In trying to pass laws that restrict the sales of certain video and PC games with violent content to minors, lawmakers of all types have always said there is valid scientific evidence that there is a link between playing these types of games and the people playing them committing real violent acts. Now Christopher J. Ferguson, Ph. D and faculty member at the Texas A&M International University’s Department of Behavioral, Applied Sciences and Criminal Justice, has taken a look at these studies and analysed them for a new research paper entited, "Evidence for publication bias in video game violence effects literature: A meta-analytic review." As the title suggests, Ferguson's research casts doubt that the studies that have been performed on this subject have really found a link between playing games with violent content and actually committing violent acts. FiringSquad got a chance to chat with Ferguson about his study and what it might mean for the future of this kind of research:
FiringSquad: First, we have to ask; are you an avid player of video and PC games
yourself and if so what are your current favorite games?
Christopher J. Ferguson: Well, I did grow up with the Atari 2600 and like many of my generation, was glued to that thing. I have continued playing video games to the present, mainly strategy games such as the Civilization series and Europa Universalis. I like RPG games sometimes too and was a big fan of the Baldur’s Gate series and spin-offs. I don’t play too many of the first-person shooter games that tend to cause the most fuss nowadays related to aggression, although I do sometimes like the WWII games such as Call of Duty and Medal of Honor. As can happen with age, work and family commitments take up a lot more of my time now, and so I don’t play nearly as often as when I was younger.
FiringSquad: How did the idea for your research study come about?
Christopher J. Ferguson: There were several reasons behind the current meta-analysis. Related to video games and aggression, two main meta-analyses had been published, one by Anderson and Bushman (2001) and one by John Sherry (2001) and these meta-analyses came to somewhat different conclusions. I thought that a third meta-analysis, done somewhat differently, may help resolve some of the differences between these two. In particular neither of the previous metas had examined “publication bias” or the tendency for scientific journals to publish manuscripts that support a particular hypothesis, and to ignore those with “null” findings. Secondly, I also tried to return the focus on the interpretation of “effect size” rather than assuming that any effect size was meaningful. Thirdly, I looked at whether the use of standardized and reliable or unstandardized measures of aggression mattered in regards to outcome and it did. Unstandardized measures tended to produce higher effects than standardized measures.
FiringSquad: How hard was it to examine all of the published studies that concerned
violence and their possible link to games? Did the studies vary widely in
their methods?
Christopher J. Ferguson: Meta-analyses are indeed time consuming and laborious procedures, but they can be fun to do as well. One of the challenges, indeed, is that studies of aggression do vary widely in their methods. This is one thing that meta-analyses don’t always do well and that is to examine the validity of the studies included in the meta-analysis. One way around this is to attempt a “best practices” approach in which a particular facet of methodology is examined in relation to the effect sizes that are produced by that methodology. That is what I did with the use of standardized measures. However, methodologies can vary in many different ways and it’s difficult to consider all of them in a single manuscript.
FiringSquad: Your conclusion stated that these studies actually show little evidence
of any real links between violence and playing games. Where do you think the
studies go wrong, in your opinion?
Christopher J. Ferguson: I may be cynical here, but I suspect that most people who read scientific articles (including perhaps most scientists who are not in the immediate field) tend to read two parts of the manuscript, namely the abstract and (less often) the discussion at the end. The methods section and even the results can be rather dry, but that’s where the most crucial information is. This is where I think that “the field” has “gone wrong”…most scientists outside the area of media violence are familiar with “the gist” of the research (mainly through abstracts) but have not examined the details. Some studies of video game effects are pretty solid in their results (one way or another). Other studies produce results that are rather ambiguous, and thus subject to interpretation. It’s part of human nature that researchers often would interpret ambiguous results as more supportive of their hypotheses than these results actually are. In all fairness to people in the video game effects research area, this is a problem I see for all research that is dependent on statistical probability “null hypothesis testing” (that is social and medical sciences). The concept of “statistical significance” simply means much less than most people realize.
FiringSquad: Where there any specific studies in particular that you felt were
completely off the mark in terms of their methods and their conclusions?
Christopher J. Ferguson: It would not be appropriate or fair for me to discuss the work of specific authors outside of a peer-reviewed format. The manuscript of the meta-analysis is, however, available from the journal web-site. This manuscript has been peer-reviewed in the scientific journal Aggression and Violent Behavior and is “in press” although an early copy is being disseminated through the journal’s web-site. I do discuss some specific studies in the manuscript. Overall I am less concerned that there are imperfections in the studies (arguably all studies have imperfections), and more concerned with the absoluteness with which causal attributions are being made. Certainly when one looks through the results of some studies carefully, they don’t always seem to support the conclusions that authors are making.
FiringSquad: Do you personally believe that this subject needs more study and if so
what methods would you use to acurrately measure the effects of violent
content in games to a person's real aggressive behevior?
Christopher J. Ferguson: Absolutely, this is an important and valid area for pursuit. Understanding the causes of violent behavior (which by the way is something different from “aggressive” behavior) is a very important area of research. Despite my concerns regarding the “tone” of the debate on this issue, I think that the research in this field could be improved rather easily. Honestly there are some researchers (including among those whom I might disagree regarding the causal attribution of video games) who are doing solid work. There are several points that I think are important: 1.) Standardized, reliable and well-validated measures need to be adopted. Laboratory measures of “aggression” need to be validated against real-life outcomes within samples of individuals. Researchers need to routinely report the reliabilities of their measures and discuss validity limitations more openly. 2.) Conflicting and ambiguous results, including results that produce very low effect sizes need to be discussed more critically rather than rushing to confirm a hypothesis. 3.) Researchers should focus less on “statistical significance” and more on “effect size” and begin examining the confidence interval around an effect size. Statistical significance is too easily produced merely by increasing sample size and I would argue is, in and of itself, uninterpretable. 4.) It would be helpful to be careful about drawing inferences between a study and real life violent criminal events. Discussing such events can help make a paper “interesting” but authors should be clear to point out that they are not studying the same phenomenon (laboratory aggression and violent crime are two very different things).
FiringSquad: We have seen so many lawmakers try to pass bills at the city, state and
federal levels that regulate sales of games with violent content to minors,
saying that there is ample research evidence that there is a real link to
these games and aggresive tendencies. Do you feel your work might put these
claims to rest for the time being or do you think it will be ignored?
Christopher J. Ferguson: This is a very good question, and honestly I don’t know what impact this study may have. If I had to guess, it would be that it will be largely ignored. Primarily I suggest this is true because John Sherry’s (2001) meta-analysis of video game effects as well as some of his subsequent manuscripts seem to get less attention, despite that they are published in reputable journals. His results are less supportive of a causal video game-aggression link. The thing about video game effects, and media effects in general is that these provide the appearance of an easy explanation for a complex phenomenon and one which looks deceptively easy to fix. This is also an issue that appeals to people on both sides of the aisle. I suspect that this is an issue that appeals equally to individuals with a liberal or conservative perspective as both groups may “disapprove” of sex or violence in the media (although perhaps for different reasons). Thus I suspect that there is much political appeal in asserting that the media violence – aggressive behavior link is “true” and less political appeal in a more nuanced examination of the issue. Politicians can appear to be “concerned for children” and cast their opponents in a negative light. I suspect that this is why statements made by groups such as the American Academy of Pediatrics regarding media effects research that inflate the number of media effects studies (I believe the AAP has claimed 3500) go largely unchallenged despite that most meta-analyses of media effects in general include closer to 200 studies.
I should point out that there are steps that the video game industry itself could take to avoid being cast so easily as “against children”. The ESRB should take care to present accurate ratings of video game content and educate parents about what games are appropriate for children at different ages. Video game companies also should consider offering advice to parents on how to say no to their children when children ask for a game that the parent is uncomfortable with. Video game companies would also be wise to invest in research examining how games, including games with violent content, may be used for educational purposes so that the positive benefits of the games clearly outweigh any potential negatives. And lastly the industry should avoid “passive aggressive” moves such as including hidden sexual or violent content in a game that the purchaser (or parent) may be unaware of. Such behavior is inexcusable, and casts the entire industry in a negative light.