I often wonder what kind of impact my blog has and to whom it affects…
Cheryl Olson and Lawrence Kutner (with colleagues) have published a study based on data that created their book “Grand Theft Childhood” and propelled their fame with the video game community. I became aware of this study through gamepolitics, the study didn’t show up in a database search through informaworld. Oh well…
This research examined the potential relationship between adolescent problem behaviors and amount of time spent with violent electronic games. Survey data were collected from 1,254 7th and 8th grade students in two states. A ‘‘dose’’ of exposure to Mature-rated games was calculated using Entertainment Software Rating Board ratings of titles children reported playing ‘‘a lot in the past six months,’’ and average days per week of video game play. Analyses were conducted using simultaneous logistic regression for binary outcome variables, and simultaneous multiple linear regression for continuous outcome variables, controlling for a series of potential confounders. M-rated game dose predicted greater risk for bullying (p<.01) and physical fights (p<.001), but not for delinquent behaviors or being a victim of bullies. When analyzed separately, these associations became weaker for boys and stronger for girls.
A full-text link to the study is available through gamepolitics.
Their introduction section is nothing out of the ordinary and in line of their writing pattern: Inconsistent methodologies and results in the video game literature, legislative failures in restricting mature video games sales to minors, vague definitions, stretched out generalizations of media effects to severe violent criminal activities.
There are certain parts of the introduction that has piqued my attention. The measurement for video game violence exposure and video game use has no “gold standard”. The authors cited several studies with contrasting measures making them difficult to compare. I thought the General Media Habits Questionnaire (developed from Iowa State University) was the gold standard, although there are flaws I personally experienced during one of my projects that had me questioning its validity. One flaw that I agree with Olson and Kutner is that participants were asked to name their favourite video games, however the question is left quite vague and very much open to interpretation. A participant of mine wrote down Duck Hunt as her favourite game, but had rated having played quite rarely. Another mentioned a game that he hasn’t played for years, but is still his favourite. Another flaw is they decried participants’ violent video game exposure is not accurately measured due to the reliance on participants’ self-report.
They cited a meta-analysis paper by Sherry which found longer play sessions with a violent video game has a smaller effect on aggression than shorter sessions. If you refer back to the abstract, participants of this study were asked to list the top 5 games “they played a lot”. Those that play a heavy dose of violent video games were more likely to bully and get involved in physical fights. The connection between Sherry’s paper and the study is rather tenuous when I gave some serious thoughts, although my initial thoughts were that children who play a lot shouldn’t be different from others because they played longer hours based on Sherry’s paper. Unfortunately, the study didn’t have many video game use variables, so it’s rather difficult to gauge the quantity, quality or type of interactions present in a video game session.
Given their research background are in child development and public health, the study is more applied and relevant to the average person. Everyone has a better knowledge of bullying than aggression which is more basic research than applied, from a certain point of view. They certainly argued bullying is a more pressing concern than school shootings since the latter are rarer phenomena. The problems associated to bullying, both in victims and perpetrators, range in emotional adjustments, peer relationships, physical health and academic performance. I’m curious as to the possible implications when they enter the workforce. Will the school bullies become workplace bullies? Will we have more assholes to interact with? Will the screaming 13-years old we find in online games become future trolls? While it may seem trivial compared with the extreme violent behaviours, bullies can make our lives miserable.
Participants: 1,254 7th and 8th graders from Pennsylvania and South Carolina participated in Olson and Kutner’s 2004 multi-faceted study. 90% are white. Average age is around 13 years old.
Unspecified questionnaire: participants answered questions on amount of time spent playing video games, game preferences, context of and motivations for game use. They also responded about non-media activities, attitudes, beliefs, and experiences related to aggression and conflict and school performance.
Amalgamated child aggression questionnaire: they adapted several elements of previously validated questionnaires as their variables for child aggression: The Olweus Bully/Victim Questionnaire, the Profiles of Student Life: Attitudes and Behaviors Survey and the Youth Risk Behavior Survey.
Aggressive personality: used the Attitude Toward Conflict Scale. Doesn’t the number of items, but participants respond questions on likert scales.
Trait anger: They created a 3 item on a 5-point scale to assess trait anger. Asking how often they feel angry, showing their anger and recognizing anger in others. I’m very irritable to the number of items in a measure, having fewer items means measuring less complete parts of a psychological construct.
Exposure to video game content: Participants listed their five games that “they played a lot in the past six months”. The authors contend that this allows them to analyze on media content that children are exposed to recently and frequently, in other words, a specific self-report question, but self-report nevertheless. The kids were also asked to estimate the number of days per week they play video games.
The study measured the violent content of the video games based on the ESRB rating scheme. You know, E for everyone, T for teens, M for mature, etc. If a video game series or a versions existed between different consoles is listed, then it is assigned a rating based on the least violent version in the last two years. Why not the average or median? I’m pretty that’s an acceptable assessment. It’s fine to evaluate video game content based on the ESRB rather than relying on children’s perceptions, it standardizes the media content. But others would argue that personal evaluation (e.g. how often they notice violence) is important to determine exposure. This is related to media violence desensitization.
There is a neck-cracking issue about their description of the M-rating. They described M-rated games as often featuring socially undesirable or criminal activities. I disagree that M-rated game often feature criminal activities, many M-rated games are war-themed games, games with adult themes (e.g. complex moral dilemmas, sexual themes, etc.) and graphic violence. To support their claim, a content analysis is needed to ascertain how often M-rated games feature socially undesirable or criminal activities. Therefore, it should be written as “games featuring criminal activities are often rated as mature”. The differences in wording show that criminal activities are often rated mature, but most mature-rated games do not feature criminal activities.
Descriptive statistics are as follows, 80 children (6% of sample) haven’t played any video games in the past six months. 1,126 children wrote down at least one video game and will be included in the analysis. From those numbers, 48.8% listed at least one M-rated video games on their top 5 list. Separating by gender, 67.9% of boys listed a M-rated game, whereas 29.2% of girls do. The most popular is Grand Theft Auto (44% among boys, while 20% for girls). If anyone has their book, you can find the top 10 popular video games among boys and girls.
There is a potential limitation (IMO) about listing or including games as mature. It’s not that the ESRB has done a bad job. It’s the popularity and social factors concerning certain video games. Two M-rated games popular among boys are Grand Theft Auto and the Halo series (Medal of Honor is included, but is rated Teen). From a content analytical point of view, it sure is violent. However, these games are blockbusters and are thus should be treated differently. My reasoning is people’s perception of a game’s content changes according to peer influences or opinions towards said game. Blockbuster games draw on gaming and story elements (knowing what is probably a trade secret) known to sell well to a broad audience, explaining why Halo is popular to all ages. Further, the context of the game matters: an analysis of video game popularity by a wider age range might reveal preference differences; Grand Theft Auto is a sandbox-type game, therefore it is very hard to quantify and qualify its M-rating in comparison to the war-themed games (See Lachlan & Maloney, 2008). Additionally, the online gaming environment could become a significant socializing agent for young children and the reliance on the ESRB would be a limiting factor for calculating violence exposure.
The abstract mentioned of an approximate “dose” of violent video game exposure, they calculated a participant’s exposure with the following formula.
The same is done with less violent rating games as well.
Statistically controlling for confounding variables, such as gender, geographical location of school, grade level, grades, trait anger, aggressive personality and dose of exposure to nonviolent/less violent games, they found that M-rated video game use was found to be a significant predictor for bullying and physical aggression (Trait anger is also a significant predictor). Apparently, an increase to the day-per-week to M-rated games increases the probability for bullying by 45%, whereas it is 24% for physical aggression. Now, I had to ask is what is the baseline score or probability (say one or zero day per week exposure to M-rated games). The same pattern is true for less violent video games.
Fortunately, there were no statistical relationships for delinquent behaviours (predicted by poor grades, high trait anger, location, and being a boy) or being a victim of bullies (predicted by enrolment in an earlier school grade and higher trait anger).
Due to gender difference of video game use, they split the data by gender and found that the relationships for boys is weaker where M-rated games is only predictive for physical aggression, whereas for girls is stronger for both bullying and physical aggression.
So their results partially support their hypothesis about playing a lot of M-rated game being predictive for common behavioural problems, even controlling for known extraneous variables. They interpreted the boys’ weakening statistical relationship being that video games are normative activities, so good behavioural conduct is expected of boys from video game use (IMO).
I found the lack of discussion regarding girls’ statistical relationships disappointing. I could offer some thoughts on the matter, but it’s all guesswork here, so don’t use it as a reference. Video games are still a male-dominated activity (well for the hardcore part anyway), so girls are treated differently by their peers. If I recall, gender threats lead girls to be more aggressive. Perhaps, they learned or behaved to be more aggressive in order to play video games or for social approval, from both boys and girls, for playing violent video games. It could go the other way, perhaps these girls are different from the average girl, in terms of personality, traits, or socialization. A general methodological caveat is the possible gender biases of psychological measures, in particular with aggression, where the consensus among the masses is that girls are less physically aggressive than boys, although they are rather more relationally aggressive (waiting for anything from Jamie Ostrov).
A study is incomplete without limitations. Its limitations is its correlational nature (no causal relationships can be established despite my use of the word predictive because that’s what they wrote in the paper), children only listed their top 5 games and not all the games they played in the last six months (by the way, I need a study investigating the relationship between the gamer and the life cycle of a video game), self-reporting measures (i.e. can anyone accurately and reliably define “a lot”), and other potential extraneous variables (e.g. parenting).
Olson, C. K., Kutner, L. A., Baer, L., Beresin, E. V., Warner, D. E., & Nicholi, A. M. (2009). M-rated video games and aggressive or problem behavior among young adolescents. Applied Developmental Science, 13, 4, 188-198.