater elevations in heartrate, hostility, aggressive thoughts, and aggres-sive behavior compared with a nonviolent videogame. Similarly, video game competition mayinfluence heart rate, as well as aggressivethoughts and feelings (see Adachi & Wil-loughby, 2011, for a more detailed discussionregarding the relation between competition, dif-ficulty, and pace of action, and the mechanismsproposed by the GAM).

Although researchers have attempted toequate games on competitiveness, difficulty,and pace of action, no one to date has equated aviolent and nonviolent game on these charac-teristics simultaneously. For example, Carnageyand Anderson (2005) attempted to control forcompetition while examining the effect of videogame violence on aggressive behavior by ma-nipulating the game-play settings of the car-racing video game Carmageddon 2. Participantswere randomly assigned to one of three condi-tions: (a) awarded points for destroying othervehicles during the race (violence rewarded),(b) deducted points for destroying other vehi-cles during the race (violence punished), or (c)

260 ADACHI AND WILLOUGHBY

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

could not come into contact with other vehiclesduring the race (nonviolent). Because the samegame was used in all three conditions, Carnageyand Anderson assumed that the level of com-petitiveness across the conditions was equal;however, without having participants rate eachcondition in terms of competitiveness, it is un-clear whether they were actually equal. For in-stance, in the violence-punished and nonviolentconditions, there is only one competitive goal:defeat the other opponents in the race. However,in the violence-rewarded condition there aretwo competitive goals: defeat the other oppo-nents in the race and defeat the other opponentsin a battle for survival. Thus, the violence-rewarded condition contained more competitionwhich may have caused participants to feelmore competitive, and in turn, behave moreaggressively than the participants in the vio-lence-punished and nonviolent conditions.

Problems With Measures of Aggression

The most commonly used measure of aggres-sive behavior in the violent video game litera-ture is the modified Taylor competitive reactiontime test (TCRTT), in which the participant istold that he or she is competing with anotherparticipant (confederate) to see who can push abutton faster upon the appearance of a cue.After each trial, the winner chooses the intensityand duration of a punishment for the loser (suchas a loud noise blast). The level of punishmentintensity and duration that the participant setsfor his or her opponent are indicative of aggres-sive behavior.

The first problem with the modified TCRTTis that the participant’s motivation to behaveaggressively is ambiguous (see Adachi & Wil-loughby, 2011). Because aggression refers tobehavior that is intended to harm another indi-vidual, it is unclear whether participants viewtheir behavior as competitive instead of aggres-sive, in that participants’ motivation to giveintense punishments may be to slow their op-ponents’ response time on subsequent trials,thus allowing participants to win the competi-tion (Lieberman et al., 1999). Furthermore, be-cause violent games generally involve morecompetition than nonviolent games, violentvideo games may prime competitive schemasmore than nonviolent video games, making thecompetitive element of the TCRTT especially

salient. The second problem with the modifiedTCRTT is that it has been shown to lack validityas a measure of aggressive behavior. Fergusonand Rueda (2009) found that both intensity andduration scores for the modified TCRTT werenot related to paper-and-pencil measures of traitaggression, domestic violence, or violent crim-inal.

To assess direct and unambiguous aggressivebehavior, Lieberman et al. (1999) created theHot Sauce Paradigm. In the Hot Sauce Para-digm, the participant is given an already com-pleted food preference questionnaire and toldthat another participant down the hall has com-pleted this questionnaire and, as indicated bythe questionnaire, does not like hot or spicyfood. The participant is then given four bottlesof hot sauce ranked in terms of hotness and isinformed that his or her job is to choose one ofthe four bottles and mix up some hot sauce forthe other participant to drink. The amount ofhot sauce given and the degree of hotness isindicative of overt aggressive behavior (Bar-lett et al., 2009), and there are no competitivebenefits gained from administering a hottersauce to the confederate. Furthermore,Lieberman et al. found that scores on thisparadigm were positively related to both traitand physical aggression scores on the Bussand Perry (1992) Aggression Questionnaire,supporting the convergent validity of the HotSauce Paradigm, although to date no studyhas measured its association with aggressivebehavior outside the lab.

Aggression-Related Video GameCharacteristics

Research regarding the effects of video gamedifficulty and pace of action on aggressive be-havior is scarce. In contrast, competition hasbeen shown to be related to aggression. Forexample, Anderson and Morrow (1995) exam-ined whether giving participants competitiveversus cooperative instructions led to differ-ences in how aggressively they played the videogame Super Mario Brothers. In this game, var-ious creatures try to attack the main character(Mario or Luigi), and the main character in turnhas the option to either attack the creatures ortry to avoid them. Participants played the gamein pairs, and each participant took turns playingthe game. In the competitive condition, partici-

261VIDEO GAME COMPETITION AND VIOLENCE

This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

pants were told that their goal was to get furtherin the game than the other participant, and theparticipants each used a different character (Ma-rio or Luigi). In the cooperative condition, par-ticipants were told that their goal was to get asfar in the game as possible together, and theytook turns using the same character (Mario).Anderson and Morrow (1995) found that par-ticipants in the competitive condition killed sig-nificantly more enemy creatures than partici-pants in the cooperative condition. Thus, theyconcluded that competition elevated aggressivebehavior compared with cooperation (see alsoWilliams & Clippinger, 2002).

Although Anderson and Morrow (1995) useda video game that is competitive in nature, thegame is not very violent (also true for Williams& Clippinger, 2002). Specifically, the violenceis cartoonish and unrealistic. Schmierbach(2010) addressed this gap in the literature byexamining the effect of competitive video gameplay on aggressive cognition using the violentfirst-person shooter video game Halo. In Halo,the main character must compete in a battle forsurvival with the other opponent characters us-ing a variety of different guns. Schmierbachrandomly assigned pairs of participants to oneof three conditions: (a) participants played thegame on their own against computer opponents(solo mode), (b) participants played againsteach other in a one-on-one battle (competitivemode), and (c) participants played on the sameteam against computer opponents (cooperativemode). Immediately after video game play, par-ticipants completed a word completion task toassess aggressive cognition. The results showedthat participants in the competitive conditionhad the highest aggressive cognition scores, fol-lowed by participants in the solo condition,while participants in the cooperative conditionhad the lowest aggressive cognition scores.Thus, consistent with past research, it appearsthat competition in a video game elevates ag-gression compared with cooperation.

The Current Studies

The purpose of Pilot Study 1 was to testwhether two games chosen through pilot testingdiffered in terms of violence, but were matchedin terms of competitiveness, difficulty, and paceof action. The two matched games were thenused in Experiment 1 to test the effect of video

game violence on aggressive behavior (the HotSauce Paradigm). Pilot Study 2 was conductedto test whether four video games chosenthrough pilot testing were matched on difficultyand pace of action, but differed on violence andcompetitiveness, such that two of the gameswere equally violent but one was more compet-itive than the other, while the other two gameswere equally nonviolent but one was more com-petitive than the other. These four games werethen used in Experiment 2 to test the effect ofvideo game competition on aggressive behav-ior, and whether there was an interaction be-tween competition and violence.

Pilot Study 1

The goal of Pilot Study 1 was to test whethera violent action video game called Conan(THQ, 2007) and a nonviolent racing videogame called Fuel (Codemasters, 2009) werematched on game characteristics, but differed inviolence. The goal was then to use these gamesin Experiment 1. Conan is a violent game inwhich the main character must compete in abattle for survival using swords and axes againstthe opponent characters in order to progressthrough the levels. Fuel is a nonviolent racinggame in which the main character must competein several different races using vehicles such asmotorcycles and ATVs.

Fourteen introductory psychology studentsfrom a midsized university in Ontario, Canada(6 men, 8 women; M age � 20 years, 1 month)played the violent video game Conan and thenonviolent video game Fuel for 12 min each ina counterbalanced order. This study was ap-proved by the University Ethics board, and allparticipants provided active consent before par-ticipation. The games were played on an XBOX360 gaming system and 42-inch TV screen. Inorder to accurately compare participants’ rat-ings of game characteristics, we controlled fortheir previous experience with the relevantvideo game genres of action (Conan) and racing(Fuel). Participants also completed a demo-graphic questionnaire that assessed age and gen-der.

Participants rated each game in terms of thefour video game characteristics. Violence wasmeasured on a scale from 1 (very low) to 7 (veryhigh) by asking “how violent was the game.”Difficulty was measured by asking “how diffi-


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

cult was the game” on a scale of 1 (very low)to 7 (very high). Pace of action was assessed ona scale from 1 (very slow) to 7 (very fast) byasking “how was the pace of the game.” Com-petitiveness was assessed on a scale from 1(very low) to 7 (very high) using Anderson andCarnagey’s (2009) four questionnaire items: “towhat extent did this video game involve com-petition,” “to what extent did you feel like youwere competing with your opponents (i.e., in abattle or in a race),” “how competitive was thisvideo game,” and “how hard were you trying towin the game/match/contest.” Alphas for thecompetitiveness scale were acceptable for bothConan (� � .86) and Fuel (� � .78).

Participants’ experience with the two genresof games (action and racing) did not signifi-cantly differ, F(1, 12) � 3.41, p � .05, partial�2 � .22. A repeated measures analysis of vari-ance (ANOVA) was then conducted to comparethe two video games on the four video gamecharacteristics of violence, competitiveness,difficulty, and pace of action. Gender was in-cluded as a between-subjects variable, and ex-perience with racing and action games wereentered as covariates. Only the Type of game(i.e., Conan and Fuel) � Game characteristics(i.e., violence, competitiveness, difficulty, andpace of action) interaction was significant, F(3,30) � 7.59, p � .01, partial �2 � .43. To assessthis interaction, follow-up analyses were con-ducted to compare each of the four video gamecharacteristics between the two games (see Ta-ble 1 for mean ratings). Conan was rated assignificantly more violent than Fuel. However,the two games did not differ on competitive-ness, difficulty, or pace of action. Consequently,

Conan and Fuel were used in Experiment 1 totest whether video game violence alone couldproduce elevations in aggressive behavior.

Experiment 1

The purpose of Experiment 1 was to testwhether Conan and Fuel produced differencesin aggressive behavior when using an unambig-uous measure of aggressive behavior, the HotSauce Paradigm. Similar to Barlett et al. (2009),deception was used to disguise the purpose ofthe Hot Sauce Paradigm so that participantswould not be aware that we were assessing theeffect of violent video games on aggressivebehavior. Specifically, participants were toldthat they were participating in two unrelatedstudies looking at (a) video game play and eye-gaze and (b) food preference and personality(the Hot Sauce Paradigm). In terms of expecta-tions: (a) No specific hypothesis was made as towhether there would be a main effect of game.That is, it was not clear whether differences inhot sauce scores (i.e., standardized summationof the amount of sauce and the degree of hot-ness) would be found between the two videogames, as this was the first study to match aviolent and nonviolent game on competitive-ness, difficulty, and pace of action; (b) Wepredicted a main effect for gender, with menexpected to have higher hot sauce scores thanwomen, consistent with past literature showingthat men are more aggressive than women (e.g.,Coie & Dodge, 1998); (c) We did not expect tofind an interaction between game and gender aspast research has shown that the relation be-tween video game play and aggression does not

Table 1Pilot Study 1 and Experiment 1 Mean Ratings of Video Game Characteristics for Conan and Fuel

Pilot Study 1 Experiment 1

Game rating

Video game

Fa Partial �2

Video game

Fb Partial �2Violent Nonviolent Violent Nonviolent

Violence 5.36 (1.28) 1.50 (0.52) 19.31�� 0.66 5.14 (1.35) 2.05 (1.20) 61.75�� 0.63Competition 5.07 (1.23) 5.54 (1.30) 0.21 0.02 5.00 (0.99) 5.19 (1.19) 0.15 0.00Difficulty 3.93 (1.07) 3.71 (1.44) 0.35 0.03 3.00 (1.14) 3.48 (1.17) 2.54 0.07Pace of action 5.07 (0.83) 4.93 (1.21) 0.64 0.06 4.74 (1.08) 5.00 (0.89) 2.56 0.07

Note. N � 14 for Pilot study, and N � 42 for Experiment 1. Means are unadjusted; SDs are in parentheses. Violent �Conan, Nonviolent � Fuel.a df � 1, 10. b df � 1, 34.��p � .01. �� p � .001.


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

differ for men and women (e.g., Anderson et al.,2010); and (d) We also included a measure oftrait aggression at the end of the study to test theconvergent validity of the Hot Sauce Paradigm.It was hypothesized that hot sauce scores wouldbe moderately related to scores on a trait ag-gression questionnaire, consistent with past re-search (e.g., Lieberman et al., 1999, r � .30;Ferguson & Rueda, 2009, r � .25).

Method

Participants. Participants consisted of 42introductory psychology students from thesame university as in Experiment 1 (25men, 17 women; M age � 18 years 6 months).There were 21 participants that played Conan(13 men and 8 women), and 21 that playedFuel (12 men, 9 women). This study was ap-proved by the University Ethics board, and allparticipants provided active consent before par-ticipation. Students were recruited using thepsychology participant pool and earned coursecredit in exchange for their participation.

Materials.Demographics. As in Pilot Study 1, a de-

mographic questionnaire was used to assessage, gender, and experience with action andracing games. Participants indicated how manyhours per weekday and weekend (1 � not at allto 5 �5 or more hours per day) that they playedaction and racing games.

Video games and equipment. Conan andFuel were played using an XBOX 360 consoleon a 42-inch TV screen.

Aggressive behavior. The Hot Sauce Par-adigm (Lieberman et al., 1999) was used tomeasure overt aggressive behavior. Participantswere asked to prepare some hot sauce for an-other participant to drink who does not like hotsauce (note that there actually was no otherparticipant). Participants were asked to choosethe intensity of hot sauce (ranging from 1 �least hot to 4� most hot) and the amount,knowing that the other participant had to drinkwhatever amount was in the cup. Participantscould also taste the sauce in order to see howhot it was. Aggressive behavior was operation-alized as the sum of the standardized number(i.e., hotness rating) of sauce selected and theweight in grams (Barlett et al., 2009). Hot saucewas purchased from a local food establishmentthat has a ranked system of hotness for the

sauces and four sauces were selected thatranked in order from least to most hot. Eachsauce was transferred into a plastic squeezebottle and was given a number from one to four(with one being the least hot and four being thehottest). In addition, a styrofoam cup to placethe hot sauce into, a cup of water, popsiclesticks, and white bread (to help ease the hotsensation after tasting the sauce) were used.

Ratings of the video game characteristics.See Pilot Study 1 for a description. Alphas forthe competitiveness scale were acceptable forboth Conan (� � .74) and Fuel (� � .89).

Food preference. Food preference was as-sessed by asking “how much do you LIKE thefollowing kinds of foods” for six items (i.e.,sweet, savory, spicy, hot, bland, and saltyfoods) on a 1 (not at all) to 5 (extremely) scale(Barlett et al., 2009). Consistent with Barlett etal., we confirmed that both the degree of hotnessand the amount the hot sauce selected by par-ticipants was not a function of their liking hotfood. Liking hot food did not account for asignificant portion of the variability in the de-gree of hotness, R2 � .07, F(1, 40) � 2.78, p �.05, or the amount of sauce R2 � .04, F(1,40) � 1.44, p � .05.

Suspiciousness. Because the growing pop-ularity of research proposing a relation betweenviolent video games and aggression and the factthat deception was used, a suspiciousness ques-tionnaire was given that asked participantswhether they knew the true purpose of thestudy, whether anyone had told them about thestudy before completing it, and whether theywere aware of any deception (Barlett et al.,2009).

Trait aggression. The Buss-Perry Aggres-sion Questionnaire (Buss & Perry, 1992) wasused to measure trait aggression and to examinethe convergent validity of the Hot Sauce Para-digm. The scale consists of 29 items (e.g., “oncein a while, I can’t control the urge to strikeanother person”) and responses range from 1(strongly disagree) to 7 (strongly agree), withhigher scores indicating higher trait aggression.The internal consistency for this scale was good(� � .89).

Procedure. Participants were tested one ata time by the first author. First, they were toldthat they were participating in two unrelatedstudies: a video game study examining videogames and eye gaze and a study looking at


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

personality and food preference. For the foodstudy, participants were told that they were ran-domly assigned to the role of “food administra-tor” and that when the time came their jobwould be to prepare a certain type of food foranother participant who had been assigned tothe role of “food taster.” Participants were thengiven the demographic questionnaire followedby the food preference questionnaire.

Next, participants were told that they weregoing to begin the video game and eye-gazestudy. Participants were randomly assigned toplay either the violent or nonviolent videogame for 12 min while wearing a piece ofeye-tracker headgear which they believed wasmeasuring their eye-gaze, although we did notactually record their eye-gaze. Upon comple-tion, participants were then given the question-naire assessing the video game characteristicsand were then told it was time to complete thefood preference study. Specifically, the experi-menter explained that it was time for the par-ticipant to prepare some food for the food taster.Participants were given an already completedfood preference questionnaire and were told thatthe food taster completed this questionnaire.The food preference questionnaire clearly indi-cated that the food taster did not like hot orspicy food.

The experimenter then provided the partici-pant with the materials for the food preferencestudy (i.e., the hot sauce, a cup, a cup of water,bread, and popsicle sticks) and explained thatthe participant could choose the intensity of hotsauce (ranging from 1 � least hot to 4 � mosthot) and the amount, and that the food tasterwould have to drink whatever amount was inthe cup. As in Barlett et al. (2009), participantswere told that they could not mix sauces. Also,if they wished to know how hot the sauces werebefore choosing one, they could sample thesauces using the popsicle stick. The experi-menter left the room and watched from behinda two-way mirror as the task was performed,and then returned to retrieve the cup of hotsauce to allegedly bring to the food taster. Thetime lapse between the video game play and hotsauce preparation was 2 to 3 min, well withinthe 5- to 10-min time frame in which the effectof violent video games on aggression has beenshown to last (Barlett et al., 2009). Finally,participants completed the suspiciousness ques-tionnaire to assess whether participants knew

the true purpose of the study before being de-briefed or whether they were aware of any de-ception. Furthermore, 26 participants (13 whoplayed Conan and 13 who played Fuel) com-pleted the Buss and Perry (1992) Trait Aggres-sion Questionnaire to examine the convergentvalidity of the Hot Sauce Paradigm. The timelapse between the video game play and thecompletion of the trait aggression questionnairewas over 10 min (M � 11 min).

Results and Discussion

Suspiciousness. We originally had 48 par-ticipants, but six participants indicated that theyknew the true purpose of the study or wereaware of the deception and thus, their data werenot included in the analysis (final N � 42).

Experience. In order to include past videogame experience for both genres of games ascovariates in the main analyses, we first deter-mined that participants’ experience with actionand racing games did not differ between the twogame conditions, F(1, 38) � .93, p � .05,partial �2 � .02.

Video game ratings. A multivariate anal-ysis of variance (MANOVA) was conducted toconfirm that Conan and Fuel were equated onthe three video game characteristics of compet-itiveness, difficulty, and pace of action, but dif-fered on violence, and experience with racingand action games were included as covariates.There was a main effect for game, F(4,33) � 28.98, p � .01, partial �2 � .78. Thevideo games differed only in ratings of vio-lence, as Conan was rated as more violent thanFuel (see Table 1 for mean ratings). Thus, con-sistent with Pilot Study 1, participants ratedConan as more violent than Fuel, but not sig-nificantly different in terms of competitiveness,difficulty, and pace of action.

Aggressive behavior. A univariateANOVA was conducted with the summation ofthe standardized amount of hot sauce given andthe standardized degree of hotness as the depen-dent variable, video game condition (i.e., vio-lent vs. nonviolent) and gender as the indepen-dent variables, and game experience for bothgenres of games as the covariates. Participantswho played Conan did not differ in hot saucescores (M � .09) compared with participantswho played Fuel (M � �.09), F(1, 36) � .00,p � .05, partial �2 � .00. As predicted, men


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

(M � .40) had higher hot sauce scores thanwomen (M � �.59), F(1, 36) � 6.34, p � .05,partial �2 � .15, and the interaction betweengame and gender is not significant, F(1, 36) �.23, p � .05, partial �2 � .01. The resultssuggest that the violent content alone was notsufficient to produce elevations in aggressivebehavior compared with a nonviolent videogame.

Convergent validity of the Hot Sauce Par-adigm. We decided to test the convergentvalidity of the hot sauce paradigm after thestudy had begun, so only the final 26 partici-pants completed the trait aggression question-naire. As predicted, hot sauce scores were pos-itively correlated with trait aggression (r � .32),although this correlation was not statisticallysignificant because of the small sample size.However, the size of the moderate correlation isconsistent with previous results (e.g., Ferguson& Rueda, 2009; Lieberman et al., 1999).

Experiment 1, therefore, demonstrated thatvideo game violence alone is not sufficient toproduce elevations in aggressive behavior in alab setting. Using an unambiguous measure ofaggressive behavior, participants did not havehigher hot sauce scores after playing a violentgame compared with a nonviolent game thatwas equated on competitiveness, difficulty andpace of action. This finding suggests that thelevel of violence in video games may be lessinfluential in elevating aggression than previ-ously believed. In addition, the present studyprovided support for the validity of the HotSauce Paradigm as a measure of aggressivebehavior, because of its positive relation with ameasure of trait aggression.

An alternative explanation for this findingmay be that neither video game elevated aggres-sive behavior from baseline because the gamesmay not have been sufficiently violent, compet-itive, difficult, or contained fast enough actionto influence aggression. To assess this hypoth-esis, we compared hot sauce scores for partici-pants who played Conan and Fuel with hotsauce scores from participants in Barlett et al.’s(2009) violent and nonviolent video game con-ditions. In terms of intensity, scores for bothConan (2.52) and Fuel (2.33) were larger com-pared with scores found in both violent (2.12)and nonviolent (1.76) games reported by Barlettet al. In terms of weight, scores for Conan(1.01) and Fuel (1.01) were larger compared

with the scores for Barlett et al.’s nonviolentvideo game (.60) and slightly smaller than thescores for the violent video game (1.27). Thus,hot sauce scores for Conan and Fuel were verysimilar to Barlett et al.’s violent video game,and larger than their nonviolent video game.The fact that Barlett et al.’s violent video gameproduced more aggressive behavior than theirnonviolent video game suggests that in ourstudy, Conan and Fuel likely elevated aggres-sive behavior from baseline.

Another possible criticism of the currentstudy is that the sample size may have been toosmall to find a significant effect for video gamecondition on aggressive behavior. However, apower analysis using G*Power 3.1.2 revealedthat with the current sample size of N � 42, wehad power of .755 to detect the significance ofan effect size equivalent to Barlett et al. (2009;partial �2 � .15). In addition, the effect size forgame in the current study was zero (partial �2 �.000), and thus increasing the sample size wouldnot have made the effect statistically significant.Given our finding that video game violencealone is not sufficient to produce elevations inaggressive behavior in a lab setting, the nextstep is to examine which video game character-istics have the largest impact on aggressive be-havior and how these characteristics interact.Although a few researchers have attempted tocontrol for the level of competitiveness whentesting the effect of video game violence onaggressive behavior, no one has examined theeffect of video game competition on aggressivebehavior. Thus, Pilot Study 2 and Experiment 2were conducted to test the effect of the level ofcompetitiveness in video games on aggressivebehavior.

Pilot Study 2

The purpose of Pilot Study 2 was to isolatecompetitiveness by matching four games ondifficulty and pace of action, and to systemati-cally control for violence, so that we could usethese games in Experiment 2 to examine theeffect of competitiveness on aggressive behav-ior. After extensive testing by the first author,four games were selected that appeared to bematched on difficulty and pace of action. Twoof the games appeared to be equally violent, andthe other two games appeared to be equallynonviolent. Of the two violent games, one was


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

more competitive than the other, and of the twononviolent games, one was more competitivethan the other. The two violent games that werechosen were Mortal Kombat versus DC Uni-verse (Midway Games, 2008) and Left 4 Dead 2(Valve, 2009). Mortal Kombat versus DC Uni-verse is a fighting game in which the maincharacter must battle another opponent charac-ter in hand-to-hand combat. The goal of thegame is to defeat the opponent character in a3-round fight so a new opponent can be faced.Because of the competitive nature of the game(i.e., three rounds of one-on-one combat), thecompetitive element of this game is quite sa-lient. Thus, Mortal Kombat versus DC Universewas hypothesized to be the more competitiveviolent game. Left 4 Dead 2 is a first-personshooter in which the main character must battlezombies using guns and other weapons. Al-though the main character must compete in abattle for survival with every other character inthe game, Left 4 Dead 2 was hypothesized to beless competitive than Mortal Kombat versus DCUniverse. Unlike most other first-person shoot-ers, the opponent characters in Left 4 Dead 2 arezombies, and hence they do not possess weap-ons. Consequently, instead of engaging in acompetitive shoot-out against other armed char-acters as in most first-person shooters, manyscenarios in Left 4 Dead 2 involve standing at adistance and shooting a barrage of chargingzombies.

The two nonviolent games were Fuel (Code-masters, 2009) and Marble Blast Ultra (Ga-rageGames, 2006). As previously described,Fuel is a racing game in which the main char-acter must compete against other characterswhile in a series of races while driving a varietyof vehicles such as motorcycles and ATVs. Be-cause of the competitive nature of the game,Fuel was hypothesized to be the competitivenonviolent game. In contrast, Marble Blast Ul-tra involves controlling a marble through a se-ries of labyrinth-like mazes as quickly as pos-sible. As there are no other characters in thegame with which to compete with, Marble BlastUltra was hypothesized to be less competitivethan Fuel. In terms of comparing the violent andnonviolent games, Fuel was hypothesized to beequally competitive to Mortal Kombat versusDC Universe, and more competitive than Left 4Dead 2. Marble Blast Ultra was hypothesizedto be less competitive than Mortal Kombat ver-

sus DC Universe and Left 4 Dead 2 (becauseLeft 4 Dead 2 does contain some competitionagainst opponent characters). The games wereplayed on an XBOX 360 gaming system and42-inch TV screen.

Nineteen undergraduate students from thesame university as in Experiment 1 (12 men, 7women; M age � 22 years 2 months) played thefour games for 10 min each in a counterbal-anced order and rated each game in terms of thefour characteristics after playing it. This studywas approved by the University Ethics board,and all participants provided active consent be-fore participation. Instead of using a 4-itemcomposite to assess competitiveness, only twoitems were used, “to what extent did this videogame involve competition” and “to what extentdid you feel like you were competing with youropponents (i.e., in a battle or in a race).” Theremaining two items, “how competitive was thisvideo game” and “how hard were you trying towin the game/match/contest,” were not usedbecause they did not discriminate between com-petitive and noncompetitive games. For exam-ple, after playing a noncompetitive game, aparticipant might report that he or she tried veryhard to win the game, even though there was nocompetition. Correlations for the two competi-tiveness items were acceptable for Mortal Kom-bat versus DC Universe (r � .73), Left 4 Dead2 (r � .77), Marble Blast Ultra (r � .57), andFuel (r � .64). As in Experiment 1, a demo-graphic questionnaire was used to assess age,gender, and past video game experience.

A repeated measures ANOVA revealed thatparticipants’ experience across the differentgenres did not significantly differ, F(3,51) � 1.32, p � .05 partial �2 � .07. A repeatedmeasures ANOVA was conducted to comparethe four video games on the four video gamecharacteristics of violence, competitiveness,difficulty, and pace of action. Gender was in-cluded as a between-subjects variable, and ex-perience with all game genres were entered ascovariates. Only the Type of game (i.e., Fuel,Left 4 Dead, Marble Blast Ultra, Mortal Kom-bat versus DC Universe) � Game characteris-tics (i.e., violence, competitiveness, difficulty,and pace of action) interaction was significant,F(9, 117) � 5.01, p � .01, partial �2 � .28 (seeTable 2 for mean ratings of the video gamecharacteristics for the four video games). Thus,participants’ ratings of video game characteris-


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

tics differed between the four video games. Fol-low-up comparisons indicated that the fourgames did not significantly differ in their ratingsof difficulty, F(3, 42) � .17, p � .05, partial�2 � .01, or in pace of action, F(3, 42) �.48, p � .05, partial �2 � .03. Thus, all fourgames were matched on difficulty and pace ofaction.

Follow-up comparisons were then conducted toexamine differences in violence ratings betweenthe four games. The two nonviolent games, Mar-ble Blast Ultra and Fuel, were not rated differ-ently in terms of violence, F(1, 16) � 2.24, p �.05, partial �2 � .12. Similarly, the two violentgames, Mortal Kombat versus DC Universe andLeft 4 Dead 2, did not differ in terms of violenceratings F(1, 16) � 2.00, p � .05, partial �2 � .11.It is important to note, however, Mortal Kombatversus DC Universe was significantly more vio-lent than both Fuel, F(1, 16) � 50.96, p � .01,partial �2 � .76, and Marble Blast Ultra, F(1,16) � 52.19, p � .01, partial �2 � .77. Similarly,Left 4 Dead 2 was significantly more violent thanboth Fuel, F(1, 16) � 71.73, p � .01, partial �2 �.82, and Marble Blast Ultra, F(1, 16) � 84.32,p � .01, partial �2 � .84. Therefore, as predicted,the two violent games (Mortal Kombat versus DCUniverse and Left 4 Dead 2) were equally violent,and both were more violent than the two nonvio-lent games (Fuel and Marble Blast Ultra). Inaddition, the two nonviolent games were bothequally nonviolent.

Follow-up analyses were then conducted to ex-amine differences in competitiveness ratings be-tween the four games. Fuel and Mortal Kombatversus DC Universe were the two most competi-tive games, and they did not differ in competitive-ness ratings F(1, 16) � 1.45, p � .05, partial �2 �.08. Fuel was more competitive than Marble BlastUltra, F(1, 16) � 80.01, p � .01, partial �2 � .83,and Left 4 Dead 2, F(1, 16) � 5.39, p � .05,partial �2 � .25. Similarly, Mortal Kombat versusDC Universe was more competitive than Left 4Dead 2, F(1, 16) � 7.12, p � .05, partial �2 �.31, and Marble Blast Ultra, F(1, 16) � 94.85,p � .01, partial �2 � .86. Finally, Left 4 Dead 2was more competitive than Marble Blast UltraF(1, 16) � 6.06, p � .05, partial �2 � .28.Thus, we confirmed that of the four games, twowere significantly more competitive than the othertwo.T

able

2P

ilot

Stud

y2

and

Exp

erim

ent

2M

ean

Rat

ings

ofV

ideo

Gam

eC

hara

cter

isti

csfo

rth

eF

our

Vid

eoG

ames

Gam

era

ting

Pilo

tSt

udy

2vi

deo

gam

eE

xper

imen

t2

vide

oga

me

C,

NV

C,

VL

C,

VN

C,

NV

C,

NV

C,

VL

C,

NV

NC

,N

V

Vio

lenc

e1.

52(0

.70)

b5.

37(0

.90)

a6.

42(0

.84)

a1.

00(0

.00)

b1.

33(0

.82)

b5.

20(0

.86)

a5.

87(1

.13)

a1.

07(0

.26)

b

Com

petit

iven

ess

5.86

(0.7

6)a

6.32

(0.5

8)a

3.18

(1.6

7)b

1.36

(0.4

4)c

5.40

(1.0

2)a

6.13

(0.9

3)a

3.10

(1.6

7)b

1.40

(0.6

9)c

Dif

ficul

ty3.

63(1

.38)

a4.

47(1

.58)

a4.

42(1

.74)

a4.

68(1

.16)

a3.

27(1

.10)

a3.

67(0

.82)

a3.

53(0

.92)

a3.

60(1

.18)

a

Pace

ofac

tion

4.74

(1.0

5)a

5.42

(0.9

0)a

5.11

(1.2

9)a

4.57

(1.2

6)a

4.73

(1.3

3)a

5.20

(0.7

7)a

4.87

(0.9

2)a

4.47

(0.9

9)a

Not

e.N

�19

for

Pilo

tSt

udy

2an

dN

�60

for

Exp

erim

ent

1.M

eans

are

unad

just

ed;

cells

with

ina

row

with

com

mon

subs

crip

tsfo

rth

epi

lot

stud

yan

dex

peri

men

tar

eno

tsi

gnifi

cant

lydi

ffer

ent

atth

e.0

5le

vel.

SDs

are

inpa

rent

hese

s.C

�co

mpe

titiv

e,N

C�

nonc

ompe

titiv

e,L

C�

less

com

petit

ive,

V�

viol

ent,

NV

�no

nvio

lent

,C,N

V�

Fue

l,C

,V

�M

orta

lK

omba

tve

rsus

DC

Uni

vers

e,L

C,

V�

Lef

t4

Dea

d2,

NC

,N

V�

Mar

ble

Bla

stU

ltra

.


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

Experiment 2

The purpose of Experiment 2 was to examinethe effect of video game competitiveness onaggressive behavior (using the Hot Sauce Par-adigm), as well as test whether competitivenessinteracts with violence to influence aggression,using the four video games from Pilot Study 2.Although previous research has attempted toequate video games on competitiveness (e.g.,Anderson & Carnagey, 2009; Carnagey & An-derson, 2005), this was the first experiment todirectly test the effect of video game competi-tion on aggressive behavior. Moreover, as pre-viously described, video game competitivenessmay influence aggressive behavior through themechanism of physiological arousal. In order tofurther examine the relation between competi-tiveness and aggressive behavior, we recordedparticipants’ heart rate at baseline and duringvideo game play to test whether video gamecompetitiveness would produce elevations inheart rate from baseline.

Four hypotheses were tested: (1) Consistentwith research showing a relation between videogame competition and aggression (e.g., Ander-son & Morrow, 1995; Williams & Clippinger,2002; Schmierbach, 2010), we expected thatthere would be a significant main effect for typeof game. Specifically, we hypothesized that thehighly competitive, violent video game, MortalKombat versus DC Universe, and the highlycompetitive nonviolent video game Fuel wouldproduce more aggressive behavior than the lesscompetitive, violent video game Left 4 Dead 2,and the less competitive, nonviolent game,Marble Blast Ultra. As the less competitive,violent video game Left 4 Dead 2 was rated asmoderately competitive in Pilot Study 2 (i.e.,mean score � 3.18/7), it was unclear whether itwould produce greater levels of aggressive be-havior than the less competitive, nonviolentgame Marble Blast Ultra. (2) It was predictedthat there would be a significant main effect forgender, such that men would give more of ahotter sauce than women, consistent with theresults of Experiment 1. However, we did notexpect to find an interaction between game andgender, again consistent with the results of Ex-periment 1.(3) Consistent with our predictionsregarding aggressive behavior, we expected asignificant main effect for type of game on heartrate scores. Specifically, we hypothesized that

Fuel and Mortal Kombat versus DC Universewould produce the greatest elevations in heartrate from baseline. This finding would be con-sistent with the suggestion that elevations inphysiological arousal may be one mechanismthrough which video game competitivenessmay influence aggressive behavior. Also con-sistent with predictions regarding aggressivebehavior, it was unclear whether the moderatelycompetitive, violent game Left 4 Dead 2 wouldproduce greater increases in heart rate frombaseline compared with the less competitive,nonviolent game Marble Blast Ultra.

Method

Participants. Participants consisted of 60introductory psychology students from the sameuniversity as in Pilot Study 2 (32 men, 28 wom-en; M age � 18 years 4 months). Participantswere randomly assigned to play one of the fourvideo games, and thus, each video game wasplayed by 15 participants (8 men and 7 women).This study was approved by the University Eth-ics board and all participants provided activeconsent before participation. Students were re-cruited using the psychology participant pooland earned course credit in exchange for theirparticipation.

Materials.Demographics. See Experiment 1 for a

Description.Video games and equipment. Fuel, Left 4

Dead 2, Marble Blast Ultra, and Mortal Kom-bat versus DC Universe were played using anXBOX 360 console on a 42-inch TV screen.

Aggressive behavior. See Experiment 1 fora description of the paradigm.

Ratings of the video game characteristics.See Pilot Study 2 for a description. Correlationsfor the two competitiveness items were accept-able for Fuel (r � .72), Left 4 Dead 2 (r � .59),Marble Blast Ultra (r � .54), and Mortal Kom-bat versus DC Universe (r � .67).

Food preference. See Experiment 1 for adescription. Consistent with Experiment 1, lik-ing hot food did not account for a significantportion of the variability in the degree of hot-ness R2 � .00, F(1, 58) � 0.01, p � .05, or theamount of sauce R2 � .00, F(1, 58) � 0.01, p �.05.

Suspiciousness. See Experiment 1 for adescription.


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

Trait aggression. See Experiment 1 for adescription (in the current sample, � � .84).

Heart rate. Electrocardiogram (ECG) re-cordings were collected during a 5-min restperiod at the beginning of the session (baseline),and throughout the video game session. ECGsignals were recorded from two electrodesplaced on the participants’ collarbone with ahardware gain of 1000. R-R (interbeat) intervalswere visually checked in the MindWare pro-gram and edited where necessary according toprinciples advocated by Berntson and Stowell(1998).

Procedure. The procedure was identical toExperiment 1, other than participants were toldthat in addition to assessing their eye-gaze, theirheart rate would be monitored for 5 min atbaseline as well as continuously during videogame play.

Results and Discussion

Suspiciousness. We originally had 65 par-ticipants, but 5 participants indicated that theyknew the true purpose of the study or wereaware of the deception and thus, their data werenot included in the analysis (final N � 60).

Experience. A repeated measuresMANOVA was conducted to examine whetherparticipants in the four video game conditionsdiffered in their experience with the four genresof games. Experience with the four genres didnot differ between video game conditions, F(9,156) � .46, p � .05, partial �2 � .03.

Video game ratings. A MANOVA wasconducted to confirm that Fuel, Left 4 Dead 2,Marble Blast Ultra, and Mortal Kombat versusDC Universe were equated on difficulty andpace of action, but differed on competitivenessand violence, and experience with the fourvideo game genres were included as covariates.There was a main effect for game, F(12,141) � 39.10, p � .01, partial �2 � .76. Thevideo games differed only on ratings of com-petitiveness, F(3, 48) � 54.02, p � .01, partial�2 � .77, and violence, F(3, 48) � 216.78, p �.01, partial �2 � .93. Follow-up analyses re-vealed that consistent with Pilot Study 2, MortalKombat versus DC Universe was rated as morecompetitive than Left 4 Dead 2 and MarbleBlast Ultra, but not significantly different thanFuel (see Table 2 for mean ratings of videogame characteristics). Similarly, Fuel was rated

as more competitive than Marble Blast Ultraand Left 4 Dead 2. Finally, Left 4 Dead 2 wasrated as more competitive than Marble BlastUltra. Thus, ratings of the video game charac-teristics in this experiment were consistent withPilot Study 2. Consequently, we were able toisolate the effect of video game competitivenesson aggressive behavior. In addition, we wereable to examine whether competitiveness andviolence interact to influence aggression.

Aggressive behavior. A univariateANOVA was conducted with the summation ofthe standardized amount of hot sauce given andthe standardized degree of hotness as the depen-dent variable, video game condition (fourgames) and gender as the independent variables,and experience with the four genres of games asthe covariates. There was a main effect forgame F(3, 48) � 7.04, p � .01, partial �2 � .31.Follow-up analyses revealed that participantswho played Fuel (M � .79) and Mortal Kombatversus DC Universe (M � .90) had significantlyhigher hot sauce scores (Figure 1) than partici-pants who played Marble Blast Ultra (M ��.91) and Left 4 Dead 2 (M � �.78). Hot saucescores did not differ between Fuel and MortalKombat versus DC Universe, or between Mar-ble Blast Ultra and Left 4 Dead 2. Althoughmen (M � .31) had higher hot sauce scores thanwomen (M � �.35), this difference did notreach significance F(3, 48) � 1.49, p � .05,partial �2 � .03, and the interaction betweengame and gender also was not significant, F(3,48) � .41, p � .05, partial �2 � .03.

Trait aggression. To confirm that randomassignment was successful in placing equallyaggressive participants in the four video gameconditions, an ANOVA was conducted withtrait aggression as the dependent variable, andgame and gender as the independent variables.There were no differences in trait aggressionscores among the four conditions, F(3, 52) �.06, p � .05, partial �2 � .00. Thus, the fourvideo game conditions contained equally ag-gressive participants.

Heart rate. A repeated measuresMANOVA was conducted to examine differ-ences in heart rate between baseline and videogame play among the four video game condi-tions. Game and gender were entered as thebetween-subjects variables, while heart rate(baseline and during video game) was entered


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

as the within-subjects variable. A significantheart rate x video game interaction was found,F(3, 48) � 12.76, p � .01, partial �2 � .44.Follow-up analyses were then conducted foreach video game condition to see which videogames produced elevations in heart rate frombaseline (Table 3). Fuel and Mortal Kombatversus DC Universe both led to significantelevations in heart rate from baseline, whileMarble Blast Ultra and Left 4 Dead 2 didnot.

In summary, Experiment 2 confirmed the hy-pothesis that the two most competitive games,Fuel and Mortal Kombat versus DC Universe,would produce greater aggressive behaviorscores than the less competitive games, MarbleBlast Ultra and Left 4 Dead 2. Also as expected,Fuel and Mortal Kombat versus DC Universedid not produce differences in aggressive be-havior. In addition, we found that a moderately

competitive, highly violent game (Left 4 Dead2) was not sufficient to elevate aggressive be-havior compared with a less competitive, non-violent game (Marble Blast Ultra). These find-ings suggest that the level of competitivenessin video games is an important factor in therelation between video games and aggressivebehavior, with highly competitive gamesleading to greater elevations in aggressionthan less competitive games. As expected,men gave more of a hotter sauce than women,although this difference did not reach signif-icance. Also as expected, there was no inter-action between game and gender when pre-dicting aggressive behavior. Finally, wefound that only the two highly competitivegames, Fuel and Mortal Kombat versus DCUniverse, elevated heart rate from baseline.This is consistent with the theory that physi-ological arousal may be one mechanism

-1-0.8-0.6-0.4-0.2

00.20.40.60.8

1

C, V C, NV LC, V NC, NV

Video game

Stan

dard

ized

sum

mat

ion

of th

e am

ount

of h

ot s

auce

an

d de

gree

of h

otne

ss

Figure 1. The effect of video game condition on aggressive behavior in Experiment 4. C �competitive, NC � noncompetitive, LC � Less competitive, V � violent, NV � nonviolent,C, V � Mortal Kombat versus DC Universe, C, NV � Fuel, LC, V � Left 4 Dead 2, NC,NV � Marble Blast Ultra.

Table 3Experiment 2 Baseline and Heart Rate Scores for the Four Video Games

Video game

Heart rate

Fa Partial �2Baseline Game

Competitive, nonviolent 76.92 (10.18) 78.95 (11.25) 6.07� 0.32Competitive, violent 74.89 (16.15) 86.36 (17.52) 31.00�� 0.71Less competitive, violent 78.85 (11.88) 77.32 ( 8.53) 0.78 0.06Non-competitive, nonviolent 75.81 (11.88) 76.62 (12.25) 0.72 0.05

Note. SDs are in parentheses. N � 60; Means are unadjusted. Competitive, nonviolent � Fuel; competitive, violent �Mortal Kombat versus DC Universe; less competitive, violent � Left 4 Dead 2; noncompetitive, nonviolent � Marble BlastUltra.a df � 1, 13.�p � .05. �� p � .01.


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

through which video game competitivenessmay influence aggressive behavior.

General Discussion

The present study was the first to isolate theviolent content in a video game by matching aviolent and nonviolent game on competitive-ness, difficulty, and pace of action (Pilot Study1). We then demonstrated that the violent con-tent alone was not sufficient to elevate aggres-sive behavior in the short-term (Experiment 1).This finding suggests that the level of violencein video games may be less influential in ele-vating aggression than previously believed.

After demonstrating that video game vio-lence alone was not sufficient to elevate aggres-sive behavior, we examined the effect of videogame competitiveness. The present study wasthe first to isolate video game competitivenessby matching two violent games, and two non-violent games on violence, difficulty, and paceof action (Pilot Study 2). We found that videogame competitiveness elevated aggressive be-havior in the short-term, regardless of the levelof violent content, as the two most competitivevideo games, Mortal Kombat versus DC Uni-verse (violent) and Fuel (nonviolent), producedthe greatest levels of aggressive behavior. Wealso found that a moderately competitive game(Left 4 Dead 2), even when paired with a highlevel of violence, was not sufficient to elevateaggressive behavior compared with a less com-petitive, nonviolent game (Marble Blast Ultra).In terms of mechanisms, according to the GAM,violent video games may influence aggressivebehavior through aggressive thoughts, feelings,and physiological arousal. Consistent with theGAM, we demonstrated that physiologicalarousal may be a mechanism through whichvideo game competitiveness influences aggres-sive behavior, as only Mortal Kombat versusDC Universe and Fuel produced elevations inheart rate from baseline.

Limitations

There were several limitations with thisstudy. First, the present study only used samplesof university students. Findings, however, maybe different for other age groups. For example,the relation between video game competitive-ness and aggression may be different for ado-

lescents (e.g., 12 to 19 years) versus adults(e.g., 25 years and older), because of the hy-pothesis that some adolescents may experiencea temporal gap between an early maturing so-cioemotional system (hypothesized to be a re-sult of increases in dopaminergic activity, per-haps linked to puberty, leading to increases inreward seeking, need for novelty and stimula-tion), and a slower maturing self-regulatory sys-tem (hypothesized to be led by the prefrontalcortex, responsible for planning, judgment, andinhibition, which may not be fully mature untilthe mid 20s; Steinberg, 2010). Thus, adoles-cents on average may be more likely to behaveaggressively than adults after playing a compet-itive video game because of their potentiallygreater difficulty (on average) in regulating theirarousal than adults. Future research shouldcompare the effect of video game competitive-ness on aggression between these different agegroups. Second, although this study addressedthe short-term effect of video game competitionon aggressive behavior, we did not examinelong-term effects. Thus, longitudinal researchexamining the relation between video gamecompetition and aggression is needed. Finally,findings may not generalize to other geographicregions, including those with differing ethnicand/or demographic mixes.

Research Implications

We have expanded on previous researchthat has found a relation between violentvideo games and aggression (see Anderson etal., 2010) by demonstrating that when isolat-ing specific video game characteristics, com-petitiveness had a much larger impact on ag-gressive behavior than the violent content. Atfirst glance, this finding may seem to contra-dict past research which has found that vio-lent video games produced more aggressionthan nonviolent video games. However, be-cause past studies have failed to equate theviolent and nonviolent video games on com-petitiveness, difficulty, and pace of action si-multaneously, researchers may have attrib-uted too much of the variability in aggressionto the violent content. For example, since vi-olent video games are more competitive in gen-eral than nonviolent games, it was likely thecompetition, rather than the violence, that wasresponsible for the elevations in aggression in


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

past studies. Furthermore, in the only study thatsuccessfully matched a violent and nonviolentvideo game on competitiveness (Anderson &Carnagey, 2009), the violent game was ratedhigher in terms of difficulty and pace of action.Thus, it was unclear whether the violence, dif-ficulty, pace of action, or a combination of thethree influenced aggressive behavior. Future re-search should test the effect of both video gamedifficulty and pace of action on aggressive be-havior, as well as how the four video gamecharacteristics interact. For example, a compet-itive game that is more difficult, in that peoplelose more often and must exert considerableeffort in order to succeed may be more likely toinfluence aggressive behavior. Thus, an interac-tion between competitiveness and difficultymay be related to elevated aggression.

Although we found that heart rate may beone mechanism through which video gamecompetition influences aggressive behavior,future research is needed to examine otherpossible mechanisms. Consistent with Ander-son and Morrow’s (1995) finding that compe-tition produces more aggressive thoughts thancooperation, it is likely that video game com-petitiveness influences aggressive thoughts.Similarly, Anderson and Morrow stated thatcompetition usually leads to negative feelingsas only one person can reach the desired goalof winning the game. Thus, video game com-petition likely produces feelings of frustra-tion, as opponents continuously attempt toobstruct each other’s goal of becoming victo-rious (Berkowitz, 1989).

In addition, research examining the effects ofvideo game competitiveness on aggression mayalso apply to other competitive situations, suchas sports. Unlike video games, many sportscontain physical contact (e.g., football, hockey)and offer opportunities to behave aggressively,and even to become violent (e.g., fighting orunnecessary roughness). Thus, there is clearly aneed for a better understanding of the relationbetween competition and aggression, and videogames may be an excellent vehicle to investi-gate this relation.

Conclusion

Some researchers believe that they have al-ready shown that violent video games are a riskfactor for aggressive behavior (Anderson et al.,

2010) and that this effect stems from the violentcontent in the games (Anderson et al., 2004).On the contrary, results from the present studyindicate that video game competitiveness, notviolent content, is responsible for elevating ag-gressive behavior in the short-term. The presentfindings lead to a new direction in video gameand aggression research and should encourageresearchers to continue to critically examine thisissue.

References

Adachi, P. J. C., & Willoughby, T. (2011). The effectof violent video games on aggression: Is it morethan just the violence? Aggression and ViolentBehavior, 16, 55– 62. doi:10.1016/j.avb.2010.12.002

Anderson, C. A., & Bushman, B. J. (2002). Humanaggression. Annual Review of Psychology, 53, 27–51.

Anderson, C. A., & Carnagey, N. L. (2004). Violentevil and the general aggression model. In A. Miller(Ed.), The social psychology of good and evil (pp.168–192). New York, NY: Guilford Press.

Anderson, C. A., & Carnagey, N. L. (2009). Causaleffects of violent sports video games on aggres-sion: Is it competitiveness or violent content?Journal of Experimental Social Psychology, 45,731–739. doi:10.1016/j.jesp.2009.04.019

Anderson, C. A., Carnagey, N. L., Flanagan, M.,Benjamin, A. J., Eubanks, J., & Valentine, J. C.(2004). Violent video games: Specific effects ofviolent content on aggressive thoughts and behav-ior. Advances in Experimental Social Psychol-ogy, 36, 199–249.

Anderson, C. A., Gentile, D. A., & Buckley, K. E.(2007). Violent video game effects. New York,NY: Oxford University Press.

Anderson, C. A., Ihori, N., Bushman, B. J., Rothse-tin, H., R., Shibuya, A., Swing, E. L., . . . Saleem,M. (2010). Video game effects on aggression, em-pathy, and prosocial behavior and eastern andwestern countries: A meta-analytic review. Psy-chological Bulletin, 136, 151–173. doi:10.1037/a0018251

Anderson, C. A., & Morrow, M. (1995). Competitiveaggression without interaction: Effects of compet-itive versus cooperative instructions on aggressivebehavior in video games. Personality and SocialPsychology Bulletin, 21, 1020–1030.

Barlett, C. P., Branch, O., Rodeheffer, C., & Harris,R. (2009). How long do the short-term violentvideo game effects last? Aggressive Behavior, 35,1–12. doi:10.1002/ab.20301


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

Berkowitz, L. (1989). Frustration-aggression hypoth-esis: Examination and reformulation. Psychologi-cal Bulletin, 106, 59–73.

Berntson, G. G., & Stowell, J. (1998). ECG artifactsand heart period variability: Don’t miss a beat!Psychophysiology, 35, 127–132.

Bushman, B. J., & Anderson, C. A. (2002). Violentvideo games and hostile expectations: A test of thegeneral aggression model. Personality and SocialPsychology Bulletin, 28, 1679–1686. doi:10.1177/014616702237649

Buss, A. H., & Perry, M. (1992). The aggressionquestionnaire. Journal of Personality and SocialPsychology, 63, 452–459.

Carnagey, L. N., & Anderson, C. A. (2005). Theeffects of reward and punishment in violent videogames on aggressive affect, cognition, and behav-ior. Psychological Science, 16, 882–889.

Coie, J. D., & Dodge, K. A. (1998). Aggression andantisocial behavior. In W. Damon & N. Eisenberg(Eds.), Handbook of child psychology (5th ed.,Vol. 3, pp. 779–862). New York, NY: Wiley.

Ferguson, C. J., & Kilburn, J. (2010). Much adoabout nothing: The misestimation and overinter-pretation of violent video game effects in easternand western nations: Comment on Anderson et al.(2010). Psychological Bulletin, 136, 174–178. doi:10.1037/a0018566

Ferguson, C. J., & Rueda, S. M. (2009). Examiningthe validity of the modified Taylor competitivereaction time test of aggression. Journal of Exper-imental Criminology, 5, 121–137. doi:10.1007/s11292-009-9069-5

Ferguson, C. J., & Rueda, S. M. (2010). The hitmanstudy: Violent video game exposure effects onaggressive behavior, hostile feelings, and depres-

sion. European Psychologist, 15, 99–108. doi:10.1027/1016-9040/a000010

Ferguson, C. J., Rueda, S. M., Cruz, A. M., Ferguson,D. E., Fritz, S., & Smith, S. M. (2008). Violentvideo games and aggression: Byproduct of familyviolence and intrinsic violence motivation? Crim-inal Justice and Behavior, 35, 311–332. doi:10.1177/0093854807311719

Gentile, D. A. (2009). Pathological video-game useamong youth ages 8 to 18: A national study. Psy-chological Science, 20, 594–602. doi:10.1111/j.1467-9280.2009.02340.x

Lieberman, J. D., Soloman, S., Greenberg, J., &McGregor, H. A. (1999). A hot new way to mea-sure aggression: Hot sauce allocation. AggressiveBehavior, 25, 331–348.

Schmierbach, M. (2010). “Killing spree”: Exploring theconnection between competitive game play and aggres-sive cognition. Communication Research, 37, 256–274.doi:10.1177/0093650209356394

Steinberg, L. (2010). A dual systems model of ado-lescent risk taking. Developmental Psychobiol-ogy, 52, 216–224.

Williams, D., & Skoric, M. (2005). Internet fantasyviolence: A test of aggression in an online game.Communication Monographs, 72, 217–233. doi:10.1080/03637750500111781

Williams, R. B., & Clippinger, C. A. (2002). Aggres-sion, competition and computer games: Computerand human opponents. Computer in Human Be-havior, 18, 495–506.

Received February 18, 2011Revision received June 15, 2011

Accepted June 16, 2011 �


This

doc

umen

t is c

opyr

ight

ed b

y th

e A

mer

ican

Psy

chol

ogic

al A

ssoc

iatio

n or

one

of i

ts a

llied

pub

lishe

rs.

This

arti

cle

is in

tend

ed so

lely

for t

he p

erso

nal u

se o

f the

indi

vidu

al u

ser a

nd is

not

to b

e di

ssem

inat

ed b

road

ly.

the effect of video game competition and violence on aggressive behavior: which characteristic has...

Documents