Violent video games studied as a teaching tool (Gentile & Gentile, 2008)

Now Andrew Eisen asked some questions about the study in and I can’t help but to try to answer them. The best practice of getting the right answers is to e-mail the authors themselves, the contact info should be on the first page.


This article presents conceptual and empiricalanalyses of several of the ‘‘best practices’’ of learning and instruction, and demonstrates how violent video games use them effectively to motivate learners to persevere in acquiring and mastering a number of skills, to navigate through complex problems and changing environments, and to experiment with different identities until success is achieved. These educational principles allow for the generation of several testable hypotheses, two of which are tested with samples of 430 elementary school children (mean age 10 years), 607 young adolescents (mean age 14 years), and 1,441 older adolescents (mean age 19 years). Participants were surveyed about their video game habits and their aggressive cognitions and behaviors.

The first hypothesis is based on the principle that curricula that teach the same underlying concepts across contexts should have the highest transfer. Therefore, students who play multiple violent video games should be more likely to learn aggressive cognitions and behaviors than those who play fewer.

The second hypothesis is based on the principle that long-term learning is improved the more practice is distributed across time. Therefore, students who play violent video games more frequently across time should be more likely to learn aggressive cognitions and behaviors than those who play the same types of games for equivalent amounts of time but less frequently. Both hypotheses were supported. We conclude by describing what educators can learn from the successful instructional and curriculum design features of video games.

Now on to the questions:

“Why was a second aggression sample only done with the elementary school kids?”

Can’t really answer right away, need time. Maybe after mid-december when i’m done with undergrad thesis proposal.

“How significant was the aggression increase? I look at the table in the study and I have to wonder what the difference between a .33 and .44 overall physical aggression index is.” The table he’s refering is”Table 1″

Now I might be able to answer that. Now the table only show the raw correlational data, you can’t compare correlational scores to predict something. so they used logisitic regression (see in wikipedia).

Now speaking of only the 3-5th graders, they’re trying to predict variables measured at time 1 to time 2 ( measures take on november-February and measures taken April-May). So they controlled several variables from being a factor into their calculation: sex, race, age, lag(?), weekly amounts of video game play, and time 1 hostile attribution bias.

Therefore, figure 1 is what needs to look at, but I haven’t been taught on how to read logistical regressions. so i’m just reading from the text. So controlling all these variables, they found that playing multiple violent video games influence kids concept of aggression as a normal thing and will see things in aggressive terms, so with an aggressive cognition, which would result in more aggressive behaviours. Unfortunately, I do know they have a significant effect, but how large is the effect? Well, i don’t know. Maybe it’s small like Patrick Markey said?

 Update(11/02/08): I’ve finally read the article, see my comments after the break.

Gentile & Gentile are looking at conceptual similarities between educational practices and violent video games. I guess they want to show how violent video games can be teachers of aggression. Well, let’s dig in.

Here’s an interesting statement: since the average gamer is getting older, it will lead to gamers being parents themselves. IMO, I do wonder if it includes hard cores, mid-cores and casual gamers. I would just remove the casual gamers from my statistics. Anyways, they argued that parents have likely become desensitized to violent content and may be less likely to monitor children’s games. An interesting speculation, I can see why.

However, I’m also arguing that a gamer-parent might be more likely to monitor their children’s games since they are more likely to play video games (VG) with their kids and provide parental guidance through violent video games. In addition, they might have effective strategies in limiting the children’s play time than non-gamers (speculation of course). Another speculation of my own is the developmental course of violence desensitization, I haven’t read any longitudinal studies on violence desensitization, but it’s certainly true that our attitudes and behaviours change as we grow into adulthood. A final consideration is the context of the would-be parents media consumption, what is worse? A high preference of violent media or a high consumption of media with violence and non-violence in equal ratio? TV vs. video games vs. internet?

Leaving that aside, Gentile & Gentile listed some elements as to why violent video games are potential educational tools:

  1. Clear objectives, difficulty of tasks are adapted to players’ knowledge, skill and learning capabilities (game objectives are simple, but always hard to complete)
  2. Hands-on practice, repetition, and mastery (A trip is worth a thousand travel books)
  3. Overlearning (FPS pro players don’t really think much on what keys to press since they practice so much to the point that it’s second nature to them, this allows them to concentrate on strategy and tactics.)
  4. Immediate feedback and rewards (feeling pride when I beat a bot on hard mode)
  5. Difficulty of tasks is dependent on earlier successes (first levels in games are always meant to familiarize players to the game, then if they mastered the basics, they would be introduced to harder stuff, rinse and repeat; another example is game tutorials)
  6. “Close-to-optimal combination of massed and distributed practice” (Unlike cramming, you practice your skills everyday, you play your game everyday anyway)
  7. Variety (you don’t simply use your knowledge for just one game, but also in other situations or contexts)

So from these listed elements, they decided to test out two of the principles: the sixth and seventh element. So their hypotheses are that children who play violent video games more regularly and who play multiple violent video games are more likely to learn aggressive behaviours and cognitions. They do so through a mixed cross-sectional and short longitudinal study.

A side note for those interested in the educational field, I recommend writings from Paul James Gee.


Participants: Three groups of participants: 430 elementary students, designated as school children (3rd grade: 119, 4th grade: 119 and 5th grade: 192) (age M=9.65, SD=1.03)
607 secondary students, designated as young adolescents (8th grade: 496, 9th grade: 111) (age: M=14, SD=.64)
1441 college students, designated as late adolescents (age M=19.4, SD=1.73)
All participants are drawn from various areas (urban, suburban and rural) of Minnesota. Ethnic group distribution in the sample appears that above 80% are Caucasian and for the rest are other ethnic groups that occupy single digits percentage. Gender distribution is equal.


General measures for all groups

Violent video game exposure: elementary children were asked to name their three favorite video games. From these three, they were asked on often they played them on a 5-point scale and how violent they think these games are, on a 4-point scale.

For young and late adolescents, the scales are different. Frequency: 7-point scale instead of 5. How violent it is: 7-point, instead of 4. In addition, they were asked on their preference of how much violence they want in video games on a 10-point scale today and 2-3 years ago on a 5-point scale.

Frankly, from the Brenick et al. and from Gentiles’ own literature review, they should know that playing violent video games often should distort participants’ perception of graphic violence. So, a gamer might see it as not so violent, while a non-gamer might see it as very violent. However, this could go the other way too, so this balances it out. Also, the perception of violence is subjective, so there’s no other alternative I could think of.

Another minor issue is whether one play many video games, violent and non-violent, on a equal basis. Or having to name three games is too much or little for various individuals. A hypothetical situation I have in mind, is whether one individual is playing popular violent video games, but is sandwiched by a sea of non-violent video game. However, how many participants are in this situation is probably unlikely and small.

Weekly amount of video game play: a composite measure and I like it. How much time do they play during different time periods (basically morning, afternoon and evening). These are asked for weekdays and weekends for obvious reasons.

For young and late adolescents, it is the same as the children’s, they added late night (past midnight), in addition to the three time periods.

Measures for school children only

Aggression measure: three measures were taken (see below) and combined to form a single aggression score. They used statistical techniques to do that, however my worries is whether the peer assessment measure and teacher measure do agree with each other. I don’t know if the self-report measure of fights is combinable measure. I rather have a measure that is similar to the other two.

  • Peer assessment of social adjustment: a measure where participants nominate three other participants in their class that fits the description of what is being asked of. Example: Name three people who are the most altruistic in your class? In this study, only two items (physical aggression) are used for this study: those of hit or kick others and those who push and shove others around. My problem with that measure is what if no one did push or hit others?
  • Teacher ratings of aggressive behavior: a survey questionnaire in assessing children’s aggression and pro-social behaviour. The aggression subscale is used in this study, unknown numbers of items reported. Items are responded on a 5-point scale.
  • Self-reports of fights: One question was asked of the participants if they were involved in a school fight during the year.

Assessment of hostile attribution bias / social information processing: a good measure that I like. Participants read 10 stories, each describing a provocation and the intent isn’t obvious. Participants answer two questions: What is the provoker’s intent? They are given 4 answers (2 hostile vs. 2 benign) to pick from. Did the provoker intended to be mean, yes or no? so this measure is to assess participants thinking and bias towards hostile thinking. (See [social cognition])

Measures for young adolescents only

Trait hostility: a modified, age-appropriate Cook & Medley Hostility scale was used. It’s reliable.

Arguments with teacher: on a scale of 1-to-4, how often do you argue with your teacher in the past year? Range from almost daily to less than a month. How about never?

Physical fights: did you physical fight in the last year? yes or no?

Measures for late adolescents only

Trait anger, trait hostility and overall physical aggression: used the Buss-Perry Questionnaire to measure three traits pertaining to aggression. It’s reliable and it’s asking participants to agree with various statements that describes them. Example: “I have a lot of temper tantrums” and participants rate how much they agree with it.

Proactive and reactive physical aggression: used the Social interaction survey. It’s a self-report survey of which are answered on a 7-point scale. Proactive means doing aggressive acts at first, reactive means responding with aggressive behaviours.


For the school children, measures were taken at two time points. First time point (Time 1): between November and February. Second time point (Time 2): April and May. So this is a short longitudinal study. For the other groups, measures were given only once.


It’s organized according to the hypotheses they proposed to test.

After reading the results section, it makes a lot of sense if you read the fine print. So here are the fine print that the media might have missed:

  • Participants who have at least named three games are included in the analysis
  • The violence rating of the three games are averaged. So two games with a violence rating of 7 with a game of a violence of 1 would have an average violence rating of 5, in the case of the young and late adolescents groups.
  • Following the second point, therefore participants with three violent VG as their favorites would logically have a higher violence rating than their mix or low VG violence peers.

So the hypothesis that playing multiple violent video games (VG) will result in “better” (WTF?) aggressive cognitions and behaviours than playing fewer. 

At first, they used correlations in their analysis. They found that students who play multiple violent VG have a higher hostile attribution bias score, hostility personality score and physically aggressive than their mixed and low violent VG peers.

However, the issue of multicollinearity was raised and they decided to do logistic regressions for a more robust analysis.

In the analysis, they controlled several variables in their analysis for various age groups.

For the young adolescents, they controlled for sex, age, race, hostile attribution bias, and total amount of VG play per week. So what’s included in the analysis are level of violent VG exposure and amount of physical fights. The results showed a significant result between the level of violent VG exposure and the amount of physical fights.

For late adolescents, they controlled for sex, age, race and total amount of games played. What’s included in the analysis are trait hostility, trait anger, overall physical aggression, proactive and reactive aggression and level of violent VG exposure. Results: Level of violent VG exposure is significantly associated to overall physical aggression. No word on proactive or reactive aggression.

For the school children, they used regressions to predict Time 2 variables from Time 1 variables. They controlled for sex, age, race, lag (where did that come from?), amount of time of VG play and Time 1 hostile attributional bias. What’s included in the analysis are the aggression measure, amount of VG violence exposure. The results showed that there were an increase in hostile attribution bias and physical aggression from Time 1 to Time 2. Demonstrating a six month effect VG violent content on individuals.  

They also used path analysis to show the strength of this relation. VG violent exposure at Time 1 had a .16 direct effect on Time 2 hostile attribution bias. However, Time 1 physical aggression had a .64 direct on it too (with the same statistical significance rating of 0.001). So I believe that while VG violent exposure is significant and deserves attention, a person’s physical aggression score might have a greater effect and attention. Would it render VG violence effect trivial? I don’t know. Would refraining children from playing violent video games be helpful? IMHO, Yes, it would help, especially if the child is already aggressive, don’t want aggravate the situation do you?

If we look at the interactions between Time 1 variables, there are significant associations. Namely, the amount of time of game play, gender and time 1 physical aggression scores to VG violence exposure (.24, .39 and .33, respectively). Unfortunately, I don’t know the direction of these interactions, whether they are one-way or two-way effects since the data is correlational.

Now on to the second hypothesis: regularly playing violent VG over time will increase one’s aggressive cognitions and behaviours.

This hypothesis is tested only on the young adolescents and late adolescents age groups. They created the independent variable, designated as distributed practice, by multiplying frequency of game play per week or month with number of years of game play. They created two VG violent exposure groups into two, split at the middle. They used multiple regressions to analyze the data. They included hostile attribution, trait hostility, arguments with teachers and physical fights as the dependent or outcome variable.  Sex, grade and age were controlled.

Results for the young adolescents group: significant results were found in the association between distributed practice and arguments with teachers, regardless of low vs. high VG violence exposure. A significant result was found for the high VG violence exposure group for hostile attribution bias.

Results for the late adolescents group: among the high VG violence exposure group, there were significant associations between distributed practice and trait anger and aggression (proactive, reactive and overall). There was a close significant association for trait hostility.  Among the low VG violence exposure group, there were no such associations.

The split between low and high violence VG exposure, for the second hypothesis, had me thinking as to why did they make such a split. What are the methodological reasons to split a continuous variable into a dichotomous one. Unless, it’s obvious to the professional readers, but I’d like to know why?

It seems that all age groups are affected by violence exposure in video games, however one remaining questions is the effect size among the age groups. Are children more affected by violent exposure than older age groups? That is the story for another study…

Some limitations the authors thought for this study:

  1. Possible measurement errors for amount of time play.
  2. Erroneous assumption in the statistical analysis that VG violence exposure is constant, correcting for it might be hellish for the authors IMO.

Anyways besides the results and discussion of the results, they wrote about the implications for teaching in comparison with the media. No, I’m not going to bother reading and commenting.

Gentile, D., & Gentile, J. (2008). Violent video games as exemplary teachers: A conceptual analysis. Journal of Youth and Adolescence, 37(2), 127-141.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s