PALGN and EveryonePlays, in partnership with GAME retailers, have set up a petition for everyone who believes that Australian video game classifications should include an adult R18 rating, in order to protect children from mature games. Politicians and the sensationalist media will often point to psychology studies and the perceived link between video game violence and aggressive behaviour, as a reason for not having an R18 rating. However, what do these studies really say? Are they really denouncing violent video games as the devil, or have the academics been taken out of context?
Since 1995, Professor Craig Anderson from the Department of Psychology at Iowa State University has been studying, among other topics, the effects of violent media on aggressive behaviour. Currently, he is a Distinguished Professor and the Director of the Center for the Study of Violence, where his attentions have turned to video games and other interactive media. Over the last 10 years or so, Anderson’s studies in particular have caught the attention of those against the introduction of an R18+ rating in Australia.
While the sensationalist and mainstream media will throw headlines such as “Violent Video Games Cause Aggression”, Psychology studies stemming from statistical methodology and practice often have a much deeper story to tell. Since we’re not all statisticians or psychologists, this article tries to put these methods and the corresponding results into layman’s terms.
Furthermore, we’ll take a look at both sides of the story: some researches have found a causal link between violent games and behaviour and some have not, and why this is the case. Importantly, we’ll see what lessons we can learn from both sides. It ought to be noted though, this is only a very brief look at the issue at hand, and there is a lot more than what is displayed here. This is the first of two features, and we’ll take a look at the methods used in these studies as well as the latest study from Professor Anderson.
The low-down on Meta-Analysis
In any statistical analysis, the size of your sample is extremely important. In psychology experiments, it can be extremely difficult and expensive to conduct them with large sample sizes. So what academics and researchers such as Anderson will do, is gather several published results and then analyse the results of these experiments using a technique known as ‘Meta-Analysis’. This essentially involves taking all these smaller but conceptually identical experiments and analysing them as a group. This is the primary methodology for adding strength and statistical validity to the link between violent video games and aggressive behaviour.
Aside from being able to make better use of small sample experiments, Meta-Analysis also is able to control for a set of different sized experiments by weighting them appropriately and has a bunch of mathematically favourable properties, including the control of variation between studies. However, there are also downsides to using this method. Basically, unless your smaller studies are all well designed, your Meta study will not have meaningful results. Furthermore, this methodology often relies on the availability of other studies, so the quality can sometimes be out of your control.
Occasionally, there are two studies pointing in one direction will give an opposite effect together – like two magnets repelling each other. This may lead to the subjective rejection of a particular study from the analysis. Finally, when a researcher is using studies other than their own, they have to be wary that generally only ‘statistically significant’ studies are published and available. As for a note on statistics in general (and not just Meta-analysis), if the person conducting the study has a particular agenda, be it political, social or economic, it will adversely affect the credibility of their study. But that goes without saying.
Two schools of thought
According to the 2008 paper from Dr. Christopher Ferguson of Texas A&M International University titled “Violent Video Games and Aggression – Causal Relationship or Byproduct of Family Violence and Intrinsic Violence motivation”, there are two main schools of thought on the relationship between violent video games and aggressive behaviour.
The first school of thought is derived from ‘Social Learning Theories’, and centres around the work of Anderson and his colleague, Professor Brad Bushman, called the ‘General Aggression Model’ or GAM. Essentially, this theory stipulates that being exposed to violent media (be it games, film or other) will write ‘aggressive cognitive scripts’ in our mind. Apparently, we then use these scripts as a guide on how to act under particular social circumstances. So, the more violent the material a person is exposed to, the more violent or aggressive their responses to particular situations may be. The theory also stipulates that should you come across an ambiguous situation, you’re more likely to resort to behaviour from these scripts.
General Aggression Model or 'GAM'.
While it is believed that this approach ought to demonstrate short and long-term effects, it disregards factors such as genetics, personality and family environment. A lot of the studies so far have identified short term effects, including an apparent increase in aggression and desensitising to violence. However, the long term effects remain unidentified. Overall, the theory implies that even if you had no pre-existing violent motivations, you can acquire them through exposure to violent media.
The second school of thought derives from ‘Biological/Innate Motivational Theory’. In contrast to the first school, which suggests a causal relationship between violent video games and aggressive behaviour, this theory suggests that the relationship between violent and/or aggressive behaviour is more likely to be explained by genetic and environmental factors. According to this theory, the violence in video games is more likely to affect the ‘method’ of violence. So while this theory says that the desire to act violently is caused by genetic, personality and environmental factors, the way this violence is expressed may be a result of exposure to violent media. The violent media in this case is referred to as a ‘stylistic catalyst’. The ‘Catalyst Model’ that has been derived from this theory, has not been subject to much research yet though, particularly in relation to the GAM.
The still to be tested Catalyst Model.
Violent Video Games and Aggression: Studies from Professor Craig Anderson
In 2003, Anderson stated that “after 40 year’s of research, one might think that debate about media violence effects would be over”. In 2010, it’s still raging on.
As mentioned at the outset, the studies from Professor Anderson and his various colleagues are often used in the arguments against excessive violence in video games. And more recently, opposition to the R18+ rating in Australia. We will concentrate on his most recent study that has been doing the rounds in the media. For anyone who has seen the Sunrise segment on Friday 5 March, this is the study that they refer to.
Violent Video Game Effects on Aggression, Empathy, and Prosocial Behaviour in Eastern and Western Countries: A Meta Analytic Review (2010)
The significance of this study comes in its size. It gathers 130 previous studies for a total sample size of 130,000, which makes it the biggest study of its kind. As you will see though, this sample size is a slight misnomer. Also, it was done with a combination of studies from the US, Europe and Japan, being amongst the first to see whether there are any cultural differences that can be taken into account. No one was aged over 16 in the study, and genders were recorded and accounted for.
The study concentrates on each of these six main variables, and the relationship that each (individually) has with exposure to violent video games:
- Aggressive behaviour (can be physical or verbal)
- Aggressive cognition (having aggressive thought processes)
- Aggressive affect (e.g. feeling frustrated if a game is too hard)
- Prosocial behaviour (basically, ‘helping’ behaviour)
- Empathy/desensitisation (empathy towards ‘victims’ and whether players become desensitised to violence)
- Physiological arousal (e.g. increased heart rate)
So, Experimental, Longitudinal (long-term over different time periods) and Cross-sectional (short-term over the one time period) were three primary kinds of studies used in the overall meta-analysis. What these studies are looking for are ‘effect-sizes’, which are then summarised with the ‘correlation coefficient’ denoted by the letter ‘r’.
Doug Gentille and Professor Craig Anderson - two leading authorities in their research fields.
Results
In this section, we provide a summary of the results.
As mention, the key statistic is the value of ‘r’, which ranges between -1 and +1. If it’s equal to zero, that means that there is no effect, or that the two variables in question are unrelated. The closer it is to +1 or -1, the stronger the effect. Below are all of the six variables tested. We provide you with the figures for each of the three different study methods (where applicable) and with the overall figure, which is constructed from all three. We also provide results for some of the other effects that were tested in conjunction with each variable.
It ought to be noted, that when a statistical study talks about a statistic being ‘significant’, that just means that it is a value other than zero, or that an effect is present. So while an effect may exist, the magnitude of it is something that is a lot more ambiguous. Mainstream media and those who don’t understand statistics will often take this ‘significance’ out of context. For the stats in question, the late statistician and meta-analysis guru Jacob Cohen recommended that ‘r’ be greater than or equal to 0.1 for it to have any meaning. So even if a study shows an r=0.1 or an effect size of 10%, chances are that it still isn’t doing much more than zero.
In this study, all variables were found to have statistically significant effect sizes.
- Aggressive behaviour: r+ = 0.189
Experimental: r+ = 0.181
Longitudinal: r+ = 0.198
Cross-sectional: r+ = 0.189
This suggests that video game violence increases aggressive behaviour, across all three experiment types.
Other findings:
-There was no statistical difference in cultural effects, so no difference between in the effect of being of Eastern or Western origin on how violent games affect aggressive behaviour.
-There was no statistically significant difference between boys and girls.
-There was a slightly statistically significant effect associated with age, so a slightly smaller effect with higher age.
- Aggressive cognition: r+ = 0.162
Experimental r+ = 0.207
Longitudinal r+ = 0.110
Cross-sectional r+ = 0.164
This suggests that video game violence increases aggressive cognition, across all three experiment types.
Other findings:
-The relationship effects were found to be significantly larger in Western than Eastern cultures.
-There was no statistically significant difference between boys and girls found, due to lack of studies.
-There were no statistically significant effects associated with age found, due to lack of studies.
- Aggressive affect: r+ = 0.139
Experimental r+ = 0.181
Longitudinal r+ = 0.082
Cross-sectional r+ = 0.145
This suggests that video game violence increases aggressive affect, across all three experiment types.
There were no other findings.
- Prosocial behaviour: r+ = -0.101
Experimental r+ = -0.161
Longitudinal r+ = -0.114
Cross-sectional r+ = -0.86
This suggests that video game violence decreases prosocial behaviour, across all three experiment types.
Other findings:
-The effects were found to be significantly larger in Western than Eastern cultures, but only in cross-section studies.
-There was no statistically significant difference between boys and girls.
- Empathy/desensitisation: r+ = -0.177
Experimental r+ = -0.148
Longitudinal r+ = -0.160
Cross-sectional r+ = -0.188
This suggests that video game violence decreases empathy/increases desensitisation, across all three experiment types.
Other findings:
-The effects were found to be significantly larger in Western than Eastern cultures, but only in cross-section studies.
- Physiological arousal:
Only experimental studies were available to measure physiological arousal.
Experimental r+ = 0.184
This suggests that video game violence increase physiological arousal.
Findings:
Aspects of games such as: Player perspective, game role game violence target, average age and game playing time, were all insignificant in increasing physiological arousal.
- Note: Readers may notice that a lot of the longitudinal studies seem to have lower stats than the other studies. This suggests that the corresponding effects are weaker over time. So for example, the effect of violent video games on aggressive cognition dissipates isn’t very strong across time. Also, the overall stat for prosocial behaviour (r = -0.101) barely satisfies Cohen’s criteria for significant meaning.
Do you geddit?
Comments/Criticism
All of the variables yielded results that were “expected”. So the conclusion is that video game violence causes more aggressive behaviours and tendencies, while making players less empathic, less prosocial and more desensitised to violence. However, Anderson him self is quick to concede that while an effect exists, the size or magnitude of the effect will come under intense debate. Furthermore, he concedes that a ‘normal’ and ‘stable’ child isn’t going to become a mass murderer simply by playing a few violent games, where he says that ''If you have a child with no other risk factors for aggression and violence, and if you allow them to suddenly start playing video games five hours to 10 hours a week, they're not going to become a school shooter''.
But Dr. Christopher Ferguson, an associate professor at Texas A&M International University, who we’ll be examining in the next part of this feature, said in a critique accompanying the study that the effects found "are generally very low". He adds that the analysis "contains numerous flaws", which he said would result in "overestimating the influence" of violent games on aggression.
For the study in general, Anderson seems to be heavily involved in a number of the past publications that are being examined, as he constantly refers to proofs and previous findings from his own studies. Ferguson claims that this runs the risk of ‘publication bias’, which is something that is difficult to avoid in Meta studies. Furthermore, this study concedes that a lot of the prosocial behaviour doesn’t seem to take into account the explosion of co-operative games in recent years. Games such as Gears of War, Resident Evil 5 and Army of Two that are violent, but require you to take care of your co-operative partner, haven’t been examined. Apparently, studies on these are currently in the works.
Finally, a lot of the aggression tests have been accused of being irrelevant, too short, or foolish by researchers, including Lawrence Kutner and Cheryl K. Olson. For example, tests compared the time college students held an air horn down before and after they played a violent video game for twenty minutes. However, as Ferguson explains, it is unethical to incite people into literal acts of aggression, so such proxy tests are the only feasible way to conduct experiments. Another criticism of this study has been that the measure of aggression not been ‘standardised’, or put into the same terms (e.g. having a study with feet and metres as measurements, but putting them together without converting them into one consistent unit of measurement).
Aggression in a can?
Stay tuned tomorrow, where we’ll bring you the other side of the argument, the conclusions from both the studies and our summary on the overall argument. Until then, if you haven’t already signed a petition at GAME, why don’t you head down to your nearest store today and help protect children from mature games?
Read more at http://everyoneplays.org.au
Become a fan of EveryonePlays!
Follow EveryonePlays on Twitter!
Loading...
