Video games and aggression Essay Example for Free

Video games and ill will Essay furious computer games such as ? rst-person accelerators (e. g. , Counterstrike) get to repeatedly raised the suspicion of p arnts, teachers, politicians, and scientists alike. Given the increasingly realistic portrayals of fierceness and the substantive training of (virtual) rapacious acts in these games rather than the hands-off observation of furiousness in movies, many have been alarmed by the wide-spread use of these games Smith et al. , 2003.The discussion resembles the previous debate on the effects of passive violence exposure in TV and movies Bushman and Anderson, 2001, and in line with psychological systemal theories on aggression and based on empirical evidence, similar conclusions have been stackn regarding side effects of violence exposure in computer games Most authors would conclude that a clear consensus has been reached that a noniceable causative in? uence of contend raving mad television set games on aggressive fashion and dispositionsof young people in particularexists Carnagey and Anderson, 2004. Nevertheless, the number of studies establishing a causal link between aggresr 2009 Wiley-Liss, Inc. siveness and interactive media such as fierce computer games remains comparatively small in comparison to studies on passive media exposure. Evidence is especially scarce with regard to whether latencybased measures of cognition, so-called inherent measures, be useful for detecting any changes in aggressive cognition as a consequence of exposure to video games. unverbalised measures may be particularly suited to uncover the processes how playing dotty and nonviolent video games affects a players automatic cognitions. Implicit dispositions could play a key part in spontaneous and madcap aggressive tendencies in the piddling and long run. Conventional wisdom holds that a substantial part of aggressive behavior is carried come on in the absence of cognitive ACorrespondence to Matthias Bluemke, P sychological Institute, University of Heidelberg, Hauptstrasse 47-51, 69117 Heidelberg, Germany.E-mail Matthias. emailprotected uni-heidelberg. de Received 15 October 2007 Revised 15 September 2009 Accepted 29 September 2009 Published online 26 October 2009 in Wiley InterScience (www. interscience. wiley. com). DOI 10. 1002/ab. 20329 2 Bluemke et al. resources or in situations where people lack behavioral control (e. g. , after alcohol inspiration).Obviously, aggression does not eternally re?ect actions in line with aces conscious reasoning or explicitly endorsed attitude toward aggression and violence. Those dispositions that relate to lesscontrolled aspects of human behavior, rather than consult behavior and intend actions, may be addressed by the term unverbalised personality Banse and Greenwald, 2007 Perugini and Banse, 2007. The Media Violence ExposureAggression Link Psychological theories that foreknow increases in aggression after (repeated) media violence exposure are plentiful.Not a single psychological theory predicts positive prohibitedcomes, neither in the short nor in the long runexcept for the catharsis hypothesis which until now suffers from empirical con? rmation Bushman et al. , 1999. Among the most important mechanisms for short-term effects are (1) associatory priming of existing aggressive beliefs, well-encoded scripts, and angry emotional reactions Berkowitz, 1993, (2) emotional stimulation upon observation of violence and excitation transfer Zillmann, 1978, and (3) simple mimicry of aggressive scripts Huesmann and Kirwil, 2007.semipermanent effects are most prominently considered to be a consequence of (1) observational learning of new social scripts Huesmann, 1988, (2) development of beliefs supporting aggression or hostile schemas that accompany expectations in social interactions Anderson and Godfrey, 1987 Huesmann and Kirwil, 2007, as well as (3) conditioning of aggression-promoting emotions Bushman and Huesmann, 2006. Long- term emotional desensitization to violent scenes may also go on Carnagey and Anderson, 2004.Empirical evidence in favor of the aforementioned theories is abundant. As the violent video game debate has had a precursor in the debate on the effects of TV-violence, related evidence on the hypothesized link exists. Longitudinal research on the effects of TV-violence has shown that the amount of viewing TV-violence in childhood predicts young adults self- and other-reported aggression much more than childhood aggression predicts young adults TV-violence consumption Huesmann et al., 2003.Owing to the activity of the gamer, violent computer games may be more harmful than passive exposure to media Carnagey and Anderson, 2004 A hostile virtual reality, higher number of violent scenes in the games, symbolically enacting Aggr. Behav. cruelty instead of perceiving it, reinforcement of atrocities, replacement of aggression-inhibiting tendenciesall of these are matter for concern Gentile and Ande rson, 2003.Based on meta-analyses of several studies, Anderson and Bushman 2001 inferred a substantial causal effect of computer game violence on aggressive behavior, aggressive cognitions and emotions, cardiovascular arousal, as well as on (reduced) helping behavior Anderson, 2004 Anderson et al. , 2003. Even if only a small effect existed outside the laboratory, Bayesian logic proves that, due to the high base rate of people consuming large amounts of video game violence, consequences on a societal level would be drastic.In sum, our understanding of the matter has developed to the point where investigating the mediating mechanisms and exploring the moderating variables becomes more important than establishing any effects themselves. This having said, the same does not apply to a relatively new class of theories and measures. So far, few studies in aggression research have dealt with implicit cognition and blush fewer have utilized impudently developed implicit measures of aggres sive dispositions in media violence research.Automatic Aggression-Related Cognitions and Impulsive Aggressiveness Dispositions Beginning with Schneider and Shiffrin Schneider and Shiffrin, 1977 Shiffrin and Schneider, 1977, the distinction between automatic and controlled processes has become quite common. Huesmann 1988, 1998 apply the distinction between automatic and controlled processing to aggressive behavior see also Dodge and Crick, 1990. Also the general aggression framework Anderson and Bushman, 2002 distinguishes thoughtful action from impulsive behavior.The most extensive application of automatic processes to social behavior in general has been laid out in the re?ectiveimpulsive model (RIM) see Strack and Deutsch, 2004, for an in-depth discussion. The model summarizes many ? ndings on human automaticity based on implicit amount procedures. RIM allows for the mutual in? uence of twain cognitive systems in producing human behavior one associative and one re? ective syste m, but interconnections between both systems exist.That automatic processes can be held at least partly responsible for the process of aggressive behavior is not a new insight see Todorov and Bargh, 2002, for an overview. Situational priming of mental constructs in the range of few hundred.The In? uence of Computer Games 3 milliseconds, even below the subliminal threshold, reliably biases peoples perceptions of ambiguous behavior, and it can guide the selection of behavioral options Berkowitz, 2008 Dodge and Crick, 1990 Zelli et al. , 1995. Depending on whether the situation activates the concept of primitiveness or the concept of politeness, the likelihood to interrupt a conversation partner changeswithout mediation by an intentional stance Bargh et al. , 1996. What is less obvious from our discussion so far is how each of the re? ective and impulsive pathways can be predicted. every last(predicate) the models allow automatic associations as dispositions to behavior. Based on sp reading activation in semantic networks, associations ef? ciently predispose the organism to the spontaneous selection of behavioral scripts. Importantly, behavioral impulses can be at variance with ones personally endorsed standards, or social norms, and this may be the case even without the person be aware of it.Whether deliberate re? ection or impulses will determine behavior, depends on the cognitive capacity and motivational resources for self-regulation, which themselves might be impaired due to temporal or chronic in?uences Baumeister et al. , 2000 DeWall et al. , 2007 Fazio and Towles-Schwen, 1999 Muraven and Baumeister, 2000.With the notion of spreading activation in perspicacity implicit measures have been developed that try to tap into automatic associations in the range of a few hundred milliseconds Fazio and Olson, 2003. It was shown that explicit measures, which are based on deliberation and reappraisals, mainly determined behavior under re? ective control, whereas i mplicit measures predominantly predicted impulsive tendencies and behavior in less-controlled situations Friese et al., 2008 Hofmann and Friese, 2008 Hofmann et al. , 2008.The latter ?nding does not contradict the caprice that clever explicit measurement procedures can likewise uncover automatic in? uences in a broad sense. Behavior is the product of both types of processes to a sizable extent, and the situation is responsible for moderating their relative impact. The question is whether the idea of associative networks and priming procedures can be exploited in the domain of aggression, as it has been done in other domains, so that assessing interindividual differences in peoples proneness to impulsively aggress becomes feasible.Assuming automatic aggressive dispositions and development implicit measures to detect them is in line with recent calls to integrate neo-associationistic approaches into explanation and prediction of aggression. In doing so, both classic theoretical and newer paradigms are combined cf. Berkowitz, 2008 Bushman, 1998. Connecting Implicit Measures, Violent Video Games, and Aggression Research We suggest that implicit measurement techniques1 could be a useful addendum to the agenda of aggression research.In contrast to handed-down explicit measures such as questionnaires, implicit measures do not rely on conscious self-report, but on the measurement of hard-to-control spontaneous associations. They typically draw on reactiontimes in categorization tasks within a few hundreds of milliseconds, that is, within the fraction of a second where also automaticity effects can be observed. Implicit measures are considered to be less susceptible to distortion by demand characteristics, social desirability, and other biasing factors such as low levels of introspection Degner et al., 2006.Crucially, due to the especial(a) time for responding, information processing in implicit measures differs distinctively from responding to a questionnaire so t hat both types of measures display their merits, particularly when predicting different kinds of behavior Dissociations between implicit and explicit measures in predicting impulsive and controlled behavior typically result Asendorpf et al. , 2002 Hofmann et al. , 2007, and treatments can affect the associative and re? ective level independently Gawronski and Bodenhausen, 2007.Heavy players of violent video games may claim to be immune to side effects, and at the re? ective level this may hold, but at the associative level the picture may look quite different. Owing to the nature of the game, impulsive behavior and automatic associations, aside from intentions, could be reinforced in violent computer games. Uhlmann and Swanson 2004 observed barely such a predicted increase of aggressive cognition after 10 min of playing a violent computer game in the lab, when hostility was measured objectively via response latencies in an Implicit Association Test IAT Greenwald et al., 1998.Other research shows that these IATs are predictors of impulsive aggression which cannot be explained by end-to-end the article we stick with the common name implicit measures for indirect, latency-based measures. Note that the ideas that the constructs proper reside at an implicit level, or that the associations themselves need to be acquired implicitly, have been granted up, and there is no doubt that most measurement procedures cannot be deemed implicit Blanton et al. , 2006 Fiedler et al. , 2006 Karpinski, 2004.Aggr. Behav. 1 4 Bluemke et al.self-report and observer ratings Banse and Fischer, 2002 Gollwitzer et al. , 2007. As the study by Uhlmann and Swanson 2004 is, to our knowledge, the only promulgated research that studyd the in? uence of violent computer games on cognition as assessed by the IAT, let us describe their main ? ndings. Playing a ? rst-person winker increased implicit aggressiveness. Despite being convergent with theory, some doubts remain. The lack of a nonplay ing control crowd does not permit a conclusion whether the violent game raised aggressive cognitions or whether the playing control condition caused participants to become more still.Then, Uhlmann and Swansons games presumably differed with regard to the elicited physiological arousal and involvement. The nonviolent game (Mahjongg) was a puzzle that differs from the violent game (Doom), a ? rst-person shooter, in terms of excitation, task complexity, competition, and frustration. The missing equivalence prohibits inferring a causal link Anderson et al. , 2004. Arousal confounds need not pose a problem for explicit measures of aggression, yet applying speeded-classi? cation tasks after playing arousing games might have affected take performance in the IAT.As a consequence, group differences may re?ect blurred measurement, rather than changes in cognition proper. Study Aims We had two aims in mind when planning this study First, a conceptual replication of Uhlmann and Swansons 2004 study seemed in place, while simultaneously controlling for arousal and task differences of the games.Second, presumption the small number of studies on the causal impact of violent and prosocial electronic games on implicit measures, we wanted to extend the data basis We expected that playing a violent game should prime aggressive cognitions, whereas playing a unruffled game should prime peaceful cognitions.mouse and ? ring at them with mouse clicks.In the peaceful game participants watered as many cheer? owers popping up in the woods at the same rate and pace as the soldiers in the violent condition. An abstract game required the clicking of disconsolate triangles without any meaning attached to these triangles, but with identical timing parameters and reinforcement stakes. This allowed us to picture whether violent content and watering sun? owers distinctively sway associations as compared with a control condition.Finally, nonplaying participants worked on a reading task of a nonarousing publisher report, constituting a service line for potential arousal differences due to playing vs. not playing. In line with Uhlmann and Swanson 2004, we predicted that, controlling for pretest differences among participants, the implicit aggressive selfconcept should be highest after ? rst-person shooting, followed by abstract playing period, then by sun? ower watering. Implicit measures should be particularly informative on alterations of associative structures.Associating oneself with violent acts should give rise to aggressive cognitions, whereas associating oneself with peaceful acts should render peaceful associations active. As we kept the virtual environment, the psychomotor task, and the gaming parameters constant, we also expected that the level of physiological arousal among the three game conditions should converge. This accompaniment would demonstrate the equality of the game contexts and render explanations of post-treatments effects in terms of plain arousal differences improbable.Sample A sample of 96 students at Heidelberg University of non-homogeneous majors took part in a study on the in?uence of computer games on (unspeci? ed) cognitive performance parameters in exchange for course cite or a chocolate bar. After controlling for high error rates 20% of errors at most in the full of life IAT and Single-Target IAT (ST-IAT) impedes see Greenwald et al. , 1998, 89 participants (68. 5% females) remained in the sample. 2 Mean age amounted to 24. 64 yrs (SD 5 5. 35).Most participants were skilled in computer usage and gaming Many reported owning a Personal Computer (N 5 86), Sony Playstation (9), Microsoft Xbox (2), or a Nintendo Gameboy (9). Daily computer usage was 2.53 (SD 5 2. 65) hr on average, and the average weekly consumption of video games 2 METHOD Hypotheses We compared three groups, relative to a control condition, with regard to changes of aggressiveness following violent gaming, nonviolent gaming, or not gaming at all. In the violent game, participants acted as ? rst-person shooters and targeted a virtual weapon at hostile soldiers, popping up in a virtual wood, by moving the hairlines of the gun with the Aggr. Behav. Owing to technical problems, the recording of one participants physiological data failed.The In? uence of Computer Games 5 summed up to 5. 16 (SD 5 7. 90) hr. Participants were randomly assigned to one of the four conditions under the constraint of keeping gender proportions across the conditions equal. This resulted in 68 males and 1416 females in each condition. Independent Variable Although the control group encountered a reading task, that is, an article from the German magazine DER SPIEGEL which was judged as emotionally neutral, the data-based groups encountered one of three computer games.Irrespective of the speci?c treatment condition, the virtual environment (a forest scene) and the actions (a left-side mouse click of the right hand) were identical (Fig. 1). In the vio lent game, participants were exposed to a war scenario that required shooting enemy soldiers from a ? rstperson perspective in order to set high. Soldiers returned ? re if they were not eliminated today. The goal was to shoot as many enemies as quickly as possible by ? ring at them with mouse clicks (hits), before they ? red back and disappeared, resulting in score losses (misses).The mean rate of soldiers per minute could be determined by the programmer and was kept constant across participants (and conditions), but the program implemented a random component with regard to timing and location of the targets so that players could not routinely counter the attacks. Misses after the fraction of a second resulted in being injured and decreased the score, signaled by a different sound than for hits, which were visually emphasized by blood spills.By contrast, in the peaceful game sun? owers popped up in the same wood in the same speed like the soldiers in the violent game, yet the play ers task was to water the ?owers with their watering can, else they died visually due to water shortage. Whenever this happened, a sad sound occurred and reminded a participant to water the sun? owers continuously and fast. On success, a players score increased, as indicated by a sound of accomplishment.Misses resulted in the same loss of points as in the violent game. Finally, in the abstract game participants removed the colored triangles that popped up in the woods by pinpointing them with a small cursor triangle before clicking the mouse button. Acoustic and visual signals added relevance to hits and misses.Dependent Variables gigantic ? ve. A German 40-item version of the International Personality Item Pool IPIP40 Goldberg, 2001 Hartig et al. , 2003 provided a basic description of personality in terms of the ? ve-factor model extraversion, neuroticism, conscientiousness, openness, and agreeability John and Srivastava, 1999. Both the ? ve-factor structure of the IPIP40 as well as its construct validity have been demonstrated beforehand Buchanan et al. , 2005. We used the IPIP40 in order to control for pretest differences among the sub-samples. inhering consistencies of the scales were adequate, Cronbachs a 5 . 74. 90. Aggression questionnaire. We administered Buss and Perrys 1992 29-item aggression questionnaire BPAQ German version by Amelang and Bartussek, 2001 to control for pre-existing group differences and to investigate postexperimental changes of aggressiveness. The German version ? ts the well-validated four-factorial structure physical aggression, verbal aggression, anger, and hostility Herzberg, 2003 von Collani and Werner, 2005.Reliability of the sub-scales, a 5 .62. 78 (. 67. 85), Aggr. Behav. Fig. 1. Screenshots of the violent, peaceful, and abstract game (printed in greyscale). 6 Bluemke et al. and the total scale, a 5 . 85 (. 87), proved adequate (post-test values in brackets). Implicit Association Tests. The computerbased IAT and its deri vate, the ST-IAT Karpinski and Steinman, 2006 Wigboldus et al. , 2004 unpublished, were administered as implicit measures of cognitive antecedents of impulsive aggression, cognize as the aggressive self-concept Banse and Fischer, 2002.In the IAT, the main dependent variable, response latencies, resulted from two sorting tasks that cross the two focal attribute conceptsaggressive and peacefulwith the two target categories, self and other. After 20 practice trials for attributes and targets each, 40 stimuli of both targets and attributes had to be categorized (see Table I binge 47) In one block self1peaceful (and other1aggressive) were mapped to identical response keys, in the other block the category response-key assignment reversed, self1aggressive (and other1peaceful). Stimuli were randomly drawn from one of the four categories.Both blocks were administered in equilibrate order across the sample to control for block order effects. The difference between the mean response latenci es of the two critical blocks (i. e. , IAT effect), served as an magnate of the association of the self to the aggressive vs. peaceful pole of the dimension. Typically faster responses for the self1peaceful than for the self1 aggressive block result. Hence, when taking individual block differences of zero milliseconds as a part point, positive IAT hit indicate a peaceful self-concept and negative IAT oodles indicate an aggressive self-concept.Previous studies showed that IATs predicted the amount of violent game playing Uhlmann and Swanson, 2004, accounted for extraordinary variance in the aggression of ice hockey players as indicated by penalty time-outs Banse and Fischer, 2002, and detected the impact of a social competence intervention Gollwitzer et al. , 2007. Because of the reproof pertaining to the relative nature of the IAT Blanton et al. , 2006 Fiedler et al. , 2006 Karpinski, 2004, we additionally applied an aggressiveness-ST-IAT that omitted the contrast category oth er, as it is unclear what exactly testtakers associate to an unspeci?ed IAT category, such as other.The measurement of latencies, the block structure, and the stimuli of the two critical blocks remained the same as in the IAT, yielding one compatible block with self1peaceful on the one key and aggressive separately on the other key, and one incompatible block with self1aggressive (peaceful separate). The simpler task structure usually decreases latencies, but, crucial for the calculation of block difference scores, across both blocks there is Aggr. Behav. always one un conjugated category. For nonrelative target objects, such as the self-concept, an ST-IAT may contain less nuisance variance than an IAT.In our own pilot study, a self-concept ST-IAT re? ected past violent video game exposure better than an attitude-toward-aggression-IAT, probably as a result of range restriction of the true-score variance of participants evaluative associations in the latter measure Bluemke and Zumbac h, 2007. Successful ST-IAT applications have shown that the ST-IAT can do almost as good in psychometric terms as the IAT. Nevertheless, research on this tool is still warranted as the evidence for the usefulness of an aggressiveness-ST-IAT is scarce.We reduced the in?uence of the asymmetric nature of the task by drawing 10 self-related stimuli, 11 stimuli of the coupled category, and 14 stimuli of the unpaired category, resulting in 35 stimuli per critical block and an almost equal number of left-hand and righthand responses 40 vs. 60% see Table I, Block 13 cf. Bluemke and Friese, 2008 Friese et al. , 2007. 3 The ST-IAT always preceded the IAT so as not to prime the category other before taking the ST-IAT. Irrespective of whether participants encountered the compatible or the incompatible block ? rst, they received the same order of blocks for the post-test.Latencies were treated according to the D5-algorithm regarding the treatment of missing data and error penalties Greenwald et al. , 2003, resulting in metrics equivalent to z-standardized scores or d-scores. Thus, ST-IAT and IAT effects are expressed in units of an individuals standard deviation pooled across both (task-speci? c) critical blocks. Individual differences were assessed with boundary reliability, a 5 . 68 (. 64) and . 64 (. 73). Again, to summarize, positive IAT or ST-IAT scores indicate a peaceful self-concept and negative IAT or ST-IAT scores indicate an aggressive self-concept.physiologic arousal parameters. As we wanted to preclude any arousal differences between groups, we assessed heart rate (HR) and skin conductance (SC) as parameters of emotion-related physiological arousal by using the Biopac student lab PRO 3. 6. 7. e. g. , Carnagey et al. , 2007 Clements and Turpin, 1995 Malmstrom et al. , 1965. The measurement procedure was divided into six sections. Data were continuously gathered, and aggregates of 30-second intervals were analyzed for each of the following phases a baseline imme diately after 3The disproportionate number of categories in the ST-IAT prevents that both the number of left and right key-strokes and the number of peaceful and aggressive stimuli in the two critical blocks can be balanced. We chose a solution between both extremes. TABLE I. Structure of ST-IAT and IAT Including (ST-)IAT Items (Translated From German) Block 2 Initial combined task (here compatible) Others egotism Me You mine Yours I Self They Their Them My Others Self Me You Mine Yours I Self They Their Them My Target-concept discrimination Initial combined task (here compatible) Reversed target discrimination Block 3 Block 4 Block 5 Block 6 Block 7.Sequence Block 1 Task Attribute discrimination Task operating instructions Stimuli Aggressive Peaceful Compromise Fight Agree Blow Reconciliation Give in Hurt revenge Hit Make peace exit of trials 10 aggressive 10 peaceful Aggressive Peaceful1Self Me Fight Agree Mine Blow Self Reconciliation Give in Hurt I Revenge Hit Make peace My C ompromise 14 aggressive 11 peaceful 10 self-related 10 self-related 10 other-related 10 10 10 10 Reversed combined task (here incompatible) Aggressive1Self Peaceful Me Fight Agree Mine Blow Self Reconciliation Give in Hurt I Revenge Hit Make peace My Compromise 11 aggressive 14 peaceful 10 self-related Aggressive1Others Peaceful1Self Me Fight Agree Mine Blow Self Reconciliation Give in Hurt I Revenge Hit Make peace My Compromise 10 aggressive 10 peaceful 10 self-related 10 other-related self-related other-related self-related other-related Reversed combined task (here incompatible) Aggressive1Self Peaceful1Others Me Fight Agree Mine Blow Self Reconciliation Give in Hurt I Revenge Hit Make peace My Compromise 10 aggressive 10 peaceful The In? uence of Computer Games 7 Note Within the task instructions, spacial position of the categories indicates the left or right response key.Target and attribute stimuli alternated in critical IAT blocks (here depiction of an arbitrary sequence of stimuli). Aggr. Behav. 8 Bluemke et al. attaching the electrode (Pre-1), a pre-treatment baseline (Pre-2), a treatment phase subdivided into one early, one mid-term, and one ? nal interval (T1T3), and a post-treatment phase before the detachment of the electrode (Post). Procedure After entry in the lab, we obtained written informed consent that participants might randomly end up in a violent game condition and stressed that they could opt out at any point in time without giving any reasons.None of the participants used this option, neither in response to the initial information, nor during the course of the experiment. At ? rst, participants reported on socio-demographic variables, and then took a personality questionnaire related to the ? ve-factor model, before they encountered baseline measures of an aggression-speci? c questionnaire, an ST-IAT, and ? nally an IAT. Next, the experimenter attached the devices for measuring HR and SC at the index ? nger of the left hand. Followi ng a short introduction to the randomly chosen game condition, participants played, or read, for a period of only 5 min.Arousal measurement continued until a re-test of the aggressiveness questionnaire was completed, but the devices were detached before we administered the implicit measures a second time. Subsequent to questions on computer usage derived ? from Krahe and Moller, 2004, the session ended by ? careful debrie? ng of participants. In sum, all phases lasted about 30 min altogether. Z2 5 . 04. Scores of BPAQ subscales likewise did not change (all F-valuesr1. 21). Arousal All groups displayed a typical pattern of initial excitement and habituation (Fig. 2).As expected, when testing the equivalence of games in terms of physiological arousal, according to a 4 (experimental condition) A 6 (time Pre-1, Pre-2, T1, T2, T3, and Post) analysis of variance (ANOVA) with repeated measurement on the latter factor, no group differences on HR emerged, Fo1 (Z2r. 01). Importantly, there wa s no interaction between time trends and experimental treatment, Fo1 (Z2r. 03). Running the same analyses on SC as a more sensitive measure of arousal also showed no reliable differences between groups, Fo1 (Z2r. 03), and time trends were not moderated by experimental condition, Fso1 (Z2r. 02).As could be expected, the violent game showed a slight numerical increase in SC (from Pre-2 to T1). We therefore examined each of the six measurement occasions separately. Only at the beginning of the play (T1) did signi? cant variation exist, F(3, 84) 5 5. 04, P 5 . 003, Z2 5 . 15. Post hoc tests according to Tukey (HSD) revealed that the violent game resulted in somewhat higher excitement compared with the abstract game and the reading task (Psr. 01). Importantly, violent and peaceful games did not differ signi? cantly, P 5 . 14. Only 1 min later, the initial startle-like reaction had vanished (Fo1 at T2).Implicit Measures The impact of games was analyzed by a onefactorial ANOVA on change sc ores between IAT pre- and post-test (Table II). Replicating the ? ndings by Uhlmann and Swanson 2004, type of game signi? cantly in? uenced implicit aggressiveness, F(3, 85) 5 2. 93, P 5 . 04, Z2 5 . 09. 4 Introducing participant sex as a control factor resulted in an interaction between sex and game content, F(3, 81) 5 3. 33, P 5 . 02, Z2 5 . 11. Whereas change scores did not differ as a function of sex, Fo1, the impact of game content became clearer at the same time, F(3, 81) 5 4. 00, P 5 . 01, Z2 5 . 13.The pattern of IAT change scores and the signi? cance of the contrasts between games within sex indicated that the sex by game interaction was particularly driven Introducing Order of Block Compatibility did not change conclusions on the game factor, F(3, 81) 5 2. 92, P 5 . 04, Z2 5 . 10, other Fso1. Also using BPAQ pretreatment scores as covariates in ANCOVA models did not alter the conclusions, though some of the covariates tended to explain small portions of IAT variability, Ps Z. 08, Z2sr. 04. 4 RESULTS Explicit measures To preclude any pre-existing group differences, we examined the Big Five personality scores before treatment.According to a multivariate analysis of variance on the IPIP40 scales, participants were comparable F(15, 249)o1, Z2 5 . 04, regardless of the speci? c Big Five scale, Fsr1. 26, PsZ. 30, Z2sr. 04. We also study whether the random assignment to experimental conditions worked by analyzing trait aggressiveness. As expected, neither before, Fo1 (Z2 5 . 03), nor after the treatment, F(3, 85) 5 1. 36, P 5 . 26, Z2 5 . 05, did substantial group differences in self-reported aggressiveness on the BPAQ total scale exist. Replicating Uhlmann and Swansons 2004 ? ndings, trait questionnaires did not respond to video play, according.