At the time of the initial posting, I don’t have the article yet, but google alert went hot with this news. So the only thing I managed to get is the abstract and news. Andrew Przybylski (grad student) and Dr. Richard Ryan (his graduate advisor/supervisor) of the Self-determination lab of University of Rochester had previous research experience on video games research (I know it’s not grammatical correct and long). Update (17/01/09): I just got the article, am reading it and posting up more news links and a youtube video.
Six studies, two survey based and four experimental, explored the relations between violent content and people’s motivation and enjoyment of video game play. Based on self-determination theory, the authors hypothesized that violence adds little to enjoyment or motivation for typical players once autonomy and competence need satisfactions are considered. As predicted, results from all studies showed that enjoyment, value, and desire for future play were robustly associated with the experience of autonomy and competence in gameplay. Violent content added little unique variance in accounting for these outcomes and was also largely unrelated to need satisfactions. The studies also showed that players high in trait aggression were more likely to prefer or value games with violent contents, even though violent contents did not reliably enhance their game enjoyment or immersion. Discussion focuses on the significance of the current findings for individuals and the understanding of motivation in virtual environments.
I believe the adage “sex sells” and “violence sells” is one of the key factors for a successful television program or movie. I suppose the video game industry simply assumed that adage equally applies to interactive entertainment. However, Przybylski and colleagues cast that adage into doubt with their comprehensive studies.
Based on the self-determination theory, they argued that the level of autonomy, competence and relatedness are primary factors of why video games are so satisfying for players, what makes players intrinsically motivated to play and, ultimately, fills some psychological needs. They noted that another theory, uses-and-gratification theory, share similar arguments. IMO, there’s some similarities to Csikszentmihalyi’s flow theory when they wrote that players play them because the activity itself is rewarding rather than an external reward.
Autonomy is the freedom of choice within a game environment. Examples of high autonomy games are the Sim series games (Simcity, The Sims, etc.), Grand Theft Auto series, and role-playing games. Examples of low autonomy games are, IMO, the adventure games, such as the Longest Journey and its sequel, to some degree, fighting games, side-scroll games, and rail-shooters.
Competence is the level of challenge a player faces, it can be a matter of motor skills to creative problem solving; it is also feedback of the player’s performance, how it is delivered and significance to the player. Examples of high competence games are the Guitar Hero series and series related to it (I haven’t played them, but I assume there are complex score rules, i.e. score multipliers and such), side-scroll games, and any game that provide immediate feedback. Games of low competence games are, again IMO, adventure games, they don’t really tell if you’re progressing or not.
Relatedness is the level of attachment to the game (IMO). This factor is not examined since (IMO) it involves the use of narratives which would probably involve a higher level of research resources and different approaches. Half-Life series best exemplifies a high relatedness game, Final Fantasy series too and probably any games that creates hardcore fans and cosplayers. A game of low relatedness would any games that are easily forgettable or have no storylines at all.
According to Zillman (1998) and Jones (2002), violence motivates players because it makes players feel empowered, heroic, excitement, high status, among other things… Maybe they should’ve elaborated a bit more. Another is trait aggression affects individuals’ gaming choices and possibly enjoyment because like attracts like. (not a good line)
Besides what the abstract described, they also looked at the relationship with telepresence or immersion. I know there are no studies between autonomy, competence and immersion, but I believe there are some studies on the relationship between immersion and violence (See James Blascovich and James D. Ivory). I don’t really remember the details.
Looking at the experimental studies, they all follow the same experimental paradigm with some modifications in each experiment to cover several angles and they essentially examine the same variables of interest, game enjoyment and motivation for future play. The first survey looked like a probing study to see if doing the experiments is worth it and the other survey study looks like a gap-filler for the experimental results, more details later on.
Participants: 1028 (929 males and 99 females) participated in the survey, they consisted of members from an popular online community. The online community is not named for privacy reasons. Average age is 24, range is 18 to 39. They used a raffle as an incentive for the participants, 100$. There’s the possibility that a self-selecting sample might not be representative of player population, but having a large sample size might lower the impact of such problem.
Favourite game: Participants are asked to name their current favourite game which would be used as a basis for the other measures.
Player Experience of Need Satisfaction (PENS): The authors own creation from another study. It consist of statements that participants rate on a 7-point scale. The PENS measures for in-game competence (3 items, more I say), in-game autonomy (4 items, more!), presence (9 items, okay).
Game enjoyment: 4 items statements rated on a 7-point scale from a questionnaire. I know the alpha is high (.82), but I want more items!
Sequel interest: It asked whether players would want a sequel made for their current favourite game. That’s a good one.
Word of mouth: asked whether they recommend it to friends. Also a good one.
Violence rating: they used the ESRB and their own violence coding scheme: they rated on a 5-point scale: no violence (Tetris), abstract violence (Mario), impersonal violence (Starcraft), fantasy violence (Final Fantasy) and realistic violence (Grand Theft Auto).
Correlational and regressional analyses were used. They found that violence content (ESRB and study violence coding) has only a statistical significance with telepresence (r = .08, that looks small). Competence and autonomy looked reasonably correlated with the measures, except age and violent content. So survey says violence isn’t important in what players liked in any given game.
This experiment consist of having 68 undergraduates (21 males and 47 females; avg. age is 19.5, no video game experience reported) play House of the Dead 3 for 15 minutes. They selected this game because of easy gameplay (it’s a rail-shooter and point-and-click), graphic violence and the length of time is typical, according to the authors, for this genre. If it were an RTS game, I expect longer.
Similar to study 1, they used the PENS. In addition, they used the Buss-Perry aggression questionnaire before and after playing as a measure for trait aggression. It consisted of 29 statements that are to be rated on a 7-point scale. Another addition is the preference for future play where players rate 5 statements on a 7-point scale (I think) of whether they want to play the game used in the experiment in the future. This one is good since they anticipated some variations of what it means to play in the future, like whether I want to own a copy of it or play it in my free time.
An alternative method which would be good for another study is to have a huge collection of video games and ask participants to choose 5 games. Ferguson had a similar method in one of his studies.
Using correlation and regressional analyses, competence and autonomy was found to be associated with presence, enjoyment and future play preference. As for trait aggression, it has only a statistical significance with future play preference, but not presence as was indicated in study 1; curiously it is also associated with autonomy. There’s some sex difference in terms of competence and autonomy.
This experiment looks like a replication study from Anderson et al. (2004) by having 99 undergraduates (41 males, 58 females, avg. age is 20.1, no video game experience reported) play either a non-violent or violent video game for 20 minutes. The non-violent game is Glider Pro 4 and the violent game is Marathon 2. Wuh?! I guess Anderson et al. chose them because nobody knows those games, eliminating any previous experiences to mess up the results. However, the gameplay differences would be a confound in the results.
Using the exact measures from study 2, again regressional analyses found that competence and autonomy were associated with enjoyment, future play preference and telepresence. Additionally, no difference between violent and non-violent game were found for competence and autonomy and there’s only one sex difference for competence. No significance found for trait aggression, however removing competence and autonomy from the equation resulted in interaction effects for all three variables. The results showed that those with high trait aggression liked the violent game, wants to play it in future and felt higher telepresence, of course that wouldn’t matter if you considered feelings of competence and autonomy, the converse is the same for low trait aggression participants with their non-violent video game.
Study 4 is the best in show among the six studies. 101 undergraduates (36 males, 65 females, avg. age is 19.6, no video game experience reported) played a modded Half-Life 2. They created two versions: a high-violence game where they fight with a shotgun and must kill others who are out to kill the player; a low-violence game where they must use a device or whatever to tag others that when tagged they float for a while and disappear under the assumption that they teleported back to base. A similar game would be Solid Snake catching monkeys a la Ape Escape. A little tangent, Staude-Müller (2008) also used game mods for one of his studies.
They had a 20-minute training period, judging from the youtube video above, it looks they used a human trainer and not an in-game tutorial. I prefer an in-game tutorial to reduce training variability, but that’s pushy. Then, they watched a video clip explaining the gaming situation setting the mood which would lead them to play either the high-violent or low-violent game. Judging from the youtube clip, they’re both exciting. I also noticed that god mode is on and presumably for all studies too. It would be a bummer to add game deaths in the equation.
Using MANOVA for analysis, they found no differences between the low- and high-violence games in terms of competence, autonomy, enjoyment, future play preference and telepresence. There are sex differences, but not going into details. Again, they found the same associations between competence and autonomy with enjoyment, future play preference and telepresence. They also found an interaction effect between trait aggression and future play preference where players with high trait aggression preferred to play high-violent video games than the low-violent video game.
Going back to study 2’s paradigm, they used House of the Dead 3. This time, they adjusted the blood levels, so that there’s low blood condition and the high blood condition. Barlett et al. (2008) also made a similar study. Another difference from study 2 is that they had “avid video game players”, 39 (all males, avg. age is 19.54, video game experience avg. is 7.47 hours per week with 11.23 years of video game experience, a good representation of player population I’d say, but I want more participants). So play time is 15 minutes.
Using the same analyses procedures from study 4, they found no difference between the low-blood and high-blood game in terms of competence, autonomy, future play preference and telepresence. So violence isn’t really much contributing in the gaming experience I gather from now. As expected, competence and autonomy were found to be associated with enjoyment and future play preference, and telepresence, although competence was not significantly associated with telepresence. Interaction effects were found between trait aggression and future play preference and enjoyment, in that those who played the high-blood game with high trait aggression liked and enjoyed it more than the no-blood game. However, when competence and autonomy is added into the equation the interaction effects becomes non-significant.
Armed with experimental results, they returned to surveying to see how well trait aggression results applied outside of the experimental setting. 1642 (1447 males and 195 females, avg. age is 23.9) individuals were surveyed from the same online community. They used the same methods from study 1. They also added game value, whether participants rate on a 7-point scale on their favourite game’s worth. Looking at the results tables, violent content has a largely non-significant relationship with any of the variables of interest whereas competence and autonomy do. As for trait aggression, interaction effects were found for game preference in that high trait aggression participant preferred violent video games whereas low-trait aggression participant don’t figure violence into their preference calculation.
I must admit one thing: I wrote this post in one sitting, I may have written some erroneous details and suffered several writer’s block during the afternoon and evening. But the most important thing is that the abstract tells a much briefer summary. What I provided are some details into how the authors found their results. Nevertheless, it’s preferable to read the article itself to get the real stuff and
not my coffee filter.
Some implications to discuss is that I guess it relieves developers from the need for bloody special effects and concentrate on something else like moar explosions a la Michael Bay. They should concentrate on a more open-world gaming environment and one that is challenging and tells how good we are, in an honest way, of course, I don’t like the “catering to their whiny ego” approach. Let’s see… well the six studies they conducted made sure that they covered many angles and possible alternative explanations, making their claims robust in terms of external validity and convergent validity.
I guess some game-related anecdotes are in order since the paper reminded of some mods. I remember in Unreal Tournament 2003, bombing run, where one has to carry a ball to the opponent’s net while the defending team is trying to block them, either killing the ball-carrier or blocking them. I guess the players weren’t focusing on the violence, but rather on the ball. Another game mod for Quake 3 was a rocket mod where players fire rockets that had shockwave effects where players are thrown in the air. Heh, made me think of a mod where there are concussive micro-missiles used to disorient players. But to the point, there are examples of violent video games being used for non-violent goals or where violence is a secondary feature.
One serious implication is that individuals with high trait aggression have a higher preference for violent video games even though they do not enjoy the game more than anyone else. So there’s the possibility that early exposure to violent video games increases individuals’ trait aggression, which increases their preference for violent video games and therefore an increase in aggressive behaviours (need source). One limitation from this study is that they did not measure participants’ video game experience or media experience. So the relationship has not been examined, but I say they shouldn’t since participants’ memories on media exposure aren’t exactly accurate. Leave it for another study.
Some limitations and future directions the authors noted. Limitations are cultural-restrictive results, applicable only to Western society. Self-report measures, need a more objective measure like behavioural measures. Doubts about the ESRB rating system, didn’t they used their own violence coding? I thought they did that to cover the ESRB’s shortcomings… Future directions to cover are differences in culture norms (I want in!), sensation seeking, personal experience to violence (family violence, school violence, neighbourhood violence, etc.), mindfulness, nerdiness (thanks to Craig Ferguson for the idea), more precise measures of violent content, etc.
Violence does not sell to a lot of people, what remains is does “sex sells”?
Przybylski, Andrew K. and Ryan, Richard M. and Rigby, C. Scott. (2009). The motivating role of violence in video games. Personality and Social Psychology Bulletin, 35, 243-259.