THIS ARTICLE HAS BEEN CORRECTED.SEE LAST PAGE PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES Competence-Impeding Electronic Games and Players'Aggressive Feelings Thoughts,and Behaviors Usny众 tahaPd RatdM points.In e aioetbeo s serve for ndicated that mpetenc Keyrd motivation,competence.aggression,electronic games and researchers alike ha an 2001)wher thers find human aggression,proposing a multitude of possibilities.Diverse inconscquential links (e.g Elson Ferguson.in s suc as ger make-up(Fe 0).social pres .0u Deuse .&DeNeve.1995)have bee nent as a source of aggression.Indeed. ncers that some sing on gaming Motivational Perspective me SDT theorizes that human aggression and inter onal violenc ned c effects of gaming on aggression:Some researchers report finding consistent links between some forms of gaming and measures of three basic psychological needs that are tosuppo ces:Ih ceds for (ie.the mher302013 wK. of Cli cal and S ocial Scie Oxford.England:Ed vard L.D nt of Clinical and S nd provide the i Grolnick.&La Guardia.2006). 4
PERSONALITY PROCESSES AND INDIVIDUAL DIFFERENCES Competence-Impeding Electronic Games and Players’ Aggressive Feelings, Thoughts, and Behaviors Andrew K. Przybylski University of Rochester and University of Oxford Edward L. Deci University of Rochester C. Scott Rigby Immersyve, Celebration, Florida Richard M. Ryan University of Rochester Recent studies have examined whether electronic games foster aggression. At present, the extent to which games contribute to aggression and the mechanisms through which such links may exist are hotly debated points. In current research we tested a motivational hypothesis derived from self-determination theory—that gaming would be associated with indicators of human aggression to the degree that the interactive elements of games serve to impede players’ fundamental psychological need for competence. Seven studies, using multiple methods to manipulate player competence and a range of approaches for evaluating aggression, indicated that competence-impeding play led to higher levels of aggressive feelings, easier access to aggressive thoughts, and a greater likelihood of enacting aggressive behavior. Results indicated that player perceived competence was positively related to gaming motivation, a factor that was, in turn, negatively associated with player aggression. Overall, this pattern of effects was found to be independent of the presence or absence of violent game contents. We discuss the results in respect to research focused on psychological need frustration and satisfaction and as they regard gaming-related aggression literature. Keywords: motivation, competence, aggression, electronic games Philosophers and researchers alike have sought the causes of human aggression, proposing a multitude of possibilities. Diverse factors such as genetic make-up (Ferguson, 2010), social learning (Bandura, 1977), goal frustration (Miller, 1941), and ambient air temperature (Anderson, Deuser, & DeNeve, 1995) have been evaluated as potential sources of aggressive thoughts, feelings, and behaviors. More recently however, popular interest has turned to electronic game entertainment as a source of aggression. Indeed, the great majority of social science research focusing on gaming environments examines concerns that some games—violent ones in particular— have the potential to foment aggression. Despite a growing body of research, there is no clear consensus regarding the effects of gaming on aggression: Some researchers report finding consistent links between some forms of gaming and measures of aggression (e.g., Anderson & Bushman, 2001), whereas others find unreliable or inconsequential links (e.g., Elson & Ferguson, in press). Our aim in the present research was to empirically explore gaming-related aggression through the motivational lens of selfdetermination theory (SDT; Ryan & Deci, 2000)— by systematically investigating psychological need thwarting in gaming as a source of aggressive feelings, thoughts, and behaviors. Motivational Perspective SDT theorizes that human aggression and interpersonal violence result from the threatened or actual thwarting of basic psychological needs (Deci & Ryan, 2012; Ryan & Deci, 2000). SDT posits three basic psychological needs that are sensitive to supports and susceptible to impedance from environmental circumstances: The needs for competence (i.e., the experience of efficacy), autonomy (i.e., the sense of choice and volition), and relatedness (i.e., the feeling of connection and belongingness with others). When supported, these three needs form the basis of psychological health and provide the necessary and sufficient conditions for effective self-regulation. SDT researchers have argued that people are more prone to aggression when any of these three basic needs is thwarted either proximally, by situational threats or deprivations, or distally, by chronic developmental conditions (Ryan, Deci, Grolnick, & La Guardia, 2006). Several lines of research provide support for the idea that psychological need thwarting can contribute to increased aggresThis article was published Online First December 30, 2013. Andrew K. Przybylski, Department of Clinical and Social Sciences, University of Rochester, and Oxford Internet Institute, University of Oxford, Oxford, England; Edward L. Deci, Department of Clinical and Social Sciences, University of Rochester; C. Scott Rigby, Immersyve, Celebration, Florida; Richard M. Ryan, Department of Clinical and Social Sciences, University of Rochester. Correspondence concerning this article should be addressed to Andrew K. Przybylski, who is now at the Oxford Internet Institute, University of Oxford, 1 St Giles’, Oxford OX1 3JS, England. E-mail: andy.przybylski@ oii.ox.ac.uk This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. Journal of Personality and Social Psychology, 2014, Vol. 106, No. 3, 441– 457 © 2013 American Psychological Association 0022-3514/14/$12.00 DOI: 10.1037/a0034820 441 THIS ARTICLE HAS BEEN CORRECTED. SEE LAST PAGE
PRZYBYLSKL DECL RIGBY.AND RYAN basic a factor that as people ed for auto More recently,Roth,Kana aid.the motivation-foc pproach based in SDT differs widely 201 from how gaming-related aggression is typically studied ul as well as The Gaming-Related Aggression Literature s0 hlocking led to in The primary theoretical framework used to understand the links ms of GAM is a social c onitive framework in ded to model th cte late unde ned the need for relatedne 977 )the GAM argue that incidental s hat act and ple sive f of ial cont Take t for the idea This kind d the ognitiv heightening aggressiv ting n as me of competence-need thwarting in a novel context iolkentcontcantintoditio al.passive Motivational Perspective and Electronie Gaming GAM games,which othe ng.GAM-omair dies of d to rew onducte using one of and ated sm anicular.mar rovide ar A Bu y a 985) aple.the graded challe nges and det ailed which e h compu supp on and of gar olay and nay inflate ffec g the need Supportin and is quite onstratine bow in-eam e need satisfaction child relatio quality (Wa Adac rm shifts i vers'well-b The of vi xposure (e.g von Salis lence acro 等a wide ve te nt and quent research by Przybylski.Ryan.and Rigby (2009 I and prospective open th ed tha nities for ot inherently mot sive.Ins nted for the lion's share of variability in viol t eam effects of violent eamine on pla ggressive feelings and work of yan,Rigby an the labo Like the cros ational studies ha hown in nen ki et al. who played (e.g.. n&Di.200
sion. Early work by Ryan and Grolnick (1986) showed that children had higher levels of implicit aggressiveness when their basic need for autonomy was impeded. More recently, Roth, KanatMaymon, and Bibi (2011) reported that classroom environments that were autonomy-need thwarting were more likely to foster bullying as well as interpersonal aggression. Weinstein, Hodgins, and Ostvik-White (2011) also showed that the salience of autonomy-blocking led to increased accessibility of aggressive thoughts, which in turn resulted in people enjoying hostile forms of humor. Neighbors, Vietor, and Knee (2002) reported that needthwarting motivation predicted experiencing pressure and egodefensiveness while driving, a shift that in turn related to aggressive driving and road rage. Weinstein (2009) showed that those situations that undermined the need for relatedness served to increase interpersonal aggression, an effect that actually carried over from a prior environment to bias social behavior and introduce aggressive feelings into a new social context. Taken together, these studies provide preliminary support for the idea that psychological need thwarting can lead to aggressiveness. To date, the effects of competence-need deprivation have not been extensively explored from an SDT perspective. The present article builds upon the prior research by systematically examining aggression as an outcome of competence-need thwarting in a novel context. Motivational Perspective and Electronic Gaming Unlike many other pursuits, most people are not typically motivated to play electronic games for extrinsic rewards such as money or fame; instead such games are pursued because they provide inherent satisfactions. Indeed, according to SDT gaming is typically intrinsically motivated and provides individuals with rich opportunities for psychological need satisfactions (Ryan, Rigby, & Przybylski, 2006). In particular, many games provide ample competence need satisfactions—feelings of efficacy and skill growth, which are fundamental to intrinsic motivation (Deci & Ryan, 1985). For example, the graded challenges and detailed, unambiguous performance feedback in computer games can support satisfaction of the need for competence. Similarly, wide-ranging opportunities for exploration and action allow for satisfaction of the need for autonomy. Finally, the affordance of communication channels and group challenges emphasizing collaboration and competition can satisfy the need for relatedness. Supporting this idea, Ryan, Rigby, and Przybylski (2006) reported on a series of experiments demonstrating how in-game need satisfactions predicted both preferences for specific games, and game enjoyment. They also demonstrated that games that were need satisfying predicted positive short-term shifts in players’ well-being. These relations were in evidence across a wide range of games, varying in content and genre. Subsequent research by Przybylski, Ryan, and Rigby (2009) examined the motivational appeal of violent game content. They reported that opportunities for aggression in electronic games were not inherently motivating or necessarily more immersive. Instead, basic need supports, for competence and autonomy in particular, accounted for the lion’s share of variability in violent game appeal and immersion. Like the earlier work of Ryan, Rigby, and Przybylski (2006), this research underscored the predictive utility of understanding games from a motivational vantage point. Przybylski et al. (2009) highlighted the extent to which games satisfied basic needs, a factor that was linked to how people engaged in games, independent of their particular contents or genres. That said, the motivation-focused approach based in SDT differs widely from how gaming-related aggression is typically studied. The Gaming-Related Aggression Literature The primary theoretical framework used to understand the links between electronic gaming and human aggression is the general aggression model (GAM; Anderson, Deuser, & DeNeve, 1995). The GAM is a social cognitive framework intended to model the paths by which exposure to violent media influences aggressive thoughts and feelings. Based on social learning theory (Bandura, 1977), the GAM argues that incidental exposure to media featuring violence is arousing and pleasurable and thus increases the future probability of intentional violent media exposure. Acute incidents of intentional self-exposure to violent media lead to chronic selfexposure. This kind of chronic exposure to violent media snowballs by increasing the accessibility of aggressive thoughts and cognitive schemas, heightening aggressive feelings, and resulting in aggressive behavior (for a comprehensive review, see Lindsay & Anderson, 2000). Originally, the GAM was designed to model the effects of violent content in traditional, passive forms of media such as comics, movies, and music. More recently however, the GAM has been used to study violence in computer games, which in contrast is an interactive domain. Broadly speaking, GAM-based studies of gaming-related aggression have been conducted using one of three approaches: correlational studies, laboratory experiments, and longitudinal designs. A number of correlational and prospective studies have demonstrated small, yet consistent links from dispositional aggression, delinquency, and poor school performance to violent game engagement (e.g., Anderson, Gentile, & Buckley, 2007). These studies have used variants of the Violent Video Game Exposure (VVGE: Anderson & Dill, 2000) questionnaire to assess violent gaming levels, which has been somewhat controversial. Researchers have highlighted serious problems with the VVGE such as suspect validity because it does not accurately tap into everyday patterns of game play and may inflate effect-size estimates (Fikkers, Valkenburg, & Vossen, 2012). Further, an increasing number of recent prospective studies have indicated that the gamingaggression link is not conclusive and is quite small when other factors are considered (Elson & Ferguson, in press). Demographic variability, parent– child relationship quality (Wallenius & Punamäki, 2008), peer deviance (Willoughby, Adachi, & Good, 2012), consistent measures of game content (Ferguson, 2011), and expert ratings of violent game exposure (e.g., von Salisch, Vogelgesang, Kristen, & Oppl, 2011) serve to moderate most observed relations between aggression and electronic gaming. Because correlational and prospective studies leave open the causal direction of the relations—that is, whether violent play makes people somewhat more aggressive or dispositionally aggressive persons seek out violent games—an increasing number of studies have sought to address this issue of causality by examining the effects of violent gaming on players’ aggressive feelings and behaviors postengagement in the laboratory. Like the crosssectional (i.e., correlational) and prospective studies, a subset of these studies have shown increased postengagement aggression for those who played violent games (e.g., Anderson & Dill, 2000; This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 442 PRZYBYLSKI, DECI, RIGBY, AND RYAN
COMPETENCE IMPEDANCE AND AGGRESSION 443 .Ci 10R6 tronic game developer Given th ve dcba of dueling meta-analyse t but do not always s Like the selves thes cases (s Anders ctal,201 eror leaming and sustained effort on the players'part are relevant for a motivation- the c the mules and halance o or example in rev h oy a new game ter son (200 reported that some of the rovidemore n-depth int ing balanc n the g rea have emerged fro ever,this investment even aggre players with impo tant gratif ication h as ver amount of expenence.may differen ally inf indicat they do not strike this balance.Games that get this bala ed the desc ribed by critics as providing "a steep learning null i in tence-impeding g he h nto gam hat th Like 200120 scaled to a ific plaver's abiliti and the by the elings and behaviors.Fo exa arc appl vers ed p ult from sud n,high-in mpetitors. Given these pote ial pitfall same de signe from an SD of de or studying the c ioral market se ation,targeting out sponta Hard b gh SDT-bae skille cha ing games in this way. de opti ing in e3mi ssary for dit ms that hav em s of the to ed sing on its effects or be ne's control inter ce t out a in virtua from foc y on violent g ent as a predicto e in eno in eu Ma a have no of pla mportant role in game motivation as a necessary.but not suffi
Anderson & Murphy, 2003), whereas others have not found these effects (e.g., Cicchirillo & Chory-Assad, 2005; Kirsh, 1998; McClure & Mears, 1986; Williams & Skoric, 2005). Given this active debate, a number of dueling meta-analyses have recently been published on the growing number of contradicting experiments. Like the experiments themselves, these metaanalyses directly conflict in some cases (see Anderson et al., 2010, and Ferguson & Kilburn, 2010). That said, some findings from these summaries are relevant for a motivation-focused analysis of gaming-related aggression. For example, in reviewing a decade of experiments assessing links between aggression and computer games, Ferguson (2007) reported that some of the conflicting results in the gaming-related aggression area have emerged from intervening variables specific to interactive gaming contexts and unstandardized aggression measures. Similarly, Sherry’s 2001 and 2007 meta-analyses reported evidence that a number of experimental factors present in the experimental literature, such as player arousal and the amount of experience, may differentially influence the relations between violent-game exposure and player aggression. Sherry’s findings indicated that participants assigned to short periods of exposure showed the strongest link between gaming and aggression outcomes, whereas medium and longer-term periods of experimental exposure showed null and in some cases, reverse effects. Sherry speculated that the initial minutes of playing a video game may be highly stimulating— explaining larger effects for brief game exposure— but that this arousal might plateau and regress to the mean under conditions where play continues for longer time periods. Like Ferguson (2007), Sherry (2001, 2007) argued that the patterns of results demonstrated by their metaanalyses indicated the need to use new theoretical perspectives to study the gaming-aggression link. A primary objective of the present research was to apply SDT to investigate gaming-related aggression through a motivation-based lens, extending this complex literature. As reviewed previously, SDT research examining acute and chronic impedance of the basic psychological needs for autonomy and relatedness showed that deprivation of these supports result in increased aggression. In contrast, the effects of competence need deprivation on aggressive thoughts, feelings, and behavior have not been as extensively explored from an SDT viewpoint. Computer gaming, a domain where human competence is of paramount importance, provides an excellent context for studying the effects of competence-thwarting experiences. Further, as past SDT-based gaming research indicates, competence is central to the interactive experiences that make games appealing. Competence thwarting would thus seem to serve as an ideal candidate for evaluation as a source of gamingrelated aggression. Importantly, although SDT-based gaming research has focused on how virtual contexts can support needs and bolster well-being, no empirical work has examined the negative consequences of need thwarting in gaming. Indeed, many of the positive motivational mechanisms that have been studied in terms of supporting experiences of competence also can be reviewed in terms of their potential to impede competence. For example, Ryan, Rigby, & Przybylski (2006) demonstrated that mastery-of-controls—the learned ability to effortlessly use a game’s control interface to carry out actions in virtual environments—was essential to building an overall sense of competence in engaging in game play. Mastery-of-controls played an important role in game motivation as a necessary, but not sufficient, condition for achieving psychological need satisfying play. Electronic game developers aim to minimize the complexity of game controls in an attempt to hook players’ interest and excitement but do not always succeed. Despite the best efforts of game creators, the controls and rules of computer games are often quite elaborate and require significant trial and error learning and sustained effort on the players’ part. Often this means that players require a number of gaming sessions to master the controls and understand the rules, and balance of challenges presented by a new game. This is especially the case with games intended for more experienced players because they provide more in-depth interactive experiences. Regarding balance, however, this investment eventually pays off in the players’ favor. For example, complex games can offer more immersive experiences and provide players with important gratifications such as stress reduction (Reinecke, 2009) and opportunities to explore ideal self-aspects (Przybylski, Weinstein, Murayama, Lynch, & Ryan, 2012). Games succeed commercially when they strike the balance between complexity and competence support and fail when they do not strike this balance. Games that get this balance wrong are described by critics as providing “a steep learning curve.” Competence-impeding gaming experiences have the potential to aggravate and demotivate players. SDT-based research into gaming provides evidence that the skill-graded challenges of games are potential motivators because they can provide optimal challenges. However, challenges provided by computer-controlled algorithms can also be poorly scaled to a specific player’s abilities, and they may present overly difficult challenges that result in repeated failures. This competence frustration anecdotally can lead to aggressive feelings and behaviors. For example, in multiplayer contexts where players vary widely in their skills, there is a phenomenon known to gamers as rage-quitting (Brook, 2009). Defined as the act of disconnecting gaming equipment, sometimes violently, rage-quitting is thought to result from sudden, high-intensity negative emotional experiences in response to feeling overwhelmed by competitors. Given these potential pitfalls, game designers strive to create games that attract and match players of similar skill levels and try to create algorithms that provide incremental challenges. Game developers also make extensive use of demographic and behavioral market segmentation, targeting games primarily to one of two main audiences. The first category, known as casual games, tends to offer simple challenges that cater to desires of players seeking out spontaneous gaming sessions. Hard-core games, by contrast, provide much more difficult challenges tailored to the preferences and abilities of highly skilled and invested players. By dividing games in this way, game developers aim to create optimally engaging experiences that appeal to different kinds of game players. Given the different levels of skill necessary for different games, it is possible that players will at times encounter challenges beyond their capacities, so the present research examines the competence dimensions of gaming, focusing on its effects on postgame aggressive ideation and behaviors. This approach, concerned with need thwarting, represents a shift away from focusing mainly on violent game content as a predictor of aggression. To date, the structural and motivational aspects of gaming have not been explored as a source of player aggression. Psychologically need satisfying play has been robustly linked to game This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. COMPETENCE IMPEDANCE AND AGGRESSION 443
PRZYBYLSKI.DECI.RIGBY.AND RYAN sders the factor in the gaming-ea this.it is plausib le that need-thwarting experiences in g bustly contribute to layer essive thoughts.feeling and ntexts th and well-be s well as stoke po aggressi n.By vational sources of aggressive feelings,thoughts,and behavior Study 1 The Present Research Stdy 1investigated how players'com tudies omn-r arting experiences of comp games influence indicators of son et al 2004)that violent and the cuive deba gam nt gam y-ofcoitrolsinncncCd th d-thwarting hypoteses across ca the en (Studies and7),that xplicitly manipulated vels of mastery Sudies4.and 6.Designing the studies in this way allowed us o disentangle game content from the motivational aspects of game would be negatively associated In Study 1.we revisited a widely cited gaming related ag Method e origin of twe duates (41 males omputer games related to player aggression.In this study ived exra course credit for participating.Questionnaires wen tent of a s participan were to play one of two targe to test how e tionnaires,aggressive feelings were ass sed before and afte gam mastery-of-contro was mea dig deeper into the factor s that shan randomly of tv target gan n et al and zle game's interface to test how this ange inuen ed a emplarenl-playc game. pacted the e c positivity of attitude abo ut the gam Thi skill ca dily mastered by gaming ndermine s me meant for dedicaed c puter game plavers.Designed fo ive tigate by and in sudy 6.we All items nipulated a puz s leve of cha nge to evaluate how scales and utilized their scale as appropriate response ancho nces player aggressive feeling Aggre aim was to test the competen dance hypothesis in a com o explode and "I feel friendly"(reversed)Jtems ability in player compe stplay ssive feelings g enio ss the =0.95.= .96 we are inte rested in short-term seven studies.our aim was to investigate a perspective that con- shifts in aggressive feelings from pre-to postengagement,we
engagement, immersion, and positive short-term shifts in player wellbeing (Przybylski et al., 2009; Ryan, Rigby, & Przybylski, 2006). Given this, it is plausible that need-thwarting experiences in games have the opposite effect. Specifically, gaming contexts that impede the fundamental need for competence may undermine game appeal and well-being, as well as stoke postgame aggression. By empirically evaluating this idea, our aim is to bring a new perspective to the current literature and advance understanding about the motivational sources of aggressive feelings, thoughts, and behavior. The Present Research Seven studies were conducted to explore how competencethwarting experiences of computer games influence indicators of aggression and gaming motivation. Given the active debate surrounding the links between violent content and aggression, the present work evaluated these need-thwarting hypotheses across a range of gaming contexts. We examined the effects of competence deprivation using specific designs that compared games varying in their violent content (Studies 1 and 7), that explicitly manipulated violent content (Studies 2 and 5), or that had nonviolent content (Studies 3, 4, and 6). Designing the studies in this way allowed us to disentangle game content from the motivational aspects of game structure. In Study 1, we revisited a widely cited gaming related aggression study from a motivational perspective. The original work focused on how differences between the violent contents of two computer games related to player aggression. In this study, we explored how the varied motivational features of the games used in the experiment related to short-term shifts in aggressive feelings. In Study 2, we manipulated the violent content of a single game, holding the motivational features of the game constant. Our goal was to test how players’ experiences of mastery related to aggressive feelings independent of the violent contents of the game. In Studies 3 and 4, we shifted the focus to nonviolent gaming contexts to dig deeper into the factors that shape player experiences of competence as well as influence aggressive thoughts and feelings. In both experiments, we manipulated the complexity of a puzzle game’s interface to test how this change influenced assessments of competence and aggression. First, we evaluated how complex game controls impacted the accessibility of aggressive thoughts and positivity of attitudes about the game. Second, we tested the idea that increases in aggression rooted in competence deprivation would undermine game enjoyment. In Studies 5 and 6, we examined additional mechanisms through which games might influence player competence, aggressive feelings, and aggressive behavior. In Study 5, we investigated how different levels of player experience influenced aggressive feelings as mediated by players’ felt competence, and in Study 6, we manipulated a puzzle game’s level of challenge to evaluate how competence-impedance influences players’ aggressive feelings and behavior. Finally in Study 7, we shifted from experimental methods to a field study, recruiting self-selecting computer-game players. Our aim was to test the competence-impedance hypothesis in a community sample. We evaluated how variability in player competence as well as between-game variability in violent content related to postplay aggressive feelings and gaming enjoyment. Across the seven studies, our aim was to investigate a perspective that considers the interactivity of games as a factor in the gaming-related aggression area and to clarify how some aspects of gaming robustly contribute to players’ aggressive thoughts, feelings, and behaviors. We did that by evaluating the need-thwarting hypothesis derived from SDT in the gaming domain, an area that is principally concerned with performance, interaction, and competence. Study 1 Study 1 investigated how players’ competent use of game interfaces related to gaming-related aggression. We adapted a widely cited experimental design focused on gaming and aggression (Anderson et al., 2004) that compared one violent game and one nonviolent game. Because the violent game had complex controls and the nonviolent exemplar had simpler controls, we tested the extent to which self-reported mastery-of-controls influenced shortterm increases in aggressive feelings while controlling for the game content. We predicted that those playing the violent game (with a complex interface) would report lower levels of masteryof-controls compared to those assigned to play the low-violence game, and we hypothesized that the players’ mastery-of-controls would be negatively associated with aggressive affect. Method Procedure. Ninety-nine university undergraduates (41 males) received extra course credit for participating. Questionnaires were administered before and after a 20-min game engagement period. Following the completion of the first group of questionnaires, participants were randomly assigned to play one of two target games. Demographic information was collected on the initial set of questionnaires, aggressive feelings were assessed before and after game engagement, and the players’ mastery-of-controls was measured following game engagement. Target games. Participants were randomly assigned to play one of two target games identified by Anderson et al. (2004; Study 2) as representative of violent and nonviolent games. The nonviolent exemplar, Glider Pro 4, is a single-player game, designed for relatively short play, in which players use two keyboard keys to navigate a paper airplane through a two-dimensional image of a home interior. This skill can be readily mastered by gaming novices. The violent game used in this study, Marathon 2, is a game meant for dedicated computer game players. Designed for extended gaming sessions, players must use a mouse and 20 keyboard keys to navigate combat in a three-dimensional environment. Measures. Survey measures were delivered in HTML format. All items, except for participant gender and age, used 7-point scales and utilized their scale as appropriate response anchors. Aggressive affect. The 35-item State Hostility Scale was used to assess aggressive feelings (Anderson et al., 1995). Participants rated the extent to which they agreed with each hostile mood statement at that moment: “I feel irritated,” “I feel like I am about to explode,” and “I feel friendly” (reversed). Items were reversed appropriately and averaged, creating before-play (M 2.41, SD 0.82, .97) and after-play aggressive affect scores (M 2.54, SD 0.95, .96). Because we are interested in short-term shifts in aggressive feelings from pre- to postengagement, we This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 444 PRZYBYLSKI, DECI, RIGBY, AND RYAN
COMPETENCE IMPEDANCE AND AGGRESSION 445 ffec Thi Mo player by way o To asse oarticipants'felt competer ontrols (the mediating var arch (Ryan Righy Items included"The game otal effect observed relati game type to change ted to do s the game it was easy to (path C).B hA) aggres as predicted by ma Results AB) .00 ranged fror 0.02t cant while the A"B path was significant. females tended to have les and I short-term shifts in aggressive feelings insofar as the more com cted a plex control interface impeded player competence includi raphics as c ntrol variable s and once le n the pa ut.The Study 2 raphics tedwhincrceaaomaslepy-0 ated only the graphics and nrative conceit game to ould be esd of the sam the t when controlling fo ne game sed le this byr sidualized chang ng 001.R2= 10 +.B97 06 data Method ed us Procedure.One hundred one unc aduates (36 males dit i gar (viole specifically for training purposes.Following this.dem and gar intcractcdt ted in order ided initia nt for the of in nterface get gar cont desig ld pr idable challenge btained by ototheganc-conditioncoteonviolknt--lhiebyiolcnc ctence in using the tudy allowe test a c d by a.200 R sing ch e in aggressive affec game.B(9s 004.p= Figre.Conceptual mediation model
regressed postplay scores onto preplay scores and saved the standardized residual scores. This enabled us to quantify change in aggressive affect for each participant. Mastery-of-controls. To assess participants’ felt competence using game controls, we used an intuitive controls assessment used in previous motivation and gaming research (Ryan, Rigby, & Przybylski, 2006). Three items assessed how effortlessly participants found the control interface for carrying out their intentions in the game. Items included “The game controls are intuitive” and “When I wanted to do something in the game, it was easy to remember the corresponding control.” The scale demonstrated acceptable reliability (M 4.59, SD 1.59, .72). Results Demographics. Participants’ age and gender were collected in all seven of the present studies. Where significant differences emerged, the relations were relatively small in magnitude (average R2 .03) or were not of interest in the present inquiry; for example, females tended to have less gaming experience and lower levels of competence during play than did males. We conducted all analyses presented in this article twice, once including participant demographics as control variables and once leaving them out. The direction, magnitude, and significance levels did not vary between these parallel analyses, so we did not report the analyses including demographics to conserve space. Player competence. We hypothesized that poor mastery-ofcontrols would be associated with increased levels of aggressive feelings following game engagement when controlling for type of game and the interaction of game type and mastery-of-controls. To test this hypothesis, we regressed residualized change scores in aggressive feelings simultaneously on mastery-of-controls, (97) .33, p .001, R2 .10, and the target game type (nonviolent 1, violent 1), (97) .06, p .55. The data collected for Study 1 allowed us to evaluate the relation of players’ felt competence (i.e., mastery-of-controls) to aggression when controlling for game content (violent vs. nonviolent). We evaluated a hierarchical moderation model to test the idea that game content moderated the relation between mastery-of-controls and aggressive feelings and found the interaction term was not significant, (95) .07, p .80. We thus did not find evidence that player felt competence and game content interacted to account for additional variability in player aggression. Results from this provided initial support for the competence-impeding hypothesis: Struggling with feelings of incompetence at the game interface relates to increased levels of aggressive affect. Target game differences. We predicted that the violent game used in this classic design would present a formidable challenge to mastery-of-controls, thus impeding participants’ experiences of competence. Results obtained by regressing mastery-of-controls onto the game-condition code (nonviolent 1, high violence 1), (98) .20, p .05, R2 .04, showed the violent game presented a barrier to players’ competence in using the controls. The present study also allowed us to test a conceptual replication of the aggression effects reported by Anderson et al. (2004; Study 2). Results derived by regressing change in aggressive affect onto game condition-code did not show higher aggression for those randomly assigned to play the violent game, (98) .004, p .97. We also examined the indirect effect that game content (violent vs. nonviolent) had on players’ aggressive feelings by way of player competence. More specifically, we evaluated a mediation model in which we tested mastery-of-controls (the mediating variable in Figure 1) as an intervening factor linking game type (the independent variable) to short-term shifts in aggression (the outcome variable) following the bootstrapping approach outlined by Preacher and Hayes (2008). As noted previously, there was no total effect observed relating game type to change in aggressive feelings (path C), .08, p .41; mastery-of-controls was predicted by game type (path A), .25, p .001; and change in aggressive feelings was predicted by mastery-of-controls (path B), .32, p .001. Bootstrapping indicated a significant indirect effect (path A B). The 95% confidence interval for the indirect path, based on 10,000 resamples, ranged from 0.02 to 0.33, and it accounted for 10.61% of the variability (R2 ) in aggressive feelings. Given that the C and C= paths were nonsignificant while the A B path was significant, results indicated that being assigned to play the violent game had a positive influence on short-term shifts in aggressive feelings insofar as the more complex control interface impeded player competence. Study 2 The aim of Study 2 was to conceptually replicate Study 1 in a more elegant way. Specifically in Study 2, we carefully manipulated only the graphics and narrative conceit of a single game to have a violent version and a nonviolent version of the same game. As in Study 1, we tested the hypothesis that players who felt incompetent using the game interface would report increased levels of aggressive feelings postengagement, over and above variability attributable to violent game content. Method Procedure. One hundred one undergraduates (36 males), mean age 19.6 years (SD 1.32), received extra course credit in exchange for participating. Upon entering the lab, all participants engaged in a 20-min practice period with a version of the game created specifically for training purposes. Following this, demographic information and aggressive affect were assessed, and participants were randomly assigned to a high or low violence 20-min play period. These versions were created in order to present two levels of violence, holding constant other aspects of the game such as landscape, visual complexity, and gameplay mechanics. Target game content. We created two additional gaming environments based on the popular and commercially available game Half-Life 2 using a programmer tool kit. The first environFigure 1. Conceptual mediation model. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. COMPETENCE IMPEDANCE AND AGGRESSION 445
PRZYBYLSKI.DECI.RIGBY.AND RYAN characte grammed by the expe nenter verbally supp ported less mastery of the game interface also reported in- he original version nd a ve reated).which wereused for he The 41 did no cant interaction.B(97)=04.p=65 anmsofleelOfvioicentcoaite Study 3 alle nt age an petenc-impedinginrfa Results eport I check oent conten and not ga ations of the gaming experience To les whe nanipula on of co Method Particinants and One hundred fou ,B(100) 27,P ates (30 males) ed th tenin To test if indvertently manipulated player Par ere randomly assigned t play. participants completed a lexica mastery- Plaver competence. us the et in ce feec gre g the lab,participants were rand ned to play (sce ure 2A) violent =+1).B(99)=-09.p =32.Results from this Measures.After play.participants Table 1 of-controls (M-3.01,SD-1.30.93).Table 2 presents Correlations Observed Between Variables in Study 2 eil reAgolo-go lex 2 3 Check) objective of this task was to evaluate how quickly participants -31 ant
ment was crafted to teach participants how to use the game-control interface to interact with the gaming environment. A friendly character programmed by the experimenter verbally supported participants, providing them with instructions, encouragement, and suggestions when they faltered. The other two environments (i.e., the original version and a version we created), which were used for the primary game play, were structurally identical but varied with respect to how the participant removed a competitor from play. In the low-violence condition, competitors (i.e., the computer) used a nonlethal “marker” to tag participants, and participants used a psychic-ball power that teleported competitors away. The power lifted and evaporated competitors to remove them. In the original high-violence condition, both the competitors (computer) and the participants used firearms to maim and dispatch the opponents, leaving them spewing blood and lying dead in the game world. Thus, the two game variants were motivationally fixed but varied in terms of level of violent content. Measures. Participant age and gender were assessed before the challenge period, aggressive feelings were measured before (M 2.88, SD 0.86, .97) and after the challenge period (M 2.53, SD 0.82, .93), and mastery-of-controls (M 5.33, SD 1.32, .87), was assessed after the challenge period. Table 1 presents zero-order correlations between observed variables. Results Manipulation check. To ensure that redesign influenced violent content and not game structure, we conducted two manipulation checks. To test whether the manipulation of content was successful, we assessed perceived threat using the single item: “I felt vulnerable during play.” We regressed these scores onto game type (low violence 1, high violence 1), (100) .27, p .01, R2 .08. This result showed the violent game was more threatening. To test if the design inadvertently manipulated player competence, we regressed mastery-of-controls scores onto game type, (100) .08, p .44. The results showed that the violent content manipulation was successful (i.e., higher player threat) without influencing game structure (i.e., invariance in mastery-ofcontrols). Player competence. We hypothesized that participants’ poor mastery-of-controls, and thus their felt incompetence, would lead them to experience increased levels of aggression. To test this, we regressed residualized change scores in aggressive feelings simultaneously onto mastery-of-controls, (99) .22, p .03, and the target game type (nonviolent 1, violent 1), (99) .09, p .32. Results from this analysis conceptually replicated those derived from Study 1. Using a more rigorous method, we found that those who reported less mastery of the game interface also reported increased levels of aggressive feelings. Player competence and target game differences. The experimental design allowed us to evaluate two additional relations: an overall effect for violent game content, in line with theory advanced by Anderson et al., 2004, and a potential interaction between player competence and content. Results derived by regressing change in aggressive affect onto game type, (99) .08, p .41 did not support a main effect for violent content. We evaluated a hierarchical moderation model to test if competence frustration and violent content interacted, and we found no significant interaction, (97) .04, p .65. Study 3 The central aim of Study 3 was to examine how competenceimpeding gaming influences aggressive thoughts. The design used in Study 3 experimentally manipulated the control interface of a nonviolent puzzle game, making it either simple and intuitive or complex and highly challenging. Of interest was the effect that this competence-impeding interface might have on players’ accessibility of aggressive thoughts and their evaluation of the game. We hypothesized that players randomly assigned to use the complex interface would report lower levels of mastery-of-controls, have faster access to aggressive thoughts, and hold less positive evaluations of the gaming experience. Method Participants and procedure. One hundred four undergraduates (30 males), mean age 19.92 years (SD 1.23), participated in exchange for course credit. Participants were randomly assigned to 10 min of game play using either a simple or a complex control interface. Following game play, participants completed a lexical decision task assessing the accessibility of aggressive thoughts and a thought-listing task measuring the positivity of evaluations of the game. Target game complexity. We used a variant of Tetris, a popular puzzle game, in Studies 3 and 4 because it provides straightforward challenges and unambiguous performance feedback that communicates an immediate sense of ability. Upon entering the lab, participants were randomly assigned to play by using either a simple button layout (see Figure 2A) or a complex layout requiring more effort to master (see Figure 2B). Measures. After play, participants completed the lexical decision task to measures aggressive thoughts, a thought listing task to measure game attitude, and a self-report assessment of masteryof-controls (M 3.01, SD 1.30, .93). Table 2 presents zero-order correlations observed between study variables. Lexical decision task: Aggressive thoughts. A go/no-go lexical decision task was used to evaluate how readily accessible aggressive thoughts were following game engagement. The objective of this task was to evaluate how quickly participants could identify words linked to aggression relative to neutral words. In full, participants completed 110 trials, the first 10 of which were practice trials. Each trial began with participants focusing on a fixation point of “” for 200 ms, which was Table 1 Correlations Observed Between Variables in Study 2 Variable 1 2 3 1. Violent Content — 2. Threat (Content Manipulation Check) .27 — 3. Mastery-of-Controls .08 .22 — 4. Aggressive Feelings .08 .48 .31 Note. n 101. p .05. p .01. p .001. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 446 PRZYBYLSKI, DECI, RIGBY, AND RYAN
COMPETENCE IMPEDANCE AND AGGRESSION 447 ed on the s all five Results We hypothesized that players using the complex interface would Rotate +】 vould be negatively related to aggressive thoughts and positively of the game's interface showed lower e thoughts and more positive attitudes about the game. e ()d pe (ay Study4 mpeding ga aced by a blank s en for 50 ms and was in tur n roduced this masu ire bayers' we wanted to evaluate it as glish word.F iating factor that linked both complexity()o legitimate English word and to wait until the next trial be ine SDT-be that the ch,we tes three hypothes e.g. abuse,damage,injure,outrage)Reaction times to legit players'competence need satisfaction and also that players'com nd the ms and above 1500ma1es 4%of trials.There were no si ficant or ach participant by subtracting em resulted in resno ime data not being orded for the participants Variable 2 oward the 4.Positive Came Auitud puzzle game.Participants then rated each thought on a 7-point s8,<0L
replaced by a blank screen for 50 ms, and was in turn replaced by a string of letters that was either a legitimate or a plausible but illegitimate English word. Participants were instructed to strike the spacebar key as quickly as possible if the string was a legitimate English word and to wait until the next trial began if the string was a nonword (2,000 ms). Five practice trials presented neutral words, and five used text strings of plausible nonwords. The remaining trials presented 60 strings that were nonwords, 20 neutral words, and 20 that related to aggression (e.g., abuse, damage, injure, outrage). Reaction times to legitimate English words below 250 ms and above 1,500 ms were discarded as outliers, and these dropped cases comprised less than 4% of trials. There were no significant condition differences in word accuracy between conditions; reaction times were log transformed and aggressive thoughts scores were calculated for each participant by subtracting reaction times to aggression words from reaction times to neutral words. A technical problem resulted in response-time data not being recorded for the first eight participants. Where applicable, all analyses were done twice, once including these individuals and once excluding them. Because the effects did not differ, the second set of analyses is not reported. Thought listing task: Game evaluation. A thought-listing task was used to tap into the participants’ attitudes toward the game. Specifically, participants were instructed to rapidly jot down the first five thoughts that came to mind when thinking about the puzzle game. Participants then rated each thought on a 7-point scale in terms of how extremely positively (3) versus extremely negatively (3) it reflected on the game. Positivity of evaluations was calculated for each participant by averaging across all five ratings (M 0.05, SD 1.43, .73). Results We hypothesized that players using the complex interface would report less mastery-of-controls, demonstrate faster access to aggressive thoughts, and report less positive attitudes toward the game compared to those who used the simple interface. To test these predictions, in three regression models, we regressed mastery of controls, (95) .76, p .001, R2 .58, aggressive thoughts, (95) .23, p .01, R2 .05, and game evaluations, (95) .27, p .01, R2 .07, onto the condition codes for our manipulation of controls (simple 1, complex 1). These results provided support for the hypotheses. We also hypothesized that players’ mastery of game controls would be negatively related to aggressive thoughts and positively related to evaluations of the game. Results from two analyses, regressing aggressive thoughts, (95) .27, p .01, R2 .07, and game evaluations, (95) .43, p .001, R2 .19, onto mastery-of-controls indicated that players who felt more mastery of the game’s interface showed lower automatic access to aggressive thoughts and more positive attitudes about the game. Study 4 Study 4, like its predecessor, manipulated the difficulty of game controls to gain a better understanding of how competenceimpeding gaming related to player aggression. New to this study was a direct measure of players’ competence-need satisfaction. We introduced this measure because we wanted to evaluate it as a mediating factor that linked both complexity (i.e., difficulty) of controls and reported mastery-of-controls to aggressive affect. In line with the competence-impedance hypothesis and previous SDT-based gaming research, we tested three hypotheses. First, we hypothesized that the difficult controls would be associated with increased aggressive feelings and decreased levels of gaming enjoyment. Second, we predicted that that mastery-of-controls would mediate the relation between complexity of the game interface and players’ competence need satisfaction and also that players’ competence need satisfaction would mediate the links between mastery-of-controls and aggressive feelings. Finally, we hypothesized that increased levels of aggressive feelings would be negatively associated with player enjoyment. Table 2 Correlations Observed Between Variables in Study 3 Variable 1 2 3 1. Complexity of Controls — 2. Mastery-of-Controls .76 — 3. Aggressive Thoughts .23 .27 — 4. Positive Game Attitude .23 .41 .13 Note. n 98. p .05. p .001. Figure 2. Simple (layout A) and complex (layout B) interfaces used in Studies 3 and 4. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. COMPETENCE IMPEDANCE AND AGGRESSION 447
448 PRZYBYLSKI.DECI.RIGBY.AND RYAN Method sing ma onto the Supported these expectationsce (simple complex We redicted that Measures.In this studv.we used a n measure of ag ings to shorten the assessment from 35 i compet after the play period. sfaction would detract from enj nent and foment aggre nge in aggressive ed satisfaction gnmeee Comp of the Int M:Ryan.Mims.Ko overall competer e-need satisfactio sed player simultaneously ont na 60.P< tem scores were averaged for ea h participant to compute 001.and omplexity-of Aggressire afect.Suate-level shifs in nitu of the effect (path A"B) 1986 nd Clark 0 faction by way of mastery. ssive b evels of player cor 139 50 -0.78.90)and after play (M 257.5D=0.86 nd change scores for ago ssion were calculated between these hon-t nced during game was as ive feelings. nc i nt su of the tal.1983). 139.=89 tisfaction,B(138) 48.p Results ndirect paths.B= 19139=d -2.94.D 001.for com We h hesized that participant h39 .fo d play To test the paue m and significance of in- ions Observed Betwveen Variables in Sudy4 t effects linking complexity of control and mastery-of Variable 1 3 45 and mastery satisfaction,B(138)=.ina second.Results demon- -30二 001,form
Method Participants and procedure. One hundred forty-one undergraduates (52 males) participated in exchange for extra course credit. The procedures of this study closely followed those of Study 3 except that participants completed self-report measures in place of the lexical and thought-listing tasks used in the previous study. Measures. In this study, we used a new measure of aggressive feelings to shorten the assessment from 35 items to six and to be able to disguise them more easily among filler items; before and after the play period, participants completed assessments of aggressive feelings (outlined below). Mastery-of-controls (M 3.56, SD 1.88, .90), players’ competence need satisfaction, and game enjoyment were assessed after game play. Table 3 presents zero-order correlations between variables. Players’ competence need satisfaction. Competence need satisfaction experienced during the game was assessed after the game using a three-item competence measure based on the widely used task competence subscale of the Intrinsic Motivation Inventory (IMI; Ryan, Mims, & Koestner, 1983), the items demonstrated good internal consistency (M 4.14, SD 1.52, .76) in keeping with past research (Ryan, Rigby, & Przybylski, 2006; .82). Item scores were averaged for each participant to compute players’ competence need satisfaction scales. Aggressive affect. State-level shifts in aggressive feelings were assessed using the hostile emotions subscale of the Positive and Negative Affect Schedule (PANAS-X; Watson & Clark, 1994). Participants rated the extent to which they felt aggressive by evaluating six adjectives (e.g., angry, hostile, scornful) before and after play based on how they felt at that moment. Before (M 2.50, SD 0.78, .90) and after play (M 2.57, SD 0.86, .92) state-level scores were computed for each participant, and change scores for aggression were calculated between these two scores using the method outlined in Study 1. Game enjoyment. How much enjoyment and fun participants experienced during game play was assessed with four items adapted from the interest and enjoyment subscale of the IMI (Ryan et al., 1983). Sample items included “I enjoyed playing the game very much” and “I thought the game was boring” (reversed). Items were averaged to create total enjoyment scores (M 4.90, SD 1.39, .89). Results Target game complexity. We hypothesized that participants assigned to use complex controls would experience lower levels of mastery-of-controls and feel a diminished sense of overall competence need satisfaction relative to those who played using the simple interface. Results derived from models regressing masteryof-controls, (140) .65, p .001, R2 .42, and player competence need satisfaction, (140) .42, p .001, R2 .19, onto the target game interface (simple 1, complex 1) supported these expectations. Player competence need satisfaction. We predicted that competence need satisfaction during gaming would serve to enhance game enjoyment, whereas diminished competence-need satisfaction would detract from enjoyment and foment aggression. Results derived by regressing change in aggressive feelings, (140) .39, p .001, R2 .15, and game enjoyment, (140) .53, p .001, R2 .28, onto player competence-need satisfaction supported our expectations. Feeling competence-need satisfaction was associated with short-term decreases in aggressive affect and higher levels of player enjoyment. Target game complexity and player competence need satisfaction. We hypothesized that the relation between interface complexity and players’ overall competence-need satisfaction would be mediated by the players’ mastery-of-controls. To test this expectation we regressed player competence simultaneously onto mastery-of-controls, (139) .60, p .001, and complexity-ofcontrols (139) .04, p .67, and then we calculated the magnitude and significance of the indirect effect (path A B) as outlined by Baron and Kenny (1986). Results demonstrated a significant indirect path linking complexity of controls to lower levels of player competence-need satisfaction by way of masteryof-controls, .39, t(139) 5.70, p .001. Mediating role of player competence need satisfaction. We next tested the hypotheses that the influence of both the complexity of the game’s interface and the mastery-of-controls on both short-term shifts in aggressive feelings and player enjoyment would be mediated by the overall levels of player competenceneed satisfaction. Aggressive feelings. To determine the magnitude and significance of an indirect effect linking complexity of controls and self-reported mastery-of-controls to aggression, we regressed change in aggression simultaneously onto complexity, (138) .18, p .05, and player competence-need satisfaction, (138) .34, p .001, in one model, and mastery-of-controls, (138) .09, p .10, and player competence-need satisfaction, (138) .48, p .001 in a second model. Results demonstrated two significant indirect paths, .19, t(139) 2.94, p .001, for complexity and .26, t(139) 4.03, p .001, for mastery. This provided evidence that both control complexity and mastery-ofcontrols related to aggressive feelings in part because they influenced overall player competence-need satisfaction. Game enjoyment. To test the pattern and significance of indirect effects linking complexity of controls and mastery-ofcontrols to gaming enjoyment, we regressed player enjoyment simultaneously onto complexity, (138) .12, p .10, and player competence, (138) .59, p .001, in one model, and masteryof-controls, (138) .07, p .30, and player competence-need satisfaction, (138) .58, p .001, in a second. Results demonstrated two significant indirect paths (A B), .23, t(139) 2.08, p .05, for complexity, and .32, t(139) 6.43, p .001, for mastery. The result indicated that the complexity of game controls and sense of mastery-of-controls related to the appeal of Table 3 Correlations Observed Between Variables in Study 4 Variable 1 2 3 4 5 1. Complexity of Controls — 2. Mastery-of-Controls .66 — 3. Player Competence .45 .61 — 4. Game Enjoyment .15 .28 .53 — 5. Aggressive Feelings .34 .20 .42 .30 — Note. n 141. p .05. p .001. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 448 PRZYBYLSKI, DECI, RIGBY, AND RYAN
COMPETENCE IMPEDANCE AND AGGRESSION 449 Method elated ags ion and eniovme Partici and found that players who bee me inc Upon enteting the laborat play by (40 y.participants were o one of four condition (a)experience and low violence chal .08 violence challenge.Participants assigned to one of the two expe Study5 The aim of Study 5 was to conceptually replicate and further Study 2. the-pricn activitie te how Study 5 ex ded on Study 1where tw o gan e controls was ant across s but those in the n all participants completed a questionnaire particinants then plaved onditions and gave players in one condition the opportunity to oe of the wo versions of the game ued in Study 2.presaged by alvtie findines v.2001.2007.whic d the ion In articular and they w (M =131,SD 048. 72)and 81 30 laver competen eed satisfaction.In line with this idea we evaluated four hypoth ed satisfaction (M=2.77.SD 1.11.a= 89),and player r 72,a=.92)wer First,we hypothe sized that opportunities to build stery-of-controls een variables. A behavioral was used as ar ience would positively influ appe ted that gardin how to spend the atory tim ed satisfaction would serve to nediate link om both e game o ng and motivation finally.we predicted that increased aggre was der Table 4 Correlations Observed Between Variables in Sudy 5 6 Experienc on Check for Content) of-Co &102<1.一p<0L
the game insofar as they influenced overall player competence satisfaction. Gaming-related aggression and enjoyment. We expected and found that players who became increasingly aggressive as a consequence of engagement would experience less enjoyment of play by regressing the former onto the latter, (140) .28, p .001, R2 .08. This result indicated that shifts in aggressive feelings were negatively associated with game enjoyment. Study 5 The aim of Study 5 was to conceptually replicate and further evaluate how experimental manipulations meant to thwart players’ competence needs would predict short-term shifts in aggression and enjoyment. Study 5 expanded on Study 1 where two games that were used incidentally differed in the difficulty of controls, and on Studies 3 and 4 where the complexity of the game controls was designed to make the controls in one condition more difficult than in the other. In Study 5, we used a complex interface in both conditions and gave players in one condition the opportunity to develop experience with the game. This manipulation was inspired by meta-analytic findings in the gaming-related aggression literature (Sherry, 2001, 2007), which revealed that players’ length of exposure to and experience with games moderated the extent to which play was associated with indicators of aggression. In particular, results showed that very brief periods of video game play were most strongly associated with aggression. Viewed from the SDT perspective we are applying, this result suggested to us that insufficient experience with games might be a source of increased aggression because limited experience would thwart mastery-of-controls and competenceneed satisfaction. In line with this idea we evaluated four hypotheses. First, we hypothesized that opportunities to build experience with a game would foster greater mastery-of-controls, leading to lower player aggression and higher game motivation. Second, we predicted that amount of player experience would positively influence overall competence-need satisfaction insofar as it bolstered mastery-of-controls. Third, we expected that players’ competenceneed satisfaction would serve to mediate links from both levels of player experience and mastery-of-controls to both aggressive feelings and motivation. Finally, we predicted that increased aggressive feelings would be negatively related to players’ game motivation. Method Participants and procedure. One hundred twelve undergraduates (33 males) received extra course credit for participating. Upon entering the laboratory, participants were randomly assigned to one of four conditions: (a) experience and low violence challenge, (b) experience and high violence challenge, (c) no experience and low violence challenge, or (d) no experience and high violence challenge. Participants assigned to one of the two experience conditions engaged in a 10-min training period, which was an abbreviated version of the practice that all participants had in Study 2. Participants in the no-experience condition activities spent the 10-min period by playing with the game used in Studies 3 and 4. Therefore, the time and activity of using some types of game controls was constant across groups, but those in the no experience condition had less exposure to the target game. Following the practice period and before the 10-min challenge period, all participants completed a questionnaire. Participants then played one of the two versions of the game used in Study 2, presaged by short films that framed play as either kill-or-be killed combat in the high-violence condition or a friendly game of tag in the lowviolence condition. After this, participants completed a second questionnaire, and the experimenter provided them with a choice of either continuing with another 10-min period of play or browsing the Internet. A technical problem prevented data from three participants being recorded, and they were dropped from the study. Measures. Measures, using 5-point scales, of aggressive feelings were assessed before (M 1.31, SD 0.48, .72) and after play (M 1.40, SD 0.51, .81), and measures of mastery-of-controls (M 3.07, SD 1.07, .83), overall player competence-need satisfaction (M 2.77, SD 1.11, .89), and player motivation (M 2.48, SD 0.72, .92) were measured only after engagement. Table 4 presents zero-order correlations between variables. Player motivation. A behavioral assessment was used as an additional method for evaluating game appeal. Following the second survey, the experimenter offered participants specific choices regarding how to spend the remaining 10 min of laboratory time: they could persist at playing the game or browse the Internet. The time participants played with the game was considered a measure of game motivation and was derived from the frequently used free-choice measure of intrinsic motivation (Deci, 1971). To create a parsimonious, overall measure of game appeal, the behavioral Table 4 Correlations Observed Between Variables in Study 5 Variable 1 2 3 4 5 6 1. Violent Content — 2. Game Experience .01 — 3. Threat (Manipulation Check for Content) .29 .04 — 4. Mastery-of-Controls .05 .27 .02 — 5. Player Competence .02 .38 .01 .63 — 6. Player Motivation .14 .21 .04 .33 .48 — 7. Aggressive Feelings .03 .20 .02 .16 .30 .23 Note. n 109. p .05. p .01. p .001. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. COMPETENCE IMPEDANCE AND AGGRESSION 449
450 PRZYBYLSKL DECL RIGBY.AND RYAN kecpledi2ofontokioleckofagercsie To determine the magnitude and signif Results eftect linking expen and mastery-of Manipulation check.To ensure that we successfully manip player e.B(107) .10.p >30 68 29.p01,R -34,p001,in ndition codins dow violence AB)8= 111081=-305n70.thus replicating findings in previous studies. ontrols.B= Study 6 terv-of- dy 6 was designed t binhercvalh e the robustness of th maive manipulatio actio odel of the t This supplen game motivation.B(108)=57.001.R Studies are tha sment additional reg we nt e igh +1).B(107 03.P=.74.We also eva behavior as well as increased levels of aggressive feelings Method ted tha eelings and higher levels of plaver motivation. s that was robust across different levels of violent game -hr We then pro to co eed satisf d mediate the ons from both amount of
measure of game motivation was combined with the self-report measure of game enjoyment (r .44) after each was standardized. Results Manipulation check. To ensure that we successfully manipulated the violent content of the computer game without affecting the motivational aspects of play, we regressed perceived threat (same measure used in Study 2), (108) .29, p .01, R2 .07, and mastery-of-controls, (108) .05, p .50, onto the game condition coding (low violence 1, high violence 1). Results demonstrated that the manipulation of violent content was successful: participants facing high violence challenges felt more threatened, and the manipulation of content did not influence players’ perceptions of mastery. Player experience. To test the prediction that the amount of players’ experience with the game would be associated with higher levels of mastery-of-controls, player competence-need satisfaction, and game motivation, as well as with lower levels of aggressive feelings, we evaluated three regression models. Results derived by regressing mastery-of-controls, (108) .27, p .01, R2 .07, player competence, (108) .38, p .001, R2 .15, change in aggression, (108) .20, p .05, R2 .04, and game motivation (108) .21, p .05, R2 .05, onto the amount of player experience (coded: no 1, yes 1), provided support for the first hypothesis. Player experience and competence-need satisfaction. To evaluate our hypothesis that mastery-of-controls would serve to mediate the relations between player experience and player competence-need satisfaction, we regressed player competence satisfaction simultaneously onto mastery-of-controls, (107) .57, p .001, and player experience, (107) .23, p .01. We then calculated the magnitude and significance of the indirect effect relating these factors. Results demonstrated a significant indirect path (A B), linking amount of experience to player competence satisfaction through its effect on mastery-of-controls, .17, t(108) 2.59, p .001. Player experience effects mediated by competence-need satisfaction. We next did a series of analyses to evaluate the links between overall player experience and both change in aggression and game motivation as mediated by competence-need satisfaction. First, in two regression models, we regressed change in aggressive feelings, (108) .30, p .001, R2 .09, and player game motivation, (108) .57, p .001, R2 .33, onto player experience. To ensure that the competence-impedance by lack of experience was independent of violent game content, we tested two additional regression models. We regressed change in aggression scores simultaneously onto player experience, (107) .30, p .001, and the game content coding (low violence 1, high violence 1), (107) .03, p .74. We also evaluated game content as a moderator of the link from player experience to aggression, but the interaction term was not significant, (106) .06, p .56. Taken together, these sets of analyses indicated that player experience was associated with lower levels of aggressive feelings and higher levels of player motivation—a pattern of relations that was robust across different levels of violent game content. We then proceeded to evaluate the prediction that competenceneed satisfaction would mediate the relations from both amount of player experience and mastery-of-controls to levels of aggressive feelings and player motivation. Aggressive feelings. To determine the magnitude and significance of the indirect effect linking experience and mastery-ofcontrols to aggressive feelings, we regressed change in aggression simultaneously onto player experience, (107) .10, p .30, and player competence satisfaction, (107) .26, p .01, in one model, and mastery-of-controls, (107) .06, p .50, and player competence satisfaction, (107) .34, p .001, in a second model. Results demonstrated two significant indirect paths (A B), .11, t(108) 3.05, p .01, for experience, and .19, t(108) 0.43, p .001, for mastery; linking these factors to shifts in aggressive feelings by way of their influence on player competence-need satisfaction. Player motivation. To test the pattern and significance of the indirect effect linking player experience and mastery-of-controls to player motivation, we regressed the composite motivation measure simultaneously onto player experience, (107) .03, p .50, and player competence satisfaction, (107) .46, p .001, in one model, and mastery-of-controls, (107) .05, p .50, and player competence satisfaction, (107) .45, p .001, in a second. Results demonstrated two significant indirect paths, .17, t(108) 3.31, p .001, for experience, and .28, t(108) 5.18, p .001, for mastery, linking these factors to game motivation by their effect on player competence-need satisfaction. Gaming-related aggression and player motivation. The final hypothesis was that increased levels of aggressive feelings would be related to less game motivation. Results derived from regressing player motivation, (108) .22, p .01, R2 .05, onto change in aggression supported this prediction. We also tested for a direct effect, linking violent game content to players’ aggressive feelings but did not find a significant effect, (108) .03, p .70, thus replicating findings in previous studies. Study 6 Study 6 was designed to further evaluate the robustness of the need-thwarting hypothesis by advancing the present research in two ways. First, we implemented an alternative manipulation to thwart players’ felt competence in which we altered the algorithms that shaped the players’ degree of challenge, making it overly difficult in one of the two conditions. This supplements the manipulation of different games (Study 1), different control interfaces (Studies 3 and 4), and different levels of experience (Study 5). Second, in Study 6, we utilized a behavioral assessment as the primary outcome, measuring the assignment of physical pain to another participant as an index of aggression. In this study we evaluated a single hypothesis: that high game difficulty intended to thwart competence would relate to higher levels of aggressive behavior as well as increased levels of aggressive feelings. Method Participants and procedure. Forty-seven undergraduates (21 males, Mage 20.31 years, SD 3.92) received £3.00 compensation in exchange for participating in this 0.5-hr experiment. Participants were informed that the purpose of the study was to understand how physiological experiences relate to computer game engagement and emotion. Participants (a) filled out a demoThis document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 450 PRZYBYLSKI, DECI, RIGBY, AND RYAN
