452 PRZYBYLSKI.DECI.RIGBY.AND RYAN ras drawn from mer bers of popular web- ed community tha ums for c gan Y=Po+P1+E dee gan rate models for each t in games were coded in two ways. variable.Results indicated that 1.919)= -3.27,p<.01,and posi Software rating board ESRB)isa group largely funded throug 38919 satisfactio 时 stry th ngs independen the following Level os for cach Y■P。+P,+P,+E ng a here Y is the oucome of In addition blind to the purpose of the four ent. study. and E is error at Level 1.Results from 01, nd P 1RQ6 7 hat ssigned to games with no violent content eith was significantly linked to postplay oever Mari Bro ith impe I violence (e.g.. .(918) sive feeli e s grand Theft auto ).To verify reliability.50 game titles were selected 94).These ratcr Level (E) ing title olent content scores were calculated for Y=P。+P,+E Participants were asked to reflec on the of thei mputer gaming and sclect the thre game on (M 366.D=089 881 General Discussion -point es To this end.r popular interest and concems abou ons entists,and the waysand t to which had to be co npleted three times by each par gmay be linked to h come a hot Results uate a compete -impedance hypothesis.Our approach wa Analytic strategy.Given the nested nature of the data-each enbush.199)This :Bryk ch allowed us test our n the pr t research.our core interest concemed the relations use of the hierarchical arrangement of dat een the Plaver comp ence.To test the hypothesis that players ed plavers'feelings of co would lead t aggressive th oughts.feelings and chavior.Importantly participated in exchange for entry into a raffle for an electronic game of their choice, worth approximately $50 to $75. This sample was drawn from members of a popular web-based community that provides online forums for discussions about electronic games and Internet culture. Because avid gamers typically play more than one computer game at a time, they were requested to provide data on the three games they had played most frequently in the past month. After data collection was completed, the levels of violent content present in games were coded in two ways. Target games coding. ESRB rating. The Entertainment Software Rating Board (ESRB) is a nonprofit group largely funded through the electronic games industry that assigns titles one of five ratings: E (Everyone), E10 (Everyone 10 or older), T (Teen), M (Mature), or AO (Adults Only), which we coded from 1 to 5, respectively. Because ESRB categories have changed slightly over time, we used the most up to date ratings for each game. In this sample no game mentioned by the participants was an AO game, effectively creat￾ing a 1– 4 scale. Content rating. In addition to using the ESRB ratings, four raters blind to the purpose of the study, coded all the target games utilizing a more stringent method used in past research to quantify the intensity of computer game content (Przybylski et al., 2009). A rating of 1 was assigned to games with no violent content what￾soever (e.g., SimCity 4), a 2 was assigned to games with abstract violence (e.g., Super Mario Bros. Wii), a 3 was assigned to games with impersonal violence (e.g., Civilization 4), a 4 was assigned to games with fantasy violence (e.g., World of Warcraft), and a 5 was assigned to games with realistic violence (e.g., Grand Theft Auto 4). To verify reliability, 50 game titles were selected at random, and scores generated by the four raters were examined for consis￾tency. Reliability was good (k
.94). These raters coded all remaining titles, and violent content scores were calculated for games by averaging their scores. Measures. Participants were asked to reflect on the past 4 weeks of their computer gaming and select the three games they had played most in that time. They also provided reports of competence-need satisfaction (M
3.66, SD
0.89,
.88), enjoyment (M
4.08, SD
0.80,
.79), and aggressive feelings postplay (M
1.48, SD
0.78,
.82) for three favored games using 5-point response scales. To this end, partic￾ipants completed abbreviated versions of game enjoyment (two items) and aggressive feelings (two items) scales to reduce burden. Because measures had to be completed three times by each par￾ticipant, we only collected reports of postplay aggressive feelings. Despite these limitations, the measures showed good internal con￾sistency. Results Analytic strategy. Given the nested nature of the data— each participant provided reports for three games—we evaluated the hypotheses using hierarchical linear modeling (HLM; Bryk & Raudenbush, 1992). This approach allowed us to test our predic￾tions and make the best use of the hierarchical arrangement of data without inflating model estimates. Player competence. To test the hypothesis that players’ ex￾periences of competence satisfaction in favored games would be related to their levels of enjoyment and postplay aggressive feel￾ings, we evaluated the following Level 1 model: Y
P0 P1 E where Y is the outcome variable, P0 reflects the intercept value of the outcome, P1 represents player competence satisfaction, and E represents error at Level 1. We evaluated separate models for each outcome variable. Results indicated that player competence satis￾faction was negatively associated with aggressive feelings, b
.11, t(919)
3.27, p .01, and positively related to player enjoyment, b
.38, t(919)
11.01, p .001. To test the hypothesis that player competence-need satisfaction was negatively related to postplay aggressive feelings independent of variability in violent content, we tested the following Level 1 model twice: Y
P0 P1 P2 E where Y is the outcome of postplay aggression, P1 represents player competence satisfaction, P2 represents violent game con￾tent, and E is error at Level 1. Results from separate models evaluating P2 using ESRB ratings, b
.01, t(918)
0.65, ns, and P2 using our violent content coding, b
.01, t(918)
0.75, ns, indicated that neither was significantly linked to postplay aggressive feelings. In both cases however, the link between player competence satisfaction and postplay aggressive feelings remained unchanged, b
.11, t(918)
3.22, p .05. Aggressive feelings. We hypothesized that high levels of ag￾gressive feelings postplay would be negatively related to player enjoyment. The following Level 1 model examined enjoyment (Y), as a function of postplay aggressive feelings (P1), and error at Level 1 (E). Y
P0 P1 E Results showed that aggressive feelings postplay were negatively related to player enjoyment, b
.13, t(919)
2.87, p .01, as predicted. General Discussion The sources of human aggression are likely rooted in a diverse array of factors. Among these, popular interest and concerns about the effects of electronic gaming on aggression have gained the attention of numerous scientists, and the ways and extent to which gaming may be linked to aggression has become a hotly debated topic. The present research applies self-determination theory to bring a new perspective to this complex and increasingly nuanced literature. In seven studies, we focused on the ties between gaming and aggression through the lens of this motivational theory to evaluate a competence-impedance hypothesis. Our approach was built on recent SDT-based research that investigates the interactive and structural aspects of games and highlights the importance of need satisfaction for understanding the psychological experience of play (Przybylski et al., 2009; Ryan, Rigby, & Przybylski, 2006). In the present research, our core interest concerned the relations between the basic psychological need for competence and human aggression. We evaluated the prediction that games that under￾mined players’ feelings of competence would lead to increases in players’ aggressive thoughts, feelings, and behavior. Importantly, 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. 452 PRZYBYLSKI, DECI, RIGBY, AND RYAN